From db045a96bbbab6c857e8a1cbffeb612ff7759b27 Mon Sep 17 00:00:00 2001 From: fxpineau Date: Mon, 1 Jun 2026 11:16:38 +0200 Subject: [PATCH 01/10] Changes for v2.1, including M. Taylor comments --- Makefile | 11 +- appendix_custom.tex | 17 + appendix_fmoc1.tex | 99 +++++ appendix_json.tex | 3 + appendix_packing.tex | 65 +++ moc.bib | 46 +++ moc.tex | 953 +++++++++++++++++++++++++++++++++---------- role_diagram.pdf | Bin 29884 -> 30008 bytes table3.tex | 13 +- tablef.tex | 41 ++ 10 files changed, 1018 insertions(+), 230 deletions(-) create mode 100644 appendix_custom.tex create mode 100644 appendix_fmoc1.tex create mode 100644 appendix_packing.tex create mode 100644 tablef.tex diff --git a/Makefile b/Makefile index 255e064..719e68c 100755 --- a/Makefile +++ b/Makefile @@ -4,19 +4,20 @@ DOCNAME = moc # count up; you probably do not want to bother with versions <1.0 -DOCVERSION = 2.0 +DOCVERSION = 2.1 # Publication date, ISO format; update manually for "releases" -DOCDATE = 2022-07-27 +DOCDATE = 2026-06-01 # What is it you're writing: NOTE, WD, PR, or REC -DOCTYPE = REC +DOCTYPE = WD # Source files for the TeX document (but the main file must always # be called $(DOCNAME).tex -SOURCES = $(DOCNAME).tex role_diagram.pdf appendix_algo.tex appendix_json.tex \ - appendix_perf.tex table1.tex table2.tex table3.tex +SOURCES = $(DOCNAME).tex role_diagram.pdf \ + appendix_algo.tex appendix_json.tex appendix_perf.tex appendix_fmoc1.tex appendix_custom.tex \ + table1.tex table2.tex table3.tex tablef.tex # List of pixel image files to be included in submitted package FIGURES = healpix.jpg kilo.png kilo1.png kilo2.png kilo3.png kilo4.png role_diagram.svg \ diff --git a/appendix_custom.tex b/appendix_custom.tex new file mode 100644 index 0000000..bd7850f --- /dev/null +++ b/appendix_custom.tex @@ -0,0 +1,17 @@ +\section{\add{User define quantity MOC conventions}\label{sec:udmoc1}} + +\add{The conventions adopted for FMOCs are generic and could be applied to any quantity +$q$ numerically represented as a 64-bit floating-point value. +A user wishing to define a custom QMOC (where $Q$ stands for Quantity) would only need to specify: +\begin{itemize} + \item the validity range of the quantity: $[q_{min}, q_{max}[$; + \item the mapping function: $g_q: [q_{min}, q_{max}[ \rightarrow [0,1[$. +\end{itemize}} + +\add{However, one of the advantages of MOCs is the use of standardized quantities expressed +in well-defined units and reference systems. +In order to perform efficient and meaningful set operations between two MOCs, both MOCs must +be defined using the same quantity, expressed in the same units, and using +the same validity interval and mapping function.} + +\add{Poorly constrained or insufficiently documented user-defined quantities may therefore compromise the interoperability of MOCs.} diff --git a/appendix_fmoc1.tex b/appendix_fmoc1.tex new file mode 100644 index 0000000..394e7b0 --- /dev/null +++ b/appendix_fmoc1.tex @@ -0,0 +1,99 @@ +\section{\add{Previous frequency indexing convention}\label{sec:apx:fmoc1}} + +\add{This section describes the frequency axis indexing scheme that was initially explored before +the current standardized version. Documenting it here serves two purposes: +\begin{itemize} + \item to document earlier versions of FMOC and SFMOC files that may still exist; + \item to record an already explored solution and thus avoid attempting to reintroduce it in the future. +\end{itemize}} + +\add{The original idea was to (almost) directly use, as the index at the deepest level, +the integer representation of the 64-bit floating-point value of a frequency expressed in Hz. +Performing MOC operations (except \textit{degrade}) on floating-point ranges is equivalent to +performing them on the corresponding integer representation ranges because: +\begin{itemize} + \item frequencies are strictly positive values; + \item for positive floating-point numbers, the numerical ordering is identical to the ordering of their + integer bit representations. +\end{itemize}} + +\add{The only difference was that the exponent range of a 64-bit floating-point value was considered too +large for the intended use. Therefore, it was restricted to raw exponent values +$\in [929, 1184]$ instead of the full IEEE~754 range $\in [0, 2047]$. +The selected interval contains 255 distinct values and therefore requires 8 bits +instead of the 11 bits used in standard double-precision representation. +Using the 52 mantissa bits together with these 8 exponent bits leads to 60 hierarchical +orders, with a maximum order of 59. This also defines the, somewhat arbitrary, validity range +$f \in [5.0487097934144756 \times 10^{-29},\; 5.846006549323611 \times 10^{48}[$. +The following pseudo-code describes the transformation, which replaces the +11-bit exponent $\in [929,1184]$ of a floating-point value with an 8-bit exponent +$\in [0,255]$.} +\lstdefinelanguage{Pseudocode}{ + keywords={input, output, return, datatype, function, in, if, else, foreach, while, begin, end + }, + sensitive=false, + comment=[l]{//}, + morecomment=[s]{/*}{*/}, + morestring=[b]', + morestring=[b]" +} +\begin{lstlisting}[language=Pseudocode,mathescape=true,caption=Frequency in Hz to order 59 index routine,label={code:freq2index}] +Input: long float freq_hz +Output: long int index +begin + // Set the 64-bits float exponent bits mask + long int emask $\gets$ 0x7FF << 52 + // Set the mantissa (+sign) bits mask + long int mmask $\gets$ !emask + // Get the bits level representation of the input frequency + long int index = freq_hz.to_bits(); + // Extract the exponent from the input frequency + long int exp $\gets$ (index & emask) >> 52; + // Replace the exponent value in [929, 1184] by a value in [0, 255] + exp $\gets$ exp - 929 + // Replace the 11 exponent bits by the new 8 exponent bits + index $\gets$ (index & mmask) | (exp << 52) + return index +end +\end{lstlisting} + + +\begin{lstlisting}[language=Pseudocode,mathescape=true,caption=Frequency in Hz from order 59 index routine,label={code:index2freq}] +Input: long int index +Output: long float freq_hz +begin + // Set the 64-bits float exponent bits mask + long int emask $\gets$ 0x7FF << 52 + // Set the mantissa (+sign) bits mask + long int mmask $\gets$ !emask + // Extract the exponent part in [0, 255] + long int exp = (index & emask) >> 52; + // Replace epxonent value in [0, 255] by value in [929, 1184] + exp $\gets$ exp + 929 + // Replace 8 exponent bits by 11 exponent bits + index $\gets$ (index & mmask) | (exp << 52) + // Transform bit representation into float value + return index.f64_value_from_bits() +end +\end{lstlisting} + + +\add{One advantage of this approach was the existence of a perfect bijection between the +64-bit floating-point representation of a frequency (in Hz) within the valid range and +its index at the deepest resolution. The conversions were also extremely fast, requiring +only a few bitwise operations. In addition, the number of orders (60) was not limited to +the number of mantissa bits.} + +\add{However, a major issue arose from the representation of the axis using WCS, as used in HiPS 2.0. +Because the representation allocates 8 bits to the exponent and 52 bits to the mantissa, +the resulting WCS cannot be expressed purely as either linear or logarithmic. +For orders 0 to 7, the WCS behaves logarithmically. +For orders 8 to 59, inside a cell or order at least 7, it behaves linearly. +For ranges spanning order 7, however, the WCS becomes a mixture of logarithmic and linear +components, which cannot be represented easily. +This limitation was the primary motivation for abandoning this approach.} + + + + + diff --git a/appendix_json.tex b/appendix_json.tex index 06a4ec8..8e259bb 100644 --- a/appendix_json.tex +++ b/appendix_json.tex @@ -11,6 +11,9 @@ \section{JSON encoding} of the MOC (MOCORDER) is greater than the greatest stored order, the MOCORDER should be provided with an empty index list. +\addb{Although we expect index arrays to be sorted, we cannot impose +the same constraint on orders, since JSON object members are considered unordered.} + \paragraph{Example of a JSON SMOC or TMOC:} \par\noindent \begin{lstlisting}[] diff --git a/appendix_packing.tex b/appendix_packing.tex new file mode 100644 index 0000000..7f0cf3b --- /dev/null +++ b/appendix_packing.tex @@ -0,0 +1,65 @@ +\section{\add{SUNIQ and ZUNIQ packaging}} + +\add{Although they are not used in this standard, the SUNIQ and ZUNIQ packing schemes +are implemented in some MOC libraries and are therefore briefly described here. +They are valid with all quantities.} + +\subsection{\add{SUNIQ packaging}} + +\add{SUNIQ provides a generalized alternative to the UNIQ packing, which is specific to HEALPix indices. +Its main drawback is that it requires one additional bit. +SUNIQ relies on a sentinel bit (hence the “S” in SUNIQ). +Given a cell index $index$, the SUNIQ value is obtained by setting a sentinel bit immediately to the +left of the most significant bit (MSB) of the index. + The resulting expressions for HEALPix and TMOC/FMOC indices are: +\begin{eqnarray*} + suniq_{hpx} & = & 4^{order + 2} + index \\ + suniq_{t/f} & = & 2^{order + 1} + index +\end{eqnarray*} +} + +\add{The inverse operation uses the {\textit leading\_zeros} function, +which returns the number of leading zeros in the binary representation of a 64-bit (unsigned) integer. +\begin{eqnarray*} + order_{hpx} & = & \frac{59 - leading\_zeros(suniq)}{2} \\ + index_{hpx} & = & suniq - 4^{order + 2} \\ + order_{t/f} & = & 62 - leading\_zeros(suniq) \\ + index_{t/f} & = & suniq - 2^{order + 1} +\end{eqnarray*}} + +\add{Sorting UNIQ or SUNIQ values produces the same ordering of cells: cells of lower order appear first, and within a given order, cells are ordered by increasing index.} + + +\subsection{\add{ZUNIQ packaging}} + +\add{ZUNIQ encodes both the cell index and its order in a single integer while preserving the +ordering of cells according to their ranges at the maximum resolution. +HEALPix indices correspond to the concatenation of 12 Z-order curves and therefore follow the +natural ordering of a multi-resoltuion Morton (Z-order) curve, which motivates the name ZUNIQ. +ZUNIQ is constructed from the starting index of the cell at the maximum resolution, left-shifted by one bit, +with the bit at the right of the least significant bit used as a sentinel. +This scheme is similar to the numbering used by +\textit{S2CellId} in the Google \textit{S2Geometry} library. +\begin{eqnarray*} + zuniq_{hpx} & = & 2 \times index_{29} + 4^{29-order}\\ + zuniq_{t} & = & 2 \times index_{62} + 2^{62-order}\\ + zuniq_{f} & = & 2 \times index_{52} + 2^{52-order} +\end{eqnarray*} +More generally: +\begin{eqnarray*} + zuniq_{hpx} & = & 2 \times index_{order_{max}} + 4^{order_{max}-order} \\ + zuniq_{t/f} & = & 2 \times index_{order_{max}} + 2^{order_{max}-order} +\end{eqnarray*} +} + +\add{The inverse transformation uses the {\textit trailing\_zeros} function, +which returns the number of trailing zeros in the binary representation of a 64-bit unsigned integer: +\begin{eqnarray*} + order_{hpx} & = & 29 - \frac{trailing\_zeros(zuniq)}{2} \\ + index_{hpx} & = & \frac{zuniq - 4^{order_{max}-order}}{2} \\ + order_{t/f} & = & order_{max} - trailing\_zeros(zuniq) \\ + index_{t/f} & = & \frac{zuniq - 2^{order_{max}-order}}{2} +\end{eqnarray*} +Sorting cells according to their ZUNIQ values enables efficient transformations between lists of cells +at mixed resolutions and lists of ranges. +More generally, such orderings allow efficient streamed operations.} diff --git a/moc.bib b/moc.bib index 290f1f0..39c969f 100644 --- a/moc.bib +++ b/moc.bib @@ -1,3 +1,49 @@ +@article{cecconi_2023, + year = 2023, + title = {{Time-frequency catalogue: JSON implementation and python library}}, + author = {Cecconi, Baptiste and Louis, Corentin K. and Bonnin, Xavier and Loh, Alan and Taylor, Mark B.}, + journal = {Frontiers in Astronomy and Space Sciences}, + doi = {10.3389/fspas.2022.1049677}, + pages = {1049677}, + volume = {9}, +} + +@misc{loh_2025, + year = 2025, + title = {{AlanLoh/nenupy}}, + author = {Loh, Alan and Coutouly, Lucile and Girard, Julien N and Mauduit, Emilie and Viou, Cedric}, + url = {https://github.com/AlanLoh/nenupy}, + publisher = {{GitHub}}, +} + + +@INPROCEEDINGS{zarka_2028, + author = {{Zarka}, Philippe and {Coffre}, Andree and {Denis}, Laurent and {Dumez-Viou}, Cedric and {Girard}, Julien and {Griessmeier}, Jean-Mathias and {Loh}, Alan and {Tagger}, Michel}, + title = "{The Low-Frequency Beamformer and Imager NenuFAR}", + booktitle = {2018 2nd URSI Atlantic Radio Science Meeting (AT-RASC)}, + year = 2018, + volume = {2}, + month = may, + publisher = {IEEE}, + eid = {308}, + pages = {308}, + doi = {10.23919/URSI-AT-RASC.2018.8471463}, +} + +@MISC{2022ivoa.spec.0727F, + author = {{Fernique}, Pierre and {Nebot}, Ada and {Durand}, Daniel and {Baumann}, Matthieu and {Boch}, Thomas and {Greco}, Giuseppe and {Donaldson}, Tom and {Pineau}, Francois-Xavier and {Taylor}, Mark and {O'Mullane}, Wil and {Reinecke}, Martin and {Derri{\`e}re}, S{\'e}bastien}, + title = "{MOC: Multi-Order Coverage map Version 2.0}", + howpublished = {IVOA Recommendation 27 July 2022}, + year = 2022, + month = jul, + pages = {727}, + doi = {10.5479/ADS/bib/2022ivoa.spec.0727F}, + adsurl = {https://ui.adsabs.harvard.edu/abs/2022ivoa.spec.0727F}, + adsnote = {Provided by the SAO/NASA Astrophysics Data System} +} + + + @MISC{2021ivoa.spec.1101D, author = {{Dowler}, Patrick and {Evans}, Janet and {Arviset}, Christophe and {Gaudet}, Severin and {Technical Coordination Group}}, title = "{IVOA Architecture Version 2.0}", diff --git a/moc.tex b/moc.tex index 6ff7fd6..9fd4fd6 100644 --- a/moc.tex +++ b/moc.tex @@ -6,6 +6,7 @@ \usepackage[normalem]{ulem} \usepackage{fancyvrb} \usepackage{appendix} +\usepackage{amsmath} \lstloadlanguages{C} \lstset{flexiblecolumns=true,numberstyle=\small,showstringspaces=False, basicstyle=\ttfamily,defaultdialect=[latex]tex,language=tex} @@ -44,44 +45,69 @@ \newcommand\daniel[1]{\textcolor{blue}{\textbf{#1}}} \newcommand\pierre[1]{\textcolor{orange}{\textbf{#1}}} +\newcommand\fxp[1]{\textcolor{teal}{\textbf{#1}}} +%\newcommand\add[1]{\fxp{#1}} +% Switch the comment between the following lines to see/hide previous versions of the text +%\newcommand\remove[1]{\xout{#1}} +%\newcommand\replace[2]{\remove{#1} \add{#2}} +\newcommand\add[1]{#1} +\newcommand\remove[1]{} +\newcommand\replace[2]{\add{#2}} + +%\newcommand\addb[1]{\textcolor{purple}{\textbf{#1}}} +%\newcommand\removeb[1]{\textcolor{purple}{\xout{#1}}} +%\newcommand\replaceb[2]{\removeb{#1} \addb{#2}} +\newcommand\addb[1]{#1} +\newcommand\removeb[1]{} +\newcommand\replaceb[2]{\addb{#2}} + \title{MOC: Multi-Order Coverage map} \ivoagroup{Applications} +\author{Francois-Xavier Pineau (CDS)} \author{Pierre Fernique (CDS)} \author{Ada Nebot (CDS)} \author{Daniel Durand (CADC)} \author{Matthieu Baumann (CDS)} \author{Thomas Boch (CDS)} +\author{Baptiste Cecconi (LESIA, Observatoire de Meudon)} \author{Giuseppe Greco (EGO-Virgo)} \author{Tom Donaldson (STScI/NASA)} -\author{Francois-Xavier Pineau (CDS)} +\author{\addb{Thomas Robitaille}} \author{Mark Taylor (University of Bristol)} \author{Wil O'Mullane (Vera C. Rubin Observatory)} \author{Martin Reinecke (Max Planck)} \author{S\'{e}bastien Derri\`{e}re (CDS)} +\editor{\add{Francois-Xavier Pineau}} \editor{Pierre Fernique} \editor{Ada Nebot} \editor{Daniel Durand} - +\previousversion[http://ivoa.net/documents/MOC/20220727]{Version2.0} \previousversion[http://ivoa.net/documents/MOC/20191007]{Version1.1} \previousversion[http://ivoa.net/documents/MOC/20140602]{Version1.0} \begin{document} \begin{abstract} -This document describes the Multi-Order Coverage map method (MOC) -version 2.0 to specify arbitrary coverages for sky regions and/or time -coverages and potentially other dimensions. The goal is to be able to +This document \replace{describes}{defines version 2.1 of} the Multi-Order Coverage map \add{(MOC)} method \remove{(MOC)} +\remove{version 2.0} \replace{to specify}{, which enables the specification of} arbitrary coverages \replace{for}{across} sky regions +\replace{and/or time coverages and potentially other dimensions}{, time intervals and frequency domains}. + +The goal is to be able to provide a very fast comparison mechanism between coverages. The -mechanism is based on a discretization of space and time -dimensions. The system is based on the definition of a specific storage +mechanism \replace{is based}{relies} on \replace{a discretization of}{discretizing} \replace{space}{the unit sphere} and \replace{time +dimensions}{one-dimensional quantities such as time or frequency}. +\add{Specifically, the discretization uses the Z-order curve for 2-dimensional sky regions and on hierarchical dyadic subdivisions for 1-dimensional quantities.} +The system is based on the definition of a specific storage of the map coverage using predefined cells hierarchically grouped which makes it easy to produce and use for exploring astronomical -collections. There are already a few applications and libraries which -are taking advantage of this major evolution of the MOC standard. +collections. \remove{There are already a few applications and libraries which +are taking advantage of this major evolution of the MOC standard.} +\add{This version 2.1 extends the use of MOCs to the three physical quantities: +space, time, frequency, as well as their pairwise combinations.} \end{abstract} \listoffigures @@ -119,17 +145,22 @@ \section*{Conformance-related definitions} \end{figure} \section{Introduction} -This document is a major release of +This document is a \replace{major}{minor} release of the already existing encoding method recommendation \emph{Multi-Order - Coverage} map \citep[MOC~1.1,][]{2019ivoa.spec.1007F}. We generalize the MOC -originally limited to space dimension (Space MOC, SMOC) to the time + Coverage} map \citep[MOC~\replace{1.1}{2.0},][]{2022ivoa.spec.0727F}. +We generalize the MOC +originally limited to \replace{s}{S}pace \add{, Time, and Space-Time} dimension\add{s} + (Space MOC, SMOC\add{; Time MOC, TMOC; and Space-Time +MOC, STMOC}) to \replace{the time dimension (Time MOC, TMOC), and space and time dimensions (Space-Time -MOC, STMOC). Figure~\ref{fig:ivoadiagram} illustrates the role MOC2.0 +MOC, STMOC)}{the Frequency, Space-Frequency and Frequency-Time dimensions +(Frequency MOC, FMOC; Space-Frequency, SFMOC; Frequency-Time, FTMOC)}. +Figure~\ref{fig:ivoadiagram} illustrates the role MOC\replace{2.0}{ 2.1} plays within the IVOA architecture \citep{2021ivoa.spec.1101D}. The encoding method described in this document allows one to define -and manipulate space and time coverage in such a way that basic -operations like union, intersection, equality test can be performed +and manipulate space \replace{and}{,} time \add{and frequency} coverage in such a way that basic +operations like union, intersection\replace{,}{or} equality test can be performed very efficiently. This methodology allows VO applications and data servers to build efficient procedures to perform such operations on observations and catalogs. @@ -142,9 +173,13 @@ \section{The rationale} order to provide very fast union, intersection and equality operations between them. In order to accomplish this task, we based the system on a regular and hierarchical discretization as -exposed below. The standard MOC1.0 was limited to space, but for a +exposed below. +\replace{The standard MOC1.0 was limited to space, but for a multitude of use cases in astronomy we need the notion of time to be -properly integrated, e.~g.: +properly integrated}{The evolution of the MOC standard -- which was initially +limited to spatial quantities in version 1.0, then expanded to include temporal quantities in version 2.0, +and finally supplemented with frequency quantities in version 2.1 +-- enables it to cover a wide range of astronomical use cases}, e.~g.: \begin{itemize} \item What are the space and time coverages of the 2MASS observations and are there any observations which are coincidental with the HST @@ -160,11 +195,12 @@ \section{The rationale} by Galex? By both missions simultaneously? If so, are there detections within the source catalogues? \item Has Neptune been observed by DSS? +\item \add{Is a radio source contaminating my observations in given frequency ranges?} \end{itemize} -It was possible to answer those questions with MOC1.0 standard and +It was possible to answer those questions with MOC 1.0 standard and other VO tools but the amount of manipulation and computation was -quite a big hurdle for the researchers. With MOC2.0 it is possible +quite a big hurdle for the researchers. With MOC 2.0 it is possible to answer these questions in a few milli-seconds. Another example of usage would be the visual inspection of the spatio-temporal coverage of PanSTARRS observations (see Figure~\ref{fig:panstarrs}). @@ -174,7 +210,7 @@ \section{The rationale} \begin{figure}[!ht] \begin{center} \includegraphics[width=\textwidth]{panstarrs.png} -\caption{PanSTARRS observations and the associated spatial and + \caption[STMOC of PanSTARRS observations]{PanSTARRS observations and the associated spatial and temporal coverage within three different periods of time. The volume of the PanSTARRS MOC at a temporal resolution of about 17 minutes and spatial resolution of 52 arcsec is 320\,MB.} @@ -195,7 +231,7 @@ \subsection{Comparing the coverage of multiple data sets} \begin{center} \includegraphics[width=\textwidth]{stmoc_op.png} \end{center} -\caption{Intersection of HST ACS observations +\caption[Intersection of HST ACS and Saturn STMOCs]{Intersection of HST ACS observations and Saturn ephemeris Space-Time-MOC} \label{fig:stmoc-saturn} \end{figure} @@ -225,17 +261,17 @@ \subsection{Gravitational Wave localisations} which the sky localization is irregularly shaped with disjoint regions. This coding technique allows for an extra fast integration in the existing Virtual Observatory structures and tools\footnote{\url{https://emfollow.docs.ligo.org/userguide/}}. -The 2D contours of a GW sky localization can be visualised and -manipulated using Aladin Desktop, allowing one to compare them with +The 2D contours of a GW sky localization can be visualised +and manipulated using Aladin Desktop, allowing one to compare them with existing surveys, overlap sky map generations with increasing accuracy and computational cost and query the VizieR database. These sets of tasks can also be performed via Python using the astropy affiliated package mocpy\footnote{\url{https://cds-astro.github.io/mocpy/}}, efficiently displayed in javascript applications with Aladin Lite, and integrated within Jupyter notebooks through the -ipyaladin widget\footnote{\url{https://pos.sissa.it/357/031/pdf}}. +ipyaladin\footnote{\url{https://cds-astro.github.io/ipyaladin/}} widget\footnote{\url{https://pos.sissa.it/357/031/pdf}}. -Adding temporal information in a space MOC encoding can provide a +\remove{Adding temporal information in a space MOC encoding can provide a systematic approach in extracting information from follow-up campaigns involving tens of ground and space-based observatories in searching for gravitational-wave counterparts. @@ -249,18 +285,18 @@ \subsection{Gravitational Wave localisations} spatial coverage (bottom right panel). This approach permits us to depict the approximate timeline of the instruments involved in the observational campaign and place constraints on the emission properties during the -source evolution. +source evolution.} -\begin{figure}[!htbp] -\begin{center} -\includegraphics[width=\textwidth]{kilo.png} -\end{center} -\caption{A mock electromagnetic follow-up campaign - of a gravitational-wave sky localization over a time period (left). - A schematic kilonova light-curve with the observations temporal - coverage (top right) and associated spatial coverage (bottom right).} -\label{fig:gw} -\end{figure} +%\begin{figure}[!htbp] +%\begin{center} +%\includegraphics[width=\textwidth]{kilo.png} +%\end{center} +%\caption{A mock electromagnetic follow-up campaign +% of a gravitational-wave sky localization over a time period (left). +% A schematic kilonova light-curve with the observations temporal +% coverage (top right) and associated spatial coverage (bottom right).} +%\label{fig:gw} +%\end{figure} %\begin{figure}[!htbp] %\begin{center} @@ -303,10 +339,11 @@ \subsection{Space and Time MOC: Einstein Telescope and Early Warning Alerts} merger\footnote{\url{https://emfollow.docs.ligo.org/userguide/early_warning.html}}. -\subsection{Multi-site positional and temporal search} +\subsection{\replace{Multi-site positional and temporal search}{Multi-site search in space, time and/or frequency}} Often, a typical query from a virtual observatory (VO) user is to request all possible records from the VO at a given sky position -and/or at a given time. While this is in principle possible by +and/or at a given time \add{, and/or at a given frequency}. +While this is in principle possible by dispatching one narrow positional query to every registered Cone Search, SSA or SIA service and filtering by time, in practice, the number of queries required leads to an unacceptable load on both @@ -315,56 +352,105 @@ \subsection{Multi-site positional and temporal search} region. If basic footprint/coverage information was available for all registered services, for instance using VODataService 1.2 \citep{2021ivoa.spec.1102D} only those with coverage in the region -and time of interest would then be queried. This would provide a great +and time\add{/frequency} of interest would then be queried. This would provide a great reduction in the number of services to be queried optimizing the response time. Using the MOC offers the opportunity to provide this coverage information in a uniform way. The MOC could be stored locally for a given service or centrally where the coverage for a number of services would be supported. +\subsection{\add{Low frequency radio astronomy}} + +\add{Ground-based low-frequency radio astronomy instruments typically span over +a wide spectral range, such as from 10 MHz to 85 MHz, with NenuFAR +\citep{zarka_2028}, which covers a factor of 8.5 in spectral domain. +The main beam of the instrument has a field of view that varies by almost one +order of magnitude. Due to its interferometric nature, the shape of the side +lobes varies in frequency. +Astronomers preparing observations with NenuFAR need to identify and assess +any contaminating radio sources that may be visible in the side lobes of +the instrument. +The current NenuFAR Python library \citep[nenupy,][]{loh_2025} implements this +assessment by cross-matching a series of spatial MOCs (for the grating lobes +and the interfering bright radio sources) with a temporal and spectral sampling. +Merging the spatial and spectral dimensions into an spatio-frequency MOC +for a series of observation times would simplify and speed up this task.} + +\add{The low-frequency radio sources observed in our solar system, +such as the Sun or the magnetised planets, but also those of active stars +and some exoplanetary systems, are showing very dynamical structures in the +temporal-spectral domain, at time scales spanning from milliseconds to hours. +The TFCat \citep[Time-Frequency catalogue,][]{cecconi_2023} has been developed +to represent observed features in the temporal-spectral domain, using, e.g., +polygons in physical coordinates. +While this framework proved to be very efficient to store, share, and display +those features, cross-matching the catalogues is still a difficult process. +Transforming each of the catalogues into an frequency-time MOC would +allow very efficient identification of possibly simultaneous events listed +in different catalogues.} + \section{MOC principles} -The MOC standard is defined using four basic building blocks: +\add{Irrespective of the quantity,} the MOC standard is defined using four basic building blocks: discretization, unique reference system, hierarchization and efficient encoding: \begin{enumerate} \item Determine a proper tessellation/discretization methodology for - each dimension axis (space, time, ...); -\item Fix a unique referential system for each dimension, to avoid + each \replaceb{dimension axis}{quantity} (space, time, \add{frequency}, ...); +\item Fix a unique referential system for each \replaceb{dimension}{quantity}, to avoid reference conversions and thus allowing to easily compare different data collections; \item Use an hierarchical procedure and a unique representation - (canonical form) for compacting and quickly manipulating each axis - coverage at any level of accuracy; + (canonical form) for compacting and quickly manipulating \removeb{each axis} + coverage\addb{s of a given quantity} at any level of accuracy; \item Implement at least one serialization in a binary encoding format (other serializations are possible, e.g. ASCII). \end{enumerate} With these principles, a MOC consists of a list of numbers which represent the indices of the cells mapping the coverage of the spatial -or temporal axis. +\replace{or}{, } temporal \add {or frequency} axis. As soon as the consecutive cells are used at order n, they will be hierarchically grouped in their parent cell at order n-1, and this recursively. This introduces the notion of orders and -associated cell index. -The cell boundary alignment implied by the hierarchical structure +associated cell index boundary alignment implied by the hierarchical structure facilitates the combination of cells at different orders. -To work efficiently on existing hardware, we encode of any pair + +\remove{To work efficiently on existing hardware, we encode of any pair (order, index) as a long integer (64 bits), and we reserve the two most significant bits to encode the type of MOC (spatial, temporal, or future usages). The earlier MOC standard was limited to spatial coverage. We are reusing these principles to manipulate temporal coverages, as well as space-time coverages where we can manipulate the two -physical dimensions simultaneously. +physical dimensions simultaneously.} + + +%To work efficiently on existing hardware, we encode of any pair +%(order, index) as a \replace{long}{64-bits} integer \rm{(64 bits)}, and we reserve the two +%most significant bits to encode the type of MOC (spatial, temporal, +%\add{frequency} or \replace{future usages}{custom}). +%The earlier MOC standard was limited to spatial\add{, time, and space-time} coverage\add{s}. +%We are reusing these principles to manipulate \replace{temporal}{frequency} coverages, +%as well as \replace{space-time}{space-frequency and frequency-time} coverages where we can manipulate the two +%physical dimensions simultaneously. \medskip \par\noindent -We will now explain the conventions chosen for the spatial and temporal axis. +We will now explain the conventions chosen for \rm{the spatial, temporal and frequency axis} +\add{single quantity MOCs, namely space, time and frequency MOCs} +\replaceb{designed}{designated} \add{by the acronyms SMOC, TMOC and FMOC. +Then, we also explain the conventions chosen for two-dimensional MOCs, +namely space-time, space-frequency and frequency-time MOCs} +\replaceb{designed}{designated} \add{by the acronyms STMOC, SFMOC and FTMOC.} + +\subsection{\add{Single quantity MOCs}} -\subsection{Space MOC conventions} +\add{In this section, we describe the convention relative to single quantity MOCs.} + +\subsubsection{Space MOC conventions} Defining a sky region by a subset of regular sky tessellation or tiles is not a new idea. In astronomy, one could find three main methods of partitioning the sphere : Q3C, HTM and HEALPix which are respectively @@ -386,12 +472,13 @@ \subsection{Space MOC conventions} \item Accuracy: HEALPix provides libraries which allow the calculation up to accuracy of 0.4 mas (order 29) (\url{http://sourceforge.net/projects/healpix/}); \item Standard: Existence of many HEALPix libraries: C++, Java, - Fortran, IDL... Also, HEALPix was selected for several all sky - missions such as WMAP, Planck and Gaia. The HEALPix main web site is - located at Jet Propulsion Lab (\url{http://healpix.sourceforge.io/}). + Fortran, \replace{IDL}{Python, Rust, Web Assembly, }... + Also, HEALPix was selected for several all sky + missions such as WMAP, Planck and Gaia. The HEALPix main web site \replaceb{is}{, previsously} + located at Jet Propulsion Lab\addb{oratory}\addb{, is now on sourceforge} (\url{http://healpix.sourceforge.io/}). \end{itemize} -The HEALPix \citep{2005ApJ...622..759G} tessellation technique divides +The HEALPix \citep{2005ApJ...622..759G} \add{NESTED} tessellation technique divides the sphere into 12 cells, each of them sub-divided into 4 cells recursively (see Fig.~\ref{fig:healpix}). Thus the sphere at order 1 will consist of 48 cells, @@ -406,7 +493,34 @@ \subsection{Space MOC conventions} \label{fig:healpix} \end{figure} -HEALPix allows three coordinate systems: galactic, equatorial and +% MOVED HERE FROM v2.1-v2 +\add{In practice, the sub-division of each base cell -- +which are squares in the Euclidean HEALPix projection plane -- consists in +iterations of a Z-order curve, and bit interleaving allows for +a non-recursive algorithm.} +\addb{When encoding a "NESTED" HEALPix index, the four Most Significant Bits (MSB) are used to code +the base cell index ($ \in [0, 12[)$ and the remaining LSBs are the 2-dimensional Morton code of the +coordinates inside the base cell.} + +% MOVED HERE FROM v2.1-v2 +\replace{This}{Thus, the "NESTED"} numbering consists of enumerating all cells in a +specific order. For instance, at order 1, there are 48 cells (12x4) +enumerated from 0 to 47. In this scheme, the 4 sub-cells of cell M +have the indices: (M$\times4)+3$, (M$\times4)+1$, (M$\times4)+2$, +(M$\times4$) in reading +order. And reciprocally, the parent index of cell N is +N/4. Each SMOC cell is coded by a pair of numbers: +(order, index) which are the HEALPix order and the HEALPix index in this +order. + +\begin{figure}[h] +\begin{center} +\includegraphics[width=\textwidth]{nested_healpix.jpg} +\end{center} +\caption{HEALPix numbering principle} +\end{figure} + +\remove{HEALPix allows three coordinate systems: galactic, equatorial and ecliptic. Allowing various coordinate systems would limit the possibility to compare efficiently SMOCs. There is indeed no equivalence between an HEALPix cell described in a given coordinate @@ -414,14 +528,14 @@ \subsection{Space MOC conventions} coordinate system. Consequently, the SMOC definition is expressed in equatorial coordinate using the ICRS reference. This choice has been motivated by looking at most catalogs and realizing that most of them -are using equatorial coordinates. +are using equatorial coordinates.} To support the encoding based on 64-bit longs, the best resolution available is provided at order 29 and according to the HEALPix equations, corresponds approximately to 0.4mas. The SMOC resolution is set by the maximum value of the HEALPix order used to define a region. Its selection depends on the accuracy chosen by the provider -to define the region. As data set boundaries are not aligned with the +to define the region. As data set boundaries are not aligned with the HEALPix cell borders, a SMOC is generally an upper-approximation of the data set coverage. The quality of this approximation depends directly on the chosen SMOC resolution (MOCORD\_S). @@ -435,7 +549,7 @@ \subsection{Space MOC conventions} \begin{center} \includegraphics[width=\textwidth]{smoc_view.png} \end{center} -\caption{SMOC: from the image +\caption[SMOC from an image]{SMOC: from the image to the list of numbers based on HEALPix hierarchy tessellation} \label{fig:smoc-view} \end{figure} @@ -443,7 +557,24 @@ \subsection{Space MOC conventions} \input{table1.tex} \input{table2.tex} -\subsection{Time MOC conventions} +\replaceb{The mechanism used to determine which HEALPix cell contains a given +The formula computing a NESTED HEALPix index provided a given $order$ +and}{The formula computing a cell index from a} sky location is described in +the main article defining the HEALPix system \citep{2005ApJ...622..759G}. +\replaceb{Several support libraries supporting the most important set of primitives are +already available. +These libraries are required if one wants to generate SMOCs, and also +wants to compare them with sky coordinates. Though please +note that these libraries do not have built-in support to performing +basic SMOC arithmetics.}{Several support libraries implementing the most +important primitives are already available. These libraries are required +for generating SMOCs and for comparing them with sky coordinates. +However, they do not provide built-in support for performing basic +SMOC arithmetic operations.} + + + +\subsubsection{Time MOC conventions} \label{sec:tmoc1} @@ -460,53 +591,251 @@ \subsection{Time MOC conventions} 1$\mu$s resolution, enough for most astronomical time events. At the deepest order (61) the TMOC cell number is the number of $\mu$s since JD=0. - -The time is a relative observation, and depends on the position of the +\addb{Hence, ab integer time value $t$ in $\mu$s since JD=0 is valid if +$t \in [0, 2^{62}[$. For a given $order$, its cell $index$ is: +\begin{equation} + index = \frac{t}{2^{order}} + \label{eq:tmoc:i} +\end{equation} +Conversly, the range of time values $[t_{min}, t_{max}[$, in $\mu$s since JD=0, +of a given $index$ of order $order$ is given by +\begin{align} + t_{min} &= index \times 2^{order} \\ + t_{max} &= (index + 1) \times 2^{order} + \label{eq:tmoc:t} +\end{align}. +} + +% MOVED HERE fom v2.1-v2 +\addb{We note that two consecutive cells at order N with the indices +(M$\times$2)+1, (M$\times$2) will be coded at order N-1 with the index +M and thus recursively. The order N-1 cell duration is 2 times more +than the N cell duration (61: 1 $\mu$s, 60: 2 $\mu$s, 59: 4 $\mu$s, etc...).} + +\remove{The time is a relative observation, and depends on the position of the observer. There are many time scales for measuring time: Terrestrial Time (TT), Barycentric Coordinate Time (TCB), Geocentric Coordinate Time (TCG), Ephemeris Time (ET), Barycentric Dynamic Time (TDB), International Atomic Time (AI), etc. We opt for the TCB -reference (see \cite{2015A&A...574A..36R} for details). Our -choice is motivated by the fact that this system is linear by -construction and has been adopted by numerous missions such as Gaia. +reference} (see \cite{2015A&A...574A..36R} for details). +\remove{Our choice is motivated by the fact that this system is linear by +construction and has been adopted by numerous missions such as Gaia.} + +% MOVED HERE fom v2.1-v2 +TMOC {\bf must} be based on \removeb{JD system,} the time +scale TCB (see section~\ref{sec:tcb}) and the Solar System Barycenter as the reference +position \citep[see also][]{2019ASPC..523..497F} and section~\ref{sec:tmoc1}. +%The best resolution supported by TMOC is $1 \mu$s. It may be necessary to convert the temporal events to the chosen scale. If the ephemeris required for this conversion are not available, opt to degrade the accuracy of the time measurement (typically around 20 minutes for observations within the Earth orbit environment to cover all possible observer positions). -Table~\ref{table:tmocorders} is showing some time values at a given order. +Table~\ref{table:tmocorders} +\replaceb{is showing some time values at a given order}{shows the time resolutions of the cells for each possible order}. In Figure~\ref{fig:tmoc-view} we show the creation of TMOC, from a time series to the list of numbers based on time discretization. -\begin{figure}[h] +\begin{figure}[h!] \begin{center} \includegraphics[width=\textwidth]{tmoc_view.png} \end{center} -\caption{TMOC: from the +\caption[TMOC from time series]{TMOC: from the time series to the list of numbers based on time discretization} \label{fig:tmoc-view} \end{figure} -\subsection{Space and Time MOC conventions} -To respond to the different use cases presented at the beginning of +\subsubsection{\add{Frequency MOC conventions}\label{sec:fmoc1}} + +\add{As for Time, representing a frequency coverage requires defining \emph{a priori} +the total frequency range to be described by the notation. However, unlike Time -- +which is expressed as milliseconds since a reference epoch and therefore has a constant +absolute resolution (1 millisecond) -- frequency values are represented as floating-point +numbers with an approximately constant \emph{relative} resolution (i.e. a fixed number of significant digits), +rather than a constant absolute resolution.} + +\add{The adopted\footnote{To handle floating-point quantities, two approaches have been proposed. +The first implemented solution is described in Appendix~\ref{sec:apx:fmoc1}. +It was later replaced by the more WCS-compliant solution presented above.} approach is fully generic. +First, we define the global valid frequency range $[f_{min}, f_{max}[$. +The frequency $f$ {\bf must} be expressed in $Hz$. +Next, we introduce a bijective mapping +\[ +g: [f_{min}, f_{max}[ \rightarrow [0,1[ +\] +such that any frequency $f \in [f_{min}, f_{max}[$ is associated with the dyadic interval containing $g(f)$.} + +\add{For a given $order$, the interval $[0,1[$ is subdivided into $2^{order+1}$ dyadic intervals (the MOC cells), +indexed from $0$ to $2^{order+1}-1$. +The index corresponding to a frequency $f$ at order $order$ is therefore:} + +\add{\begin{equation} + index = \lfloor g(f) \times 2^{order+1} \rfloor + \label{eq:fmoc:i} +\end{equation}} + +\add{The adopted bounds of the valid frequency range are: +\begin{itemize} + \item $f_{min} = 10^{-18}$ Hz, corresponding approximately to the inverse of the age of the Universe. + \item $f_{max} = 10^{+38}$ Hz, a somewhat arbitrary upper limit chosen so that the full range spans 56 decades in logarithmic scale. +\end{itemize}} + +\add{The mapping function $g$ is defined as a linear interpolation in logarithmic space: +\begin{equation} + g(f) = \frac{\log_{10}(f) - \log_{10}(f_{min})}{\log_{10}(f_{max}) - \log_{10}(f_{min})} + %= \frac{\ln(f) - \ln(f_{min})}{\ln(f_{max}) - \ln(f_{min})} + \label{eq:fmoc:g} +\end{equation}} + +\add{Combining Eq.~(\ref{eq:fmoc:i}) and Eq.~(\ref{eq:fmoc:g}) with the selected frequency range, +we obtain, for a given $order$: +\begin{equation} + index = \left\lfloor + \frac{\log_{10}(f) + 18}{56} \times 2^{order+1} + \right\rfloor + %= \left\lfloor + %\frac{\ln(f \times 10^{18})}{56 \ln(10)} \times 2^{order+1} + %\right\rfloor +\end{equation}} + +\add{The number of consecutive integers that can be exactly represented by an IEEE~754 floating-point +is limited by the number of bits in the mantissa. +For double-precision floating-points, this number is 52.} + +\add{Notably, for a fixed exponent, the fractional part of a floating-point number corresponds to +a dyadic rational. Consequently, there is no benefit in defining a MOC $order$ with more cells +than the number of dyadic rationals representable with 52 bits. Beyond this limit, the cells +would be smaller than the intrinsic precision of double-precision floating-point numbers.} + +\add{For this reason, the maximum order is limited to 52 levels, ranging from 0 to +$order_{max} = 51$. +At the maximum resolution, the index corresponding to a frequency $f$ is therefore computed as: +\begin{align} + order_{max} &= 51 \\ + index_{51} &= \left\lfloor + \frac{\log_{10}(f) + 18}{56} \times 2^{52} + \right\rfloor +\end{align}} + +\add{Conversely, the frequency range $[f_{min}, f_{max}[$ corresponding to a given index $index$ at +a given order $order$ can be obtained using the inverse transformation: +\begin{eqnarray*} + f_{min} & = & g^{-1}(\frac{index}{2^{order + 1}}) = 10^{ 56 \times \frac{index}{2^{order + 1}} - 18} \\ + f_{max} & = & g^{-1}(\frac{index + 1}{2^{order + 1}}) = 10^{ 56 \times \frac{index + 1}{2^{order + 1}} - 18} +\end{eqnarray*}} + +\add{Choosing a given order is equivalent to selecting the number of mantissa bits used to represent the mapped value +in the interval $\in [0, 1[$. +In practice, this corresponds to choosing a level of numerical precision, analogous to selecting a +number of significant bits (or digits), noting that approximately one decimal digit of precision is +lost for every 3 to 4 mantissa bits removed. +The approximate number of significant decimal digits corresponding to a given order can be estimated using +the following relation: +\begin{equation} + n = \frac{(order + 1)}{\log_2(10)} +\end{equation} +This expression follows from the conversion between binary and decimal precision, +since one decimal digit corresponds to approximately $\log_2(10) \approx 3.32$ bits. +For example, a frequency stored using a 32-bit floating-point number provides about +24 bits of significand precision, corresponding to approximately 7 decimal digits. +To obtain a comparable precision in the hierarchical representation, an order of 29 can be chosen. +Additional values are provided in Table~\ref{table:fmocorders}. +} + +\add{We can also express, for a given order $order$ the relative size of cell using $g^{-1}$: +\begin{equation} + \frac{\Delta f}{f} = \frac{g^{-1}(\frac{index + 1}{2^{order + 1}}) - g^{-1}(\frac{index}{2^{order + 1}})}{g^{-1}(\frac{index}{2^{order + 1}})} = 10^\frac{56}{2^{order + 1}} - 1 +\end{equation} +} + +\input{tablef.tex} + +\addb{Unlike time, the finest resolution is not constant across the full frequency range. +Because frequencies are represented using 64-bit floating-point values, +the achievable precision is \textit{relative} rather than \textit{absolute}. +At the highest resolution (order 51), the relative precision is approximately $10^{-15}$, +which is close to the number of significant decimal digits provided by double-precision +floating-point numbers.} + +\addb{Like for time, two consecutive cells at order $N$, with indices +$(M\times 2)$ and $(M\times 2)+1$, are represented at order $N-1$ +by a single cell with index $M$. +This hierarchical structure applies recursively across all orders.} + +\addb{Considering the dyadic structure of the intervals, a cell at order +$N-1$ spans twice the frequency range of a cell at order $N$ +Consequently, the frequency resolution at order +$N-1$ is half that at order $N$.} + + +%\paragraph{\add{Remark}} +% +%\add{To handle floating-point quantities, two approaches have been proposed. +%The first implemented solution is described in Appendix~\ref{sec:apx:fmoc1}. +%It was later replaced by the more WCS-compliant solution presented above.} + +\subsection{\add{Two dimensional MOCs}} + +\add{In this section, we describe the convention adopted\footnote{The use of +a single index obtained by interleaving the indices of two quantities, +similarly to a 2D Z-order curve, was considered but ultimately discarded. +First, the resolutions of both quantities would have been coupled, making it impossible, for example, +to describe an observational coverage with low temporal resolution and high spatial resolution. +Second, encoding two 64-bit indices into a single 64-bit index would have reduced the maximum +achievable resolution for each quantity.} +for two-dimensional MOCs, +i.e. MOCs describing the coverage of two quantities at the same time.} + +\add{To respond to the different use cases presented at the beginning of +the document, independent single quantity MOCs are not enough. +We need to link two quantities in a global mechanism. +For example, we need to be able to select a SMOC using a time or a frequency window. +We also need to select a TMOC or a FMOC using a spatial constraint. +Implementing this linkage would allow +the potential users to select and interact with the astronomical +collections which support space, time and/or frequency and use logical +operators between them.} + +\add{Our approach combines two quantities -- time and space, frequency and space, or time and frequency -- +by associating sub-MOCs of the first quantity with MOCs of the second quantity. +For the first quantity, the range of deepest-order indices bounding a sub-MOC +must not overlap with the one of another sub-MOC. +For the second quantity, MOCs are independent and may overlap. +In the various serializations, a 2D-MOC is represented as a set of tuples, +each consisting of a sub-MOC of the first quantity and a MOC of the second quantity. +Although the resolutions of the first and second quantities may differ, the resolution of each quantity remains uniform across the entire 2D-MOC.} + +\addb{In the remaining of the document we adopt the following convention: +a $YX$MOC is a 2D-MOC in which $X$ is the first quantity and $Y$ the second one.} + +%\add{Remark: the use of a single index obtained by interleaving the indices of two quantities, +%similarly to a 2D Z-order curve, was considered but ultimately discarded. +%First, the resolutions of both quantities would have been coupled, making it impossible, for example, +%to describe an observational coverage with low temporal resolution and high spatial resolution. +%Second, encoding two 64-bit indices into a single 64-bit index would have reduced the maximum +%achievable resolution for each quantity.} + +\subsubsection{Space and Time MOC conventions\label{sec:stmoc}} + +\remove{To respond to the different use cases presented at the beginning of the document, the SMOC and TMOC independently are not enough. We need to link the two dimensions in a global mechanism. In other words, we need to be able to select the SMOC using a time window or to select a TMOC using a spatial constraint. Implementing this linkage would allow the potential users to select and interact with the astronomical collections which support space and time and use logical -operators between them. +operators between them.} -Our approach is to combine these two dimensions - time and space - by +\remove{Our approach is to combine these two dimensions - time and space - by associating each time period (coded according to the TMOC convention) with its spatial region (coded according to the SMOC convention). For that, we interleave the information of time coverage -with the information of space coverage for this period. +with the information of space coverage for this period.} -This two-dimensional interleaving approach has the advantage of making +\remove{This two-dimensional interleaving approach has the advantage of making the resolutions chosen for time and for space independent. For instance, it is possible to describe observation coverage with a low resolution for time while using a high spatial @@ -516,102 +845,174 @@ \subsection{Space and Time MOC conventions} We thus proposed the interleaving algorithm which allows us to define and manipulate high resolution STMOCs of reasonable sizes for fast algorithms (see Appendix~\ref{app:perf} for STMOC performance). +} + +\add{In Space-Time MOCs (STMOCs), the first quantity is time and the second quantity is space. +Consecutive time ranges associated with the same spatial coverage +are grouped into a sub-TMOC linked to their common SMOC. +The STMOC is represented as a sorted set of (sub-TMOC, SMOC) pairs.} + + In Figure~\ref{fig:stmoc-view} we show the visual representation of an STMOC, -in which at a given TMOC range we obtain the corresponding SMOC. +in which at a given \replace{TMOC range}{sub-TMOC} we obtain the corresponding SMOC. + \begin{figure}[h] \begin{center} \includegraphics[width=0.8\textwidth]{stmoc_view.png} \end{center} -\caption{STMOC visual representation in which +\caption[Visual representation of a STMOC]{STMOC visual representation in which at a given TMOC range we obtain the corresponding SMOC} \label{fig:stmoc-view} \end{figure} -\section{SMOC and TMOC encoding} -The encoding described in this section guarantees backward compatibility -with MOCs corresponding to previous versions of this standard. - -\subsection{Space MOC or SMOC} -As introduced above, the SMOC {\bf should} be based on the -HEALPix tessellation of the sphere, expressed in the ICRS coordinate -reference system for celestial coverages. This document does not describe -the use of SMOC outside celestial scope. However, it is possible to use -SMOC for other coverages, such as planetary coverages. The definition of -the unique reference for each body will have to be defined. Two complementary -encoding formats are defined: a string serialization based on ASCII and a -binary format based on FITS. - -\subsubsection{Numbering} - -The numbering scheme used in SMOC for specifying the cell indices {\bf - must} follow the "NESTED" HEALPix numbering schemes -\citep{2005ApJ...622..759G}. This numbering consists of enumerating all cells in a -specific order. For instance, at order 1, there are 48 cells (12x4) -enumerated from 0 to 47. In this scheme, the 4 sub-cells of cell M -have the indices: (M$\times4)+3$, (M$\times4)+1$, (M$\times4)+2$, -(M$\times4$) in reading -order. And reciprocally, the parent index of cell N is -N/4. Each SMOC cell is coded by a pair of numbers: -(order, index) which are the HEALPix order and the HEALPix index in this -order. - -\begin{figure}[h] -\begin{center} -\includegraphics[width=\textwidth]{nested_healpix.jpg} -\end{center} -\caption{HEALPix numbering principle} -\end{figure} - -{\bf Note:} The order 0 is a special case, it contains only 12 -cells enumerated from 0 to 11. - - -\subsubsection{Sky coordinates} -The mechanism used to determine which HEALPix cell contains a given -sky location is described in the main article defining the HEALPix -system \citep{2005ApJ...622..759G}. Several support libraries -supporting the most important set of primitives are already available. -These libraries are required if one wants to generate SMOCs, and also -wants to compare them with sky coordinates. Though please -note that these libraries do not have built-in support to performing -basic SMOC arithmetics. - -\subsection{Time MOC or TMOC} -As introduced above, TMOC {\bf must} be based on JD system, the time -scale TCB, and the Solar System Barycenter as the reference -position \citep[see also][]{2019ASPC..523..497F} and section~\ref{sec:tmoc1}. -The best resolution supported by TMOC is $1 \mu$s. - -In the case that the time scale and the time reference position are unknown, -we recommend to set the time resolution of the generated TMOC to order -31, e.g. about 1000 seconds (see Table~\ref{table:tmocorders}) corresponding -to about twice the light travel time correction between the Earth and -the Solar Barycenter. Please refer to the VO note on TIMESYS for more -information about this limitation \citep{timesysnote}. - - -\subsubsection{Numbering} -The numbering scheme used in TMOC for specifying the time cell indices -{\bf must} reuse a similar hierarchical principle as for the SMOC -with the difference that the time line has only one dimension, so the -hierarchical progression uses a factor of 2 instead of 4, and there is -no need to use a HEALPix mapping. - -TMOC has 62 orders, and at the best -resolution (order 61), a time event will be coded by the integer value -representing the number of $\mu$s of this event since JD=0. - -Two consecutive cells at order N with the indices - (M$\times$2)+1, (M$\times$2) will be coded at order N-1 with the index - M and thus recursively. The order N-1 cell duration is 2 times more - than the N cell duration (61: 1 $\mu$s, 60: 2 $\mu$s, 59: 4 $\mu$s, etc...). - -\subsection{Serialization} +\subsubsection{\add{Space and Frequency MOC conventions}} + +\add{In some use cases, it is desirable to combine the spatial and frequency dimensions. +Similarly to STMOCs -- see section~\ref{sec:stmoc} -- SFMOCs allow the selection of a SMOC +within a given frequency window -- or more generally, within a FMOC. +Conversely, they can also be used to derive a FMOC subject to spatial constraints. +The mechanism is identical to that described in the previous section; +the only difference lies in the nature of the quantities involved.} +\addb{Thus, in Space-Frequency MOCs (SFMOCs), the first quantity is the frequency +and the second quantity is space.} +\add{The type of quantity is specified by a prefix in the ASCII representation or +by a keyword in the FITS serialization.} + +\subsubsection{\add{Frequency and Time MOC conventions}} + +\add{FTMOCs are used to combine the frequency and time dimensions. +They use the same convention as STMOCs and SFMOCs, with the appropriate quantities prefix +in the ASCII serialization.} +\addb{Thus, in Frequency-Time MOCs, the first quantity is the time +and the second quantity is the Frequency.} + + +\section{\replace{SMOC and TMOC encoding}{MOC serializations}} + +\removeb{The encoding described in this section guarantees backward compatibility +with MOCs corresponding to previous versions of this standard.} +\addb{Two complementary serialization formats are defined for all MOCs: +a string serialization based on ASCII and a binary format based on FITS.} + +\subsection{\replace{SMOC and TMOC encoding}{MOC serializations for a single quantity}} + +%\subsection{Space MOC \remove{or SMOC}} +% +%\remove{As introduced above, the SMOC {\bf should} be based on the +%HEALPix tessellation of the sphere, expressed in the ICRS coordinate +%reference system for celestial coverages. This document does not describe +%the use of SMOC outside celestial scope. However, it is possible to use +%SMOC for other coverages, such as planetary coverages. The definition of +%the unique reference for each body will have to be defined.} +%\remiveb{Two complementary encoding formats are defined: a string serialization based +%on ASCII and a binary format based on FITS.} + +%\subsubsection{Numbering} + +%The numbering scheme used in SMOC for specifying the cell indices {\bf +% must} follow the "NESTED" HEALPix numbering schemes +%\citep{2005ApJ...622..759G}. + +%\add{In a "NESTED" HEALPix index, the four Most Significant Bits (MSB) are used to code the base cell index +%($ \in [0, 12[)$ and the remaining LSBs are the 2-dimensional Morton code of the coordinates inside the base cell. +%Said differently, an HEALPix number comes from the concatenation of 12 Z-order curves, one curve per base cell.} + +%\replace{This}{Thus, the "NESTED"} numbering consists of enumerating all cells in a +%specific order. For instance, at order 1, there are 48 cells (12x4) +%enumerated from 0 to 47. In this scheme, the 4 sub-cells of cell M +%have the indices: (M$\times4)+3$, (M$\times4)+1$, (M$\times4)+2$, +%(M$\times4$) in reading +%order. And reciprocally, the parent index of cell N is +%N/4. Each SMOC cell is coded by a pair of numbers: +%(order, index) which are the HEALPix order and the HEALPix index in this +%order. +% +%\begin{figure}[h] +%\begin{center} +%\includegraphics[width=\textwidth]{nested_healpix.jpg} +%\end{center} +%\caption{HEALPix numbering principle} +%\end{figure} +% +%{\bf Note:} The order 0 is a special case, it contains only 12 +%cells enumerated from 0 to 11. + + +% \subsubsection{Sky coordinates} +% The mechanism used to determine which HEALPix cell contains a given +% sky location is described in the main article defining the HEALPix +% system \citep{2005ApJ...622..759G}. Several support libraries +% supporting the most important set of primitives are already available. +% These libraries are required if one wants to generate SMOCs, and also +% wants to compare them with sky coordinates. Though please +% note that these libraries do not have built-in support to performing +% basic SMOC arithmetics. + + +%\subsection{Time MOC \remove{or TMOC}} +%\remove{As introduced above, }TMOC {\bf must} be based on JD system, the time +%scale TCB (see section~\ref{sec:tcb}) and the Solar System Barycenter as the reference +%position \citep[see also][]{2019ASPC..523..497F} and section~\ref{sec:tmoc1}. +%The best resolution supported by TMOC is $1 \mu$s. +% +%\removeb{In the case that the time scale and the time reference position are unknown, +%we recommend to set the time resolution of the generated TMOC to order +%31, e.g. about 1000 seconds (see Table~\ref{table:tmocorders}) corresponding +%to about twice the light travel time correction between the Earth and +%the Solar Barycenter. Please refer to the VO note on TIMESYS for more +%information about this limitation} \citep{timesysnote}. +% +%\subsubsection{Numbering} +%The numbering scheme used in TMOC for specifying the time cell indices +%{\bf must} reuse a similar hierarchical principle as for the SMOC +%with the difference that the time line has only one dimension, so the +%hierarchical progression uses a factor of 2 instead of 4, and there is +%no need to use a HEALPix mapping. +% +%TMOC has 62 orders, and at the best +%resolution (order 61), a time event will be coded by the integer value +%representing the number of $\mu$s of this event since JD=0. +% +%Two consecutive cells at order N with the indices +% (M$\times$2)+1, (M$\times$2) will be coded at order N-1 with the index +% M and thus recursively. The order N-1 cell duration is 2 times more +% than the N cell duration (61: 1 $\mu$s, 60: 2 $\mu$s, 59: 4 $\mu$s, etc...). + +%\subsection{\add{Frequency MOC}} +% +%\add{As introduced in section~\ref{sec:fmoc1}, the quantity manipulated by a FMOC +%{\bf must} be a frequency, expressed in $Hz$.} +% +%\add{FMOCs supports 52 orders, conresponding to the number of explicitly stored +%mantissa bits in an IEEE 754 double-precision binary floating-point number.} +% +%\add{Unlike time, the finest resolution is not constant across the full frequency range. +%Because frequencies are represented using 64-bit floating-point values, +%the achievable precision is \textit{relative} rather than \textit{absolute}. +%At the highest resolution (order 51), the relative precision is approximately $10^{-15}$, +%which is close to the number of significant decimal digits provided by double-precision +%floating-point numbers.} +% +%\add{Two consecutive cells at order $N$, with indices +%$(M\times 2)$ and $(M\times 2)+1$, are represented at order $N-1$ +%by a single cell with index $M$. +%This hierarchical structure applies recursively across all orders.} +% +%\add{Considering the dyadic structure of the intervals, a cell at order +%$N-1$ spans twice the frequency range of a cell at order $N$ +%Consequently, the frequency resolution at order +%$N-1$ is half that at order $N$.} + + +%\subsection{Serialization} A MOC can be manipulated and serialized either as a list of cell numbers for each order (hierarchical view), or as a list of intervals -at the deepest order (range view). These two methods are used for -SMOC, TMOC and STMOC serializations and are presented below. +at the deepest order (range view). +\add{While} \replace{T}{t}hese two methods are used for +SMOCs, \replace{TMOC and STMOC serializations and are presented +below}{only the intervals approach is used for other MOCs}. +\add{The various possible MOC} serializations are presented below. \subsubsection{Binary serialization} To encode a MOC in a FITS file, each MOC pair (order, index) {\bf @@ -652,14 +1053,21 @@ \subsubsection{Binary serialization} be in ascending numerical order. The resulting list is stored in a single-column binary table extension. +\paragraph{\add{Remark on RANGE vs UNIQs packing}} + +\add{For one-dimensional quantities such as time or frequency, the RANGE representation +is often\footnote{Except for MOCs that would be dominated by disjoint cells} more compact +than any UNIQ-based representation. +In one dimension, contiguous cells can always be merged into a single continuous range, +a property that does not generally extend to higher dimensions.} \paragraph{Backward compatibility} -RANGE packaging has been introduced for MOC2.0. +RANGE packaging has been introduced for MOC\replace{2.0}{ 2.0}. This method is generally faster than the previous one for reading or writing a MOC because the internal representation of MOC in memory is often range oriented. However, we recommend to use the first method for SMOC for compatibility with existing libraries -not yet compatible with MOC2.0. +not yet compatible with MOC\replace{2.0}{ 2.0}. \subsubsection{ASCII serialization} To encode a MOC as a string each MOC pair (order, index) {\bf must} be written @@ -672,8 +1080,16 @@ \subsubsection{ASCII serialization} The usage of a range operator is allowed in the list of indices using the dash ("-": decimal ASCII code 45) as a separator: lowindex-highindex. -The list of cells {\bf must} be well-formed, and the values {\bf must} -be in ascending numerical order. +The list of cells {\bf must} be well-formed\replaceb{,}{.} +\addb{Both the orders} and \removeb{the} values {\bf must} +be \addb{sorted} in ascending numerical order\addb{, comparing first the order and +then the cell value}. +\addb{This guarantees the same cell orderering as in the UNIQ FITS serialization, +enabling easy conversions in streaming mode. +Although these rules are intentionally strict, implementations are +encouraged to follow the robustness principle: +be conservative in what they generate and liberal in what they accept.} + If the best resolution of the MOC (moc order) is greater than the greatest stored order, the moc order {\bf must} be provided, followed @@ -707,7 +1123,10 @@ \subsubsection{ASCII serialization} \begin{lstlisting}[] smoc ::= 's'? moc tmoc ::= 't'? moc + fmoc ::= 'f'? moc stmoc ::= ('t' moc 's' moc)+ + sfmoc ::= ('f' moc 's' moc)+ + ftmoc ::= ('t' moc 'f' moc)+ moc ::= ordval (sep+ ordval)* [sep+ order] ordval ::= order sep* vals order ::= int '/' @@ -719,52 +1138,39 @@ \subsubsection{ASCII serialization} Note that we use Extended BNF supporting regular expression syntax with the following rules: i) postfix * means "repeated 0 or more times”; ii) postfix + means "repeated 1 or more -times”; iii) postfix ? means "0 or 1 times". The first three +times”; iii) postfix ? means "0 or 1 times". The first \replace{three}{six} rules depend on the MOC type, i.e. SMOC for space, TMOC for -time and STMOC for space-time. - -\section{STMOC encoding} -\label{sec:stmoc} -Coding STMOC consists in the following: for each element of a -temporal coverage we list the associated spatial coverage using the -natural packaging as defined in the previous section. - -\subsection{ASCII Serialization} -The ASCII serialization of a STMOC is a string following the ASCII MOC -serialization presented below, which interleaves time coverage as a -excerpt of TMOC and associated space coverage as a SMOC. Each TMOC -element {\bf must} be prefixed by the character 't', and each SMOC -element {\bf must} be prefixed by the character 's'. -The character is thus omitted until the next dimension element is -defined. +time, \add{FMOC for Frequency} \replace{and}{,} STMOC for space-time +\add{,SFMOC for space-frequency and FTMOC for frequency-time}. -\paragraph{Example of an ASCII STMOC:} -\begin{lstlisting}[] - t61/1 s29/0-2 t61/3 s28/0 t60/2 61/6 s29/2 5 -\end{lstlisting} -\begin{center} -{\small - \begin{tabular} { l l l l l l l l l } - & t61/1 & s29/0-2 & t61/3 & s28/0 & t60/2 & 61/6 & s29/2 & 5\\ - & $\downarrow$ & $\downarrow$ & $\downarrow$ & $\downarrow$ & $\downarrow$ & $\downarrow$ & $\downarrow$ & $\downarrow$ \\ -Dimension: & Time & Space & Time & Space & Time & (Time) & Space & (Space) \\ -Order: & 61 & 29 & 61 & 28 & 60 & 61 & 29 & (29)\\ -Cell: & 1 & 0 to 2 & 3 & 0 & 2 & 6 & 2 & 5 \\ -\end{tabular}} -STMOC ASCII encoding: two independent numbering. Values in parenthesis are implicit from the previous encoding substring. -\end{center} +\subsection{\replace{STMOC encoding}{MOC serialization for quantity pairs}} +\label{sec:2dmoce} + + +%\removeb{\add{In this section, we first describe the encoding of STMOC. +%The same principles apply to the other possible combinations of quantities.}} + +%\subsection{STMOC encoding} + +Coding \replaceb{STMOC}{a 2D-MOC} consists in the following: +for each \replaceb{element ofa temporal coverage}{sub-MOC of the first quantity} +we list the associated \replaceb{spatial coverage}{MOC of the second quantity} +using the natural packaging as defined in the previous section. -\subsection{Binary Serialization} +\subsubsection{Binary Serialization} The binary serialization of a STMOC is a FITS binary table following the RANGE packaging presented previously, which interleaves time range(s) and their corresponding space coverage ranges. Following the binary RANGE serialization method described below, each range (time or space) -is coded as two 64-bit signed integers ([min..max[). To distinguish time -and space indices, the time indices {\bf must} have the 64th bits forced +is coded as two 64-bit signed integers ([min..max[). To distinguish +\replaceb{time}{the first quantity (time)} +and \replaceb{space}{the second quantity (space)} indices, +the \replaceb{time}{first quantity} indices {\bf must} have the 64th bits forced to 1. It is not a sign inversion (two's complement) but a mask affecting -only that last bit without touching any other bits. The order of -dimensions is always time first. +only that \replaceb{last}{most significant} bit +without \replaceb{touching}{affecting} any other bits. +The order of dimensions is always \replaceb{time}{the first quantity (time)} first. \paragraph{Illustration of STMOC interleaving method} @@ -791,7 +1197,51 @@ \subsection{Binary Serialization} \item{The cell order MUST also be increasing first on the temporal axis, then on the spatial axis.} \end{itemize} -This is illustrated in Figure \ref{fig:stmocbin-encoding}. +This is illustrated in Figure~\ref{fig:stmocbin-encoding}. + +%\subsection{\add{SFMOC and FTMOC encoding}} +%\label{sec:sfmoc} + +%\add{The SFMOC and FTMOC serializations are the same as the STMOC serialization}, +%\removeb{using the prefix 'f' for the Frequency quantity.} +%\addb{except that: +%\begin{itemize} +% \item the first quantity of a SFMOC is the Frequency; +% \item the second quantity of a FTMOC is the Frequancy; +% \item the prefix 'f' is used for Frequency in the ASCII serialization. +%\end{itemize} +%} + +\subsubsection{ASCII Serialization} +The ASCII serialization of a \replaceb{ST}{2D-}MOC is a string +following the ASCII MOC serialization presented below, which interleaves +\replaceb{time coverage}{first quantity sub-MOCs} \removeb{as a +excerpt of TMOC} and associated +\replaceb{space coverage as a SMOC}{second quantity MOCs}. +\removeb{Each TMOC +element {\bf must} be prefixed by the character 't', and each SMOC +element {\bf must} be prefixed by the character 's'.} +\addb{Each sub-MOC and MOC {\bf must} be prefixed by the appropriate character: +'t' for time, 's' for space and 'f' for frequency}. +The character is thus omitted until the next dimension element is +defined. + +\paragraph{Example of an ASCII STMOC:} +\begin{lstlisting}[] + t61/1 s29/0-2 t61/3 s28/0 t60/2 61/6 s29/2 5 +\end{lstlisting} + +\begin{center} +{\small + \begin{tabular} { l l l l l l l l l } + & t61/1 & s29/0-2 & t61/3 & s28/0 & t60/2 & 61/6 & s29/2 & 5\\ + & $\downarrow$ & $\downarrow$ & $\downarrow$ & $\downarrow$ & $\downarrow$ & $\downarrow$ & $\downarrow$ & $\downarrow$ \\ +Dimension: & Time & Space & Time & Space & Time & (Time) & Space & (Space) \\ +Order: & 61 & 29 & 61 & 28 & 60 & 61 & 29 & (29)\\ +Cell: & 1 & 0 to 2 & 3 & 0 & 2 & 6 & 2 & 5 \\ +\end{tabular}} +STMOC ASCII encoding: two independent numbering. Values in parenthesis are implicit from the previous encoding substring. +\end{center} \section{FITS keywords} @@ -808,16 +1258,17 @@ \section{FITS keywords} \input{table3.tex} -%\clearpage +\clearpage \paragraph{Example of FITS headers for a MOC:} \par\noindent -\begin{lstlisting}[basicstyle=\footnotesize\ttfamily] +%\begin{lstlisting}[basicstyle=\footnotesize\ttfamily] +\begin{verbatim} SIMPLE = T BITPIX = 8 NAXIS = 0 EXTEND = T END - + XTENSION = 'BINTABLE' / HEALPix Multi Order Coverage map BITPIX = 8 NAXIS = 2 @@ -840,21 +1291,63 @@ \section{FITS keywords} DATE = '2013-06-15T11:50:43' / MOC creation date EXTNAME = 'Tycho MOC' / MOC name END -\end{lstlisting} +\end{verbatim} +%\end{lstlisting} \section{MOC usage constraints} +\subsection{\add{Reference frames}} + +\subsubsection{\add{SMOC and ICRS}} + +HEALPix supports three coordinate systems: galactic, equatorial, and ecliptic. +Supporting multiple coordinate systems in SMOCs would prevent from performing +efficient comparison operations. +Indeed, a HEALPix cell defined in one coordinate system cannot be represented +exactly in another coordinate system. +For this reason, the SMOC specification adopts a single reference frame +\add{for celestial coverages}: +equatorial coordinates in the ICRS system. +This choice reflects common astronomical practice, as the majority of +astronomical catalogs are provided in equatorial coordinates. + +However, it is possible to use SMOC for other coverages, +such as planetary coverages. The definition of the unique reference for each +body will have to be defined. + + +\subsubsection{\add{TMOC and TCB} \label{sec:tcb}} + +The time is a relative observation, and depends on the position of the +observer. There are many time scales for measuring time: +Terrestrial Time (TT), Barycentric Coordinate Time (TCB), Geocentric +Coordinate Time (TCG), Ephemeris Time (ET), Barycentric Dynamic Time +(TDB), International Atomic Time (AI), etc. We opt for the TCB +reference (see \cite{2015A&A...574A..36R} for details). Our +choice is motivated by the fact that this system is linear by +construction and has been adopted by numerous missions such as Gaia. + +\subsubsection{\addb{TMOC of unknown time scale and reference}} + +In the case that the time scale and the time reference position are unknown, +we recommend to set the time resolution of the generated TMOC to order +31, e.g. about 1000 seconds (see Table~\ref{table:tmocorders}) corresponding +to about twice the light travel time correction between the Earth and +the Solar Barycenter. Please refer to the VO note on TIMESYS for more +information about this limitation \citep{timesysnote}. + + \subsection{Canonical form} \label{sec:can} The speed of MOC operations - creation, union, intersection, etc is directly dependent on the speed of the equality test. It is therefore essential to always express a MOC in a canonical way, ie -one unique representation for one coverage. Thus in the case of a -hierarchical representation a MOC \textbf{must} be "well-formed", i.e. +one unique representation for one coverage \add{, for both the possible UNIQ (SMOC) and RANGE alternatives}. +Thus in the case of a hierarchical representation a MOC \textbf{must} be "well-formed", i.e. redundant cells are not allowed, the cells must be ascending sorted and the hierarchical encoding principle must be respected. Thus it is not allowed to encode sibling cells instead of their parent (4 -siblings for SMOC, 2 siblings for TMOC). In the case of range +siblings for SMOC, 2 siblings for TMOC \add{and FMOC}). In the case of range representation, the list of ranges must be expressed without overlapping and sorted ascending. @@ -863,11 +1356,11 @@ \subsection{Compromise of Volume VS. Resolution} In order to easily handle MOCs, it is recommended to adjust the maximum resolution, i.e. the deepest order, to obtain a representation of the desired data volume even if it means degrading the accuracy of the coverage -(see Appendix~\ref{app:perf}). In the case of STMOC, it is possible to adjust -the spatial order and/or on the temporal order independently. +(see Appendix~\ref{app:perf}). In the case of STMOC \add{(or SFMOC)}, it is possible to adjust +the spatial order and/or on the temporal \add{(or frequency)} order independently. \subsection{Working resolution} -The MOC has been designed to be able to efficiently handle observation +The MOC has been \replaceb{designed}{designated} to be able to efficiently handle observation coverages (images, catalogs, ...). During the construction of the MOC, we must then ensure that at the chosen nominal resolution, any cell of the MOC contains at least one observation (no empty cell). To keep @@ -876,7 +1369,7 @@ \subsection{Working resolution} resolutions, the operations must always be done at the worst (lowest) resolution of the original MOCs in order not to lose any observations, nor to create empty cells (see -Figure \ref{fig:operation}), and finally to guarantee the set logical +Figure~\ref{fig:operation}), and finally to guarantee the set logical properties (commutativity, associativity,...). \begin{figure}[!htbp] @@ -903,26 +1396,33 @@ \subsection{Working resolution} \section{Summary and conclusion} -We have reviewed the standards for encoding the different MOC flavors, -the space MOC (already described in version 1.1 of this document), -SMOC, the time MOC, TMOC and the space-time MOC, STMOC. +We have reviewed the standards for encoding the different MOC flavors\replaceb{,}{:} +\addb{in addition to} the space MOC \replaceb{(already described in version 1.1 of this document), +SMOC, the time MOC, TMOC and the space-time MOC, STMOC}{(SMOC), time MOC (TMOC) +and the Space-Time MOC (STMOC), which were already described in version 2.0 of this document, +we introduced the Frequency MOC (FMOC), the Space-Frequency MOC (SFMOC) and the +Frequency-Time MOC (FTMOC)}. The conventions for space and time MOC are the following: \begin{itemize} -\item The SMOC is simply defined as a list of HEALPix indices (order, index). +\item The SMOC is simply defined \addb{either} as a list of HEALPix indices (order, index) + \addb{or as a sorted list of non-overlapping indices ranges}. \item According to HEALPix, the sphere is divided in cells, hierarchically grouped 4 by 4 with 30 orders and the space coverage for the deepest order is approximatively 0.4mas. \item The space reference system is ICRS. -\item The TMOC is simply a list of time interval indices (order, index). +\item The TMOC is \removeb{simply} a list of \addb{sorted and non-overlapping ranges of} time \removeb{interval} indices \replaceb{(order, index)}{at the deepest order}. \item The time scale is divided in intervals hierarchically grouped 2 by 2 with 62 orders and the time coverage for the deepest order is 1 $\mu$s. \item The time values are defined using JD = 0 as the origin, in Barycentric system. +\item \addb{The FMOC is also a list of sorted and non-overlapping ranges of frequency indices at the deepest order} +\item \addb{The conversion from Frequency, in Hz, to deepest order frequency index is provided by Eq. \ref{eq:fmoc:i}} +\item \addb{Similarly to time, frequency index intervals are grouped 2 by 2 with a total of 52 orders.} \end{itemize} Once defined and encoded for a given astronomical collection, one -can easily combine the MOCs for these two dimensions to create a -merged STMOC which can be used to navigate and access the collection -through their coverage for both time and space simultaneously. +can easily combine the MOCs for \addb{two of} these \replaceb{two dimensions}{three quantities} +to create a merged \replaceb{STMOC}{2D-MOC} which can be used to navigate and access the collection +through their coverage for both \replaceb{time and space}{its quantities} simultaneously. The possibilities are then very interesting and will be a very valuable astronomical tool. @@ -932,6 +1432,17 @@ \section{Summary and conclusion} \appendix \section{Version History} +\subsection{\add{Changes between versions 2.0 and 2.1}} + +\add{The differences between version 2.1 of MOC and the preceding version 2.0 are: +\begin{itemize} + \item Introduction of the frequency dimension to define FMOCs, SFMOCs and FTMOCs; + \item Introduction of the SUNIQ and ZUNIQ numbering in appendix. +\end{itemize} +Adding a new quantity (frequency) as an extension to the existing standard, +without modifying existing MOCs, led us to choose a minor version for the document.} + + \subsection{Changes between versions 1.1 and 2.0} The differences between version 2.0 of MOC and the preceding version 1.1 are: @@ -961,6 +1472,10 @@ \subsection{Changes between versions 1.0 and 1.1} \input{appendix_perf.tex} \newpage \input{appendix_json.tex} +\input{appendix_fmoc1.tex} +\newpage +\input{appendix_custom.tex} +\input{appendix_packing.tex} \end{document} diff --git a/role_diagram.pdf b/role_diagram.pdf index 67fb4984cfbe40e613cc3c7beecf334aa018833a..a40021b92872d0bc01ef52c92d421f6f06e394be 100644 GIT binary patch delta 21828 zcmZ^q18}6@w)Q8sZCf4NwllG9JL!q-OzdQ0+qN^YIk6|^H~&-T+&lNHd%L>2_ImcS z*LwTyw`zCoU#-p)kk(U>T0?LkCozCn1(x_a=y4w9ygGtWJ-on+An3?52Ymu20 zh?)O2C1K_4;!4bxBqW5BAOtVRstVYNJdFGW`UR2;MDL&MF0RgIMt1*I=bse+)cKQ^ znd84yw!f%H3txS6zSan-f7W>fR855I0?;@}l2!QBWRptR%+aJS-0P$Arhmt^Do zneJtQ;$oRkC>$xGwF%CrTHD!WB#DJY%Lay7&}nh+3~QK(8iZ-d;mK^yj z7`&#pd#%A%c%~N&*sr?M6M<|*wma%CGRWeDAY(%b!Z1G=!8fAAK*~g1>J1$8Uxi4b z97ExDHY($#hT3#i+G3*Ta@(eO^LLsWTQnY`5f=i125~mIg}X66ph!yBRxZ{Olq- z0UM*5Xyj?2IlhS-BR7p6AxF<5npE>N(^Y*-^_O}I@OT3ApQe0zk4$P|AL2i}F`u?Q z^ouzP@)t-3<{+&-D>(}Qn5Er{Oc|3Jiu|@V9@<>s%&>&L^zs1ts)0e0(7%Fz(~f|A z@8rw!0QsscK%#P}I*BDjJ#ujWlUTm&iR9^Ixz4$iqyXA<4 zu1%dIj#Gl|&no$^whF7M34OooV;|-;t2Di`+F2JeY zQMNS%U0?Gzkr4c;Z&(W_DHhErTvm-&NgIwM;oy4e)R@Ea^hnJZSMiEM%h9<H`X=jdHl8c?ZA)On*MZ>Wbu*qL^}hnxBX0-=xg6&` zdSvZ9crTH_?2gD(n{?qKiM5Kf=zeQuBW@5N<`5+@w5{(el#nA&k2wX;xBA;Nd3-}-^^9dr zm-5*-U?c!5?G;HEOWR5*7|5z>hm&0CS+b3 z0E$??<5Ftil>{;QKQ?i0`NI{(E*pff1Mm)owkw#XTfG;TB|lThtpr=4-id=erOP-Z zIHU)~&~q2(kXv}X6h*Th@2TUJzLnG3Ei3DRME2R@sKF$MiPCO*lsm@VbQ{6rNeV#_ zR(^ADe{Lt&Y6A7N&$`JSURMIAM9!am0?s~7M8)BFT=_F%AkBsrGguA)+9Weef?;4Z zLHD!FC%9z&^%1Xj@%9zW;IEuyV!wrW3XX-f-0m*YPq z=G!N3RmXUEPzH!+n(41?xyRgCtr?78>Abmxl--WdweVT`Hg|a?j5ZCdj6jCw{yJm2 zVs!r0*(uIpvhRM_G(h519$exL^C*zF=VZjN5NRL)z zAdWK6e_wEciXI?(sMPhkkkpO6!P=jFYuo3*Bk0fdPj=MP`=A~ExnKK4;dcZ_lUXJ4 z*(Q<}{aP(dSC=`3x!3SikKQp6a_Ia_UX#w%Y}PkS&y@nXIxE>-6f>Ju7+NnTOPn*vEAx$#4y!~-cRJZnfo3G50ex|1Q6J2i>=`2XPa$u z8jj)b@ylM9Hk(lRNbk_e3eWLN;8m+Y09#1S+J+)W9@vyRNqr4d1=LR+I;+w(GXk?A!Zq0dj~wYb0HpN5oN%JQ_zBI>#2TLI|12JC#mDzG0f{^RC>`hzQ=Ob5ty=-kj+!isoYG4X zK-}j9^FiI0iQ20V(=4kfrQ6e}yvOrHAY#5qB0|)ds<0DQ8c9l${1WDbB4VAguWB@1 zR!WJkNhXUnKr*nU2&hPIp9_xn!w%J>0vx7Ha_{GpN`z>ni500HD9;F`RR_5xYIN#2 zMXZ$4*uB(yBICvPSZU)9a4~0@j6tG!L#kh}o2YCuN~G$t8`f4qcY&^IGBTur_$prl zI#T}b&-gDa77e|MRTwj z7_M}-4gJ=1$E!r^?$ycNL_^;*as1Ld`ccrf>reolcQh zl)2s^)7vtS48ZBD0c=)vRgA4)&T&ocRoo<434n1f@)DYFbK3`GTWZnL5I7G!6C_rK zonL50ij@$lc=gx}A`a^KWs$(6SzOV6r$XcOXC1ioaNk*9%|vm>+Hc3DGU<3sPSAiQ*8i#$@6O;~hD zy1IfX_i6TpH}rfIgU<31YR6mGx$HI|TQ(MEXa;g4Y!ENj@9d{*r-?Q!$i1PH+7?^o z(qb&M;(_Rih<45zu~V4HPkiuEAlPqA%{Cv3` zu|7IS(3?ke)v44e`5$UW*gMhZY}AIyGiK@g=%suna1<&6`7GfUDUv&7?3!Y~v<;po zyy7p?wp)FfN2cTjJa)2|d(z0X#cR`}lMtg{v4QU~6aJnSj%!1BR~fET6>J6)ucbY( zdPn-c%O`OH?}4A2Gr7zhcsjY#*&qshHSK_6&s{5qbR!p?i~0$(k@=aDQI0UYDRPeC zh)*ckjbfo1(BBxY4iMJp)T)M^pvQES5Fh2eC=iI$*Rl{fN`@VvTsxQ$FPblNlgASn z&_KE!6(J?SEcLYy#Fp9)1;JdhWQZzf$q6m~?n%r9e4XtvdpY4ah(lVn+FBSYlG-!_@3&|mAvr@qqk7$P9 z5D+gILRfaD=UMO;gfRe@b;D#P}fYYMFpkh`8+e0A%O(Xs`VvyLA7%)*(K2vtOgCiCQ-qt zqPn-ruSfBGO&`6@bMNiwj};gWSijaZ!z{^2Z(8Nu<($1xX87h7uL4w%N(l||eglba zWss<5S!2zKqX2x1IZ1j2b`tVtur$&&Poom2%c?!WXANe#) zmkDQ~tJYvzeuVYNU*sb4Yw_!pWwXQcYLB2@WtytRnr(f0a5aR3i-ww(-y-lw!|B(V zJf848gi8rL!f#+I~8wa-tl);TEN6E_^mi<0wEc z)EURw?0T$s?3xotAYqpQM?Ym(8poeZ5_!*qplKmx}eC*L6@Un3MIdTZ=B zjx}NnC;zhyKGU=pqtcL0YOYYs2#1d)t7;Ad{Trk}4M5>C)H9#aSoyBv*WI?^J`8*udyduVbjZI@ZMQw?)%9&5vEojc`vx-V@x z#_!`dP~<&3bYV1METc^53UR4$WaQmaXehCjiJkX7&csqzQ1KqH+DU2OcT+Fxd%k>w zQ*u(rR6Qg9SS}P}stRdT?L4sRpv>LZ4Z4k2YfT1qs){&pmqh6ogc#c(b17nfb=)h7 zw<6^?LEx7oPa=PK!2pJ*7KZmn*pDvA8j`Kw+p!;C37IfaOzv6)aWwI*q`TjflF&tf zSD&dWnk#tqB&k`%xw#>p32!NU@~*eTs|#s0ShIzgx(8^^S)I2tt)x9)bo(=i5`zxR z>R)qboeqq!DL+A0i*nEVFhkUn1^`&URs(8(neu0ao{@xijE&V|kvi}~Z};24l8>Pz z+oN%da0oMSlRn(8hCd-)n%$!LBediK&qdRj_Ls=$rMD<*L}?Zh^l!nv;+cNTa!n}YDSX|w2A{3O}~*#KoJB94oL8YH!yn;)}8z&lXBq) zvOkiKAZ!O^cZ{iP^6Vh1)Et%gnQ-$)+ZV!#IU~tj`&jmKJ@R?M*O0A`@zo~*eKH#p zdJ>@^YG59zKT({VsUikE;Xabz@NkUlXQ*%nVhZJ2HM@5;RJ}5DWCY(h7u+ON@ z#F2GF!N}goN5s4!XmuHtu9icW$oE{ z%*~uV5mT;50NrQcM$I)xlq_k5`-zj0r>(?Yx@#C;z8gytqZ>g0?CPOOIjuO5`D85k zjy~+hD9r353E1L@JR$0fMpO)*ZEnyu25XP!ZD3Uk!|f>)z3M^jIi<}Y#LmN zzXNQt>QO~JpD24v6Rdx=-V`d}paA`aVI9dYPC7;w^sD^y2AYQIMqB1QGgcq78ZX?9 zaRCelNOjo}Vm(Q;(8Q(9MHtBtuG%7sfPsI*dKAuTU==e3w1%0o&LGH4!z2&MI}hGy zcc_IKdZXbu{?<;hTrkWfNF*sj1-d}ZA;PZyv^3R*Nw1*B6{Ip!RiP#5|p_e!itZSRi+|OzFoEMv_>h$m( z=h4`2`fkFb_H0-P_W~lMRj2h z<>b-L(X7(Mr-@Kiu&MW{A@Tlc1Qz^s;sH5yDhxL=%Us(olh8MlC%{TRx4`1%*>; z&Y=!SRHydUa3uL4!C4YEj4AYt$iOu6O-ya2LW^K)>duV*3ROq6kjy#pJ}9nEBD?KU zu={V^b;S14Zj&|dVAD)xKlZTIAhUqf=IPwP)b~Qr`8Wq#54CLK05!tLwaIEgm0)XK z-nHxHiIGmrj}*Vxqn~Yj-BYSD6Qi!{vGZ+I4z+HdqOrLlC!78^>du9;vZr-ywQrMau;M;ZR&iV(I(Z(O(-G13@S0z@8wvh+dzeX9jqf zHjzvOvCcMJ1;M$a1*kTe#Q2tqP$sE?Fg{n{!ZyKR1K0fkuCV-9bjy;w&%t4|b%kpn zw+0F|ck?~=vOl^^Y%|e?K>Pi1nBlKnm;En#DR<` z*}vECi?7aDBRsZb&(@yKP=V2Il8;eVO0J*&ou2?sFZdgOKJE2V<`IPdt2tLZSL@Br zRuEs6t1Uh^H6q2MfIq;;7X6G0agC)gmgL$B(R26NQOKv*P+QH1l8T83T$thQAo&YU z;3Wz6nhBg?q~b%VPa}T>Kfe!9Oxte5#c-yMe-F7-+nhh~QsMlt`9h#JQ*=7R&m)xo zbY-zS3h$r-RPEeWCFTbRoZ@wCY}70uU;i>QBbI9v|wO?kZqeKdK7 zN%?p+bYT;$y$|CBj=mFXqd3~a+U{MhVk7!~UOM*Hk<{k<44fk1afEwp;vlkBLXBkS zZKm&r^DIFuIkouBpjdtq%C%B>!?l1O>Bs&+l4VK>B)oq7vJE!kvy+`8_~59=k9-x7 zf52>~j#S?8y7toBG*=Jo#U8buh&63WC-cpOx%e4if_7mR;9hw&RgI*JG6gr8?x$zy ze%Jr4V{(zQz%7KOyDAe3%3zA~dh^X%-_`-7ilT?QyeWF3FO`?hTmE;!tR2rAEQ)%<9&P6QLn{*n#gVoM)Xbqdn{SGQHa! z@xMS8_Wy<~9BiDdg#X1^IQ|#T!uf}@&^AO8>40GWLzVswXINR8|7#Uj(v1Nlq_*Z) z1!3VNW|Ackwj?b9ng14WlblJJLAa8x1z4dNxmZ~MtjtNfq?}3Dq_Cu%EG&%79Gu)7 z|FEdPg8v@=v0_njaBwhkbF(mW{X?<-d+ty0cleJL$j;8e$jr{d%JvU5`YYgI2K*_` z&GkqBO9c@Br2~jrI641)h5t(ZSIEW1{pUgf%c`xm)mQgcwZw=!`sHT#3Y{uosL=5hbruK)13e-km5f1$f1 z3VobER16^rn;Z@3iLT!KsHR>c?R}*{kv+RuMoteyUJQc_?VA1r0sMz3>M^1S4>B73L0hXibP7E9m35kLqGJvg0u4 zw$Jag&7$&4{>KfM8U9y!Xk?K_axu$(hSoBYr#=Q9WMfz$ipU~EYg0((L)bVGbe(Vb zezSPr5NDM&mmxWMTLSW8_Yh~b?%l6W4lI=w;qaS_;t+%w7*jti5=2a+{S4&z#)z$G z6Q)?!OnYlB6#98<$Bn@KVZ2lZOX7nw4*Z_5{o+hq4e%3=7giB2kP%(S=YGS_YMWwX zP9bPL8ER)>7P5Fi0Y;lTN?5!Y4Mt<)QYi*o5b@(p=e`6QItGrO#IpLcdC8m=2T#mM zdI%O_wgS=D#*bQUWlxgyXV3^VVx0m58VOPc)^yqfN*uN5fI4kg*bK|F_rXc6{1Mm! z44x|FdXZUJ5IWDQZ`W-%?yOEn$V(Tv^(CZatR2X}4$RC7!>kQumIWwYvV_;zPUqum zov^DoUkfd_2dnmb<5yWp#ZCoI{m##}-!>+1=;A#ENPzFA=vL^F>f`V>8T=8*9qMBT z2b5;B^k>&E^O|5)_b=*=Q}J-*Sks3h4vcpPBE(oCZ9CHt!LpS`I^9{VdMrWQ5@^DR z^6~V*1{@O%dKm#SpQk4y;yNA1W}m)YZ2>fzqPeRj8PXDfj-}ybKo^K<&$v@$N=0f)MWjp!NGXL0FnlF{V}>i*xyr8anG&r(iG!L>_RanU;c3j<8O-v^gezhpzq zDu2lbg~k-h`_c&!&jfa)Uk0|9sP5P>r|*S4+w6+tEV^12;Jr@<)ybdRn1cz7itFXNV?H_s(mMlYz@ zvV-zjQL*fO%D~GZ*r7Nkr`_Ufo{hp{a2y$Zqfr(ZO3}xQwXOQ-*_4}qi&x4b?$hA!^k=Q(?gmK_!GJU~<&e#)c5X zfnZu|Qk#(ELc~JA!NJ2>XdRpF+F|M&eNw#LbX6*0jSz_u4&sa96>+170dEm!N$Kxu zXZu9LA;1so4C4dJy}*pnbo@57Q;@8{eg|C~l2m=joPQb(30b;Jr?`)u7d6H+cOoN9 zBJ8jr6(){k>g5&2%F&Gy{yz6LDQZ7r4YB9B;CA6crM6qePS19XyY0X)t_613KaEJ& zsOVwoueViYNY$O|v*nCM0y%%_Hma58snhlYtYbuGGy~LL1H{;8D+1srwDBYNz6hmQl~f*p8*am=UJOfmfEecg20gwX6t% zRsd=xuTM}9`f|i!<_MS)CtG?&Y%iMPeEMH`p@?!=G3{GD=D+_u?0J@vIk$u3{EEsQ z{_bZh{5mh3KNp!|RSqzxprx5Va5p zz9nHIx_fryAW|47xbN&k$7vxMGlZ?B8>tS!S0N8ultqypC%B0o!+~RFyQu>Y#F=Mz z-|)ne=@%L?BOR)x5i)as_h0d!Cy0p`YkSyuc!3*<<#fOk1(u~BCHv4VX@q^B2F`%1g*{>Tf zW^39U4;2w%9+O^-lO0M8-7<)VUlAhTZD;Ej88P@q%r+kN1??o z_Sxf#vDZGS+G`o1uB9TJHW`;6Ix!sdomB3vt^|PgYBAm|!fvqPHTJdiQ!Ix$b>1P5 zPj!?uADBh1H7WCbJWH%(4Pi{bXP|!Vr#-iw5j|#5fGXs7l3&thH7df%kY+-J55z)0 z^1sVeVMftbO&LKohRTkuU5sb)vdJ3Iri9}9RPuqltak-yyo1~<;#>-e%7|* zBv}I1X^Sae&{D~NzKHE~6k3Fkf_pt_y^-c9?2#l$ADkqkS8w&u6TF=$rN@dZ@mQjt z(7;!fJwhb|RAJ0ylzy65`W+3wt%gulq`uTEF3ai*c-T+JMT!iM&4CR8tS#a7xH#s$ z#DUC}hSQP?VjpYUg1?bHoUigtE`RBO&YauQgVQNlkX0M-y7SlDDL!YU*4THs-Q3}~ zzH(mm7gR2`RCO2nE(P1_EH!es%&dHlGqq>+h4h(yT~-*M<8`K7qL*yjpc=B9q@ag^ zPO6kBl&)eaQ&Fi)DUzq9%MOk;=1Pa$TD?V>Lm!>c6Gh&0Q@h@;CZ!?IT`I%?OXj4wkr608wkXjPFa!qG-*d|4`EYyraI z3`PPdpEOZ-`yj^l`^3-m&wwY`Z=J=EO9ZNo`Ul;gIhWt}^}x2~74oN!XH|cPs8wjk zt*_0a0vag*@H$r(llg2rfTgW>}KNj-2u`%o*W@J_|J0&>lns(96GJ^$IjA zU)1mSA7ZQ?ZAkj3dY56$)!Z)i&ZA!B`$elvWLUzV7T7mCCJkgYIlffR=RlZN6d8d4 z=H>IAQ@_KoQJn-L()Wa~BxmhlVI792R7wyPgV{qKl_WtPQLoCWxGV}%P$42^#mS8v zAZgx$-#{xeygMCe<#YoFkA4g?_hzGSWodBhX&N!xwLjgXyucm){%~l%>nXSNi;njs z%zK^fwj&RkM6H&g;>i}!Dg%Bszoo2*jR!}EqACeiqoHs$6#<~(rz1}VZ^<2xMH^U+ znYDS9YLQ`S-5^suSK4^;{S+~x|2D*k?P}Dg+Cl_`x$M{<1C3lO*$}RH`K;Z`2^-iA zThq9-E(e3FU1)88rrZ4icWAc{&rnGGbVBElE8upAJ#NK(+K%SN9}S$uqAW|69-OR~ zp)`SmJw$qK?KEj)TIYA$?JNSOkE^- z;Xpd|=+z&IVRFUfiDABH3MJNkr7)f5QhHn3-xeb^m_)Lpbflmu%RW*PFVQm5urqLvWdbkYvfLXslsSVsYMUt-4$wNtC1}d z0|OAQ2@umJE4@!1J1emz!5iG9yXSH~!+l?%g)VrKyVSi<~e{DLNuDYn{nyw-vHxFP-_<;Q`g$+FAc|#C$?BmKa2S3r=z9ikeR+IBCJY!cg!^zE z2I>#w7CNFI-3zw87}#Mx+fj@KA&0ziOk9EU{TPfmA*xp2D-?C;s}NXHkbw6DlMixE z#0}CPGDg6*DF*`KMm}dU_T;0vJd8{m62ITR)`x!24cU;8x}&plJt0VczxqX5-R^fF zUZ3fz&#@?>f%5E7BD#ihi+7u~;gf9p22&q4`rzq+=9RYbTk*!8*d}2*srB;*rqHp> zEm5lrrx#=l08F>_Y-X>4DBk2V4c!-XWE3oo$v^Ze=n3eLF8WHazGnHNS6lt#M{NG6T-NP6^0nmuI- z(}u%d*26{b9UE=-A(lw$kVMu9=Cv*UFd?fAwNy@fj8_VVzW51;zoCe^KmX;5EEn&j z0w|L4c?2xBHOmMOd9E@QoogYjNO(k25}hezQ-EHcEc^XxZBuTc;bo!gad9_prAk+Q zlbrfJMVFzqM2<0stIe_iMh${7{!c{udU zk96~!&48;Hmv5ZqB#Pv%+1}{7pvlbv+>mZ}rS4=xsm3(^M#=5Z6k^(Xcs{I;lpUuh zRs%gmt8md|tx$)E1&lHk6!VuXl?ETBrQKR!*b(2ZP-{LfZpMrYq`=1aG;G9_!q$l8 zF!J`onVOcU90dP4WXr&akEc0)RyuK)Mf_YTGq2WS$l#rtOb>-x-k2b+Eg*-y5J$hP1AZ{ zF~{vM>YIB6V@m<;WrRdy&!E!Rj?J+!d5LQ38K!;b(Zz*fimcL636kCNe7ITrA_}7* zzS(#F2X2!Qg!&lD>TpRG)NJQ&KIG4?rU-FReAsqUDsyOd0cGle*jeOdO|P&1gw>YN zdHpHLW8>(*pYkksWg(mB>eRc7)~J|(5lm0%)Da0P&A0b#GE%x z4DY7e{TTkLHT6|v@tJZECqiE8Ag%J2ZuDBfCmJ3%uXI|)MPOXv6=m+*m+2q^V0RJr zb^)zyOdJRVqbS7y1I?kYAr;ulXed`Aoq;CiNYZdBin`RHnl`hLJPksEf>p@oiO2BT zu`X;$+tYBX2m?%$u_5dLx7YmUn`WxGa8?3$Ye0G;nI6kY*sz59_cMbcn!f%L$V>qefyhYY0U5Y|&DLHz2 zXjbFrkf@G&>KD63w{B5QaA!OAZhmy79|5Squawizh)%)KEXj#^s7knwz@I_|Y-mu{ zINSbG#N^~NEG$BVOvjL<0}EgPVz~A)>`CNxYMm{qURq~iPP9xE$XcA>GLzbP_#_>y z1TjiWj6m&K=m<4bvnx=TMktaQ6Nnm+dQOzY-mQ=d^YI;fXg4cxA9%80Y?0`J<~3QJ z<)!BbN>p9S9;ZUovi!DMpnR#V9NjP7LQ`swm&|^Uo&(##(DFV5%9_UvT-rx z2_||Jv?o-U%aol4Um#|x;lsF*;uqK|?yC4I#EjHQu zQP1-wY`uwda5q&O7V~igqb7*~oZUH}WDco}rn8v6lcYPAUA900}xDYzS=}Bj@%$p^aOKib{q{(TpKIa z?Nc~3?=Ob|CAkX1de;fg5#8oD=LkvG>}apK5ZjTaxmwS6PJ?uiE(J=Bk%Sx-j|iQR zsYS6`V~Dk@hMF z+e{mZA(vr>KBd{GmSH&$Nf!&z2={m?c6q?d2%hFGgtI06UzS$xycsvd6DgRyDc-OI zWxY$qtt;B!qyGHcm;>Yb8aH#FQ-SYtN{jFI#Ez9_jX_OV{H?kxAylkAEk|KnYFnD_ zrMy4{Xo5kZhM{l`t*%AJN7V^f9N?ynzh!^=(+Xgvm>UM`)vIAx#U9wbq?A`PblQvw z&4usX(yia4J#&2T^@8(a_Tv6-<)>ETOjfC=$V3JR5*mesJr0en&N$7=8n;(Rr=$4p z5esc+tkHCbS$-;JmMLX@cKFo(z+V25#=DDHWt5Wd&P>OJN081u|U5e&Gp*BJPqlh>T1fDHpi@Oe4aF z%uH$4{!Eb6o%t-Z5qXS{f;OCGcCzflC--x{+wHPA_09hVE=j(;(2V~& zmj5?3Vdv&x`wuo@`Mag&KS%g1e_#uJ!xpbT2oB3%tuU;tod30q^BCS4^4~(4fH)`%>;HmU5P|<6Hu;N0aQ;u~@&D0?e=(N->G&(v|3zm0hyG8B zq*6hYKXBx~$PD0rAdkQ3$N#0X{EdqK$@qU6|C#vAK^ly-Z`UgGz4O;#{>CdT|4YLW%YPb<;FHXRj1b`E@S%KZi+(Gpdw{+q%?l|41qlZ*at63zT+G=y za5z?%P!qKxYWumNr6voBjD0AYS&uflcXI3~tMDqgaw@y+wk9P&v;XWS?4avs4i23q z>Dg0u{SVBxa8n-B*7lx9=s_0^!L^TQp~=YoRXc0pIV;k!2LRQe*wZY;t}cOs5*|@M zF$*#ZfkfTChP6Ck>k4iyWGE6?fF=Pvop*ruZ&tfybko;w+;C)uOC57RunNH|mxWGk z0G_YEW_o_D6)=e@W=GlkB-wN%cR~zq`#FO=GF9bd?h+biU(KRCvAnFZAr+PgDmMjn1fMJ&l>M)a*&+$A7 z`5P%G!VE^A+rZkU2l#HC%p%g#j$*R1X%pkN@AVDad0m)%{(L?xYwA7sJd(S0Z?>BY zVlwRmMxbd_+Lh9@edeeFJ<(*^xBRT^RedbSz&eN|twDERrsv{TU|GWD;vsF=1IZGK zT6G8Wja-~%&>%9BlQl3%Xr^eQ6N-z-wMz483l>!d1$Mq(Bf;nCN2gIC7{*%ivV~vd z6}Sp*B(w*Gviv#chYTjaP~r<6K9LG5at*D1`EW=&5g7s%%xkC>eFFipNU9PS00~Pv z7v~2GJ{A6$I$~f}1@IrwC zxo(-E*M(baiZWUcCc?@7A3^hdhor0!Hp11$@#rxT#bu}kSMRLtCGT^`b;sL2kG=9r zAgqO01yG;RQ&a})AC9hM%gsl-$4ASQ{9b=Ht{*`GUMDKc#Z4`U$701$S1~cVR9>h+ zuEHftfMuh0=H}s!E6hSc9pF1k)R5tlbv#`_V#Q@-)f4Rnn=9Am_ooxqE{FNK3y%-4A-iJi3k$S@won3Te1I-(W1%`bg-Xm) z>GFxIVG$4AnWHs9yc9ua=+QGV5bG02+>^fSJ2pV_A)jNl7w{F_Wp-m-#GBCm@CQzV zh{p}`u5eIcIvgMRZG+RWBwr>Dk&mAj^&NH(TFE?=7qzN{`R@=uhP5%^6lW%IHb|Zs3Y3L$J_H! z{j0gJX*J$@E$WIzN$P}rOCGU6A0)FC>FNX(SM90%D^A98 ziLa9z7!!#R#T}!BN^-D7$Qzb>*!`97Jl=hk>%LDgXImrrZf^}W=d{t|vA~?k4$dfR zMxrbh<6nif=^T1n$QKbG*@N$mb>U$k6wblP)@&!P-e`-e9ildsk}V=Y3&6@L8D#m_ zx2E3Sp|(BLGET}|^($Cfp$=!g>x@}xg8&+Dz860>-wC5y%>MUU+x40PMQ1xIM1SwH zN3`myM{VkMpY%YmsOBP+`|nb>V^u$!lE)g0f0Vqs>eqbF4**pm#sX;2db%YSm89Pb zVEvMRn;nOC$X^hX3!)_hQc?-~g~DE)t`qd|NC}vd4X#EUbnGF*1Vy5EsDKtkM6@}T zYsmJ$s-XkG#3hih37Ns{Kr16IX?#_DO)fM@f^hVJk5421I!@bB6IlF7la{P)n-6yc z1Kn}<*}ShDZFNmvr~Hd#1*(il{=V}KGm@0>{=b*$@^&tUc}vWJRX@MhbiKW_5Y63* zPY^-rjQmvRw2%78TO(&dw|hrkKx!)rXOn^$h?X7;nyLgyNpaMZMap-VLszh-Z?un- z1Lz78lMmbr@%EbG$6=2_rNFZHplBg!^N<6<)g(3Q(EWAln5s`foj|ECr4z6{j942Y z)eyq0&1SzjUp;;WmM)2zFoL1aUAQ%66`OC@FZu{0;aFU>$o2Tt`E19lGF|u&_!riV zU$Dn;Uk`86(fP*gTTqQ(e{+a&~V@av~4jAj{ zoq_gu94tqA%DpAJ2JZCiEWf-hzu@e(Jf&wcJ?%CU30w!Ahq2%<1{7*W5otnr_RYL4 zA;G-?rjQ2S;!h;^!MOH=iTtm^Hm$z~5y8GCy+zL3OP>Y%PfBOYJb-tM)tToJu{{$?s-wyglFk7s2b{%Y4Go6za&O_8|uRJ5u5>< zuO>l?33ajrpJ?vka(eXcPXT$*DOTCfPHXEGAVjR=hUnJLF0F1B_V}*2yUb z|5(rHg_s0Kz9!0x)YQp$am(dReDSlid*XL%^Tda6&qsWBS!>s5(6gngqH`v*46L*l zB_RYQ|5W(WgvmoDPQ&VGqMB@Zymop!P@subJxNGamASSePHAZT!A7RB$?biK!2sq? ziuYv!h=D(Rw-FReoZwFa{*GtG8uu^hAnHsP z8(&~x(^+j`LGMs6Y-A&@cB2(|H7%P&)7I0^c^= z|MbeHo$a%IGeK8Z0v&gY3zLnUzZwE?ct`rYj)3YSnTWkKy3SjT8Wv*D(wg82SP=UL4;~wNnt7E;R06VQ$*|J} zsW==EeH@$wwRIGHl=FJS#(VoyS)EX#fKQ4ViQ6<(?OcY#v}9dV#%^`&9Jr%PLx6u` zC>Lf7mWB{NT8syvmaz+O)m2EbJR?KZ8DpWZXf=G6!d--m;xAtqNxqob(wHT|^VGzF zJ)Mv#=450Mc{78^CO4l(n){8TcZmi3Jw_x^z18J&gnm$Ts0uJWzq!;W(!-CVZc=?^CqFL`IS!n-^1 z_9I*~DjQDaZ#0?Mr*M&ednT#YP?`TCW|qB$Lm?cL^E2uxpvS+4i#)G=PbK5Kv$;|? z;WK~v#s784b20D*d6CB$60XdGVBR3X6nI&l>V zX(&#AO-hBZVK`J%*wG8K>_Q~|H8(0Jg3yl$6X-<$;x-cI(1kcXB5=qd87J`xSJ)v= z)Z5hi*c%!LY+`D^0PgtYo(GiG5Pr(tubPXeI&YIGj9Csu7Tw8Z9J;PG!X!`Nn4%#w zWp3J9hNnf?eCb%>&2|t9Hm0`5)~Jv|e$g3c)q7wvS}!y1V0wA!SSn5@t7(a;@HvPm z+B!}*743!L{@_E2f+=R%N>jX_P3LO#*p7X;zQ5hk`w z)n4M>K2GoD|5wR*hqJlDc#;Q#uvA0@LtDM#-HELH;wALZD6(K~-DoV`S zt7_G%QMLEh8pW|kjp7@6PV0={AHOSCuItYIilfADsXVvk#aUqnz+)QX{=k7QXrOZ=F{ zCSL%xzQ#ccA#jFC4?)NfpVpBnUk+Bg#wEG%>Fcu2swEwJcX(tL;mW0Vj4G@Y7DxbEHS|RD1i|{m!8th+yTsGLvU7+| zuUGMVHw6tFVWtJvZ;k}NoeG({x?ZK|1DT3?X31LV zw;xE*{Gh*3aIz&?57wtd3~LMQ0F(%!Wq z!&?O{)cvwzlJe3Vp~K(4dXCEVFhTIX#mOhMGR=13c~DPgcsj-^!;eln2a-9|86<*4 zhbw64mpo}drI zPghz$Qz8@thRANr-wRZ#8^A6hy(rC+>rSZauoP_C+x;_l%#LAt=BPv|N+Q4hX zbfz@)SUYBtO$m$F-Q+sC4SL2pH zBx_^fNo<0Yg+}MxZcotG&*kUpxU0pH>z`lEx{_PPXRhQYkP5f6t>)tGQ+he3@o{YI zMie4XT1=N0?R+&E{sG+;wbfMYG?Mwo;Eq(GW;2FQZoZiJb+>oq^pmEo1#c3Yuv;+w zm{5i6cC~D>0n>u0ak=!To}OAXzANM_$O*4Js6stUxhp$csYonqCEVg#?AKM#Cz$&b z`QfcGfns0?m5fK>JNa^5b<&S&_q5{rf&qx{Jd`!8;}4eK$F(Dda1U6jTu(0?%0-#p ztx&Kuxn-Y6(lUx4KQpP!Rwuv$C0Ff=JZ7@W>+Ktzf`V)06ejQW`;S=&Ke9vXh0_=h za8`86F1}MV4YQ6^pQ@}etNO%EbI7`+hgSp~gXy9w?bxU~-+AD-BO%y{*TQO3%^`>_ z`*F~iEp>>Zp^|>4A+JO5$%GIG{Z#;}o7q?ma5KOnn~X>M;me0}MvdP^(~9uug)7u7 zGXA)lXb~i8`Miq6V9xnb!&3AfY~&6%Ixv1izqrn@Mpxe5^OimtfdyX?_wT`%4m4b8 ziB^yBv}IW=+D&R{u)I~(Qsc1uyRf&2#P!KfbPe_eyo#tsN7cKF8hJ!<6ZZ zchIrDCAoXawS$wH%)VC!*n@Ho*wnY!c%pj95$Y5oiU~Z>?c-E8%!JMEDpeODU~Ccf z_RO_{o!wjtu70Lpft7d zAIMw6pOX*G@+2UAV6!7%;R=sJIK?l7{YFRGB3u?KG;?Rkx}W3rL@1`lE*+oyCXzlX zDCeMFqZrwi&6aWRcST=OSYDh54S9qnUmZCYiswl ze`EC)6?2}gKrVj&ZihXHTW#pdMMz2Fi+bK7Wbpu!Z5wR$tfYu38vzSNfQ`roa~k^l z6=|a2gc`kY4=x|L`M6Qa^Ihgy#qFm%vb!9zEN`nE@0F3P1`hJq%!VVan<=%)8Zl+M z<@rg#$z)l_go9?}2TtN}y0-&C=$ zi3ZH64nlzGIRSgbt14@$Ra*Qr2b7!&?3YsF`WwEwi4n4G&q+B-#W3*}u;X2Ou}-OH zCSyYY%|ApQBn3~^#L@Srl|S@iUy98#CNACE$(mYuuR7p}NWZ$fi@!+F*mN6Y=o^X= zd*$GVd;3ajdUb6weHe!pW!735bxksL?ls4|bS)srJvtBsiv@C%Td{p|T{C?xn#tno z*gl_z=|z*`1zHvMp;Y1g02*E+?YFx(77isB;!DaF4J9FRqzFoPOO~xyGE2K29s@dV?0U6sdl8MvJpZN{zIpwavDzSh#l{W0j~Ud> z>YwPI7+1$!ZBnV=ld<+rohA5CU1yHy^3HKUC0bb4jB|a2Jf0}=zS38zJlV)QMV?eO zYZQi9s7?xVUua4ibA2`129&5cl}uie@G;fq#no#eRq3TGO?_Mg|Yz?Z@3J4B92WNw}w{J0hQQ z3?@B5C=cZmdj#f14ubO@nN0Ge5Ovzjjk#W;N;i&wBbVx>CsFd*Cna-bS6Y)N>W{q! z9a2xa-d6#IDcGhy-_mzbH>8!cWww}v04MCqBg=WyWr_|il;JW7+mP+W(tCc+$(4*w z4nbpugL-S>TF4g5CnN<%(!QoA#A!T2Ao>igp&t5(@ZkhU09NoV|DDE zFU>Bo4shhqCae@NX#;+n+OvU@u!`4Sgm5@}_pweAmCWB{*`H)NY=qHM_V z5y@+B%6uyaqa5;^r&H;O+$86zoNL@G_|7&_xO zH6TK3{ghIJLN-0*5m$E%9l;Xo(t1olU(q6K-+ek+QvMf*8Z80Qb6)MDy%8<4+9K=f z4)yBMaBf>Z(aV{fq$UR4GVJc+`wm02=qV#wzH^F{h!7O=lfM0jAq5dBVSe!T?{<{v zuN6P+C@q6Kayqxr8m8CLawgR{L#oDJmGI}J`oHWbdXirR3#T=4AkOi$bHj5W?%eQi zg3>7sJLfU|1DlEdQj>5d)i26-{y$Dq=ajH0lwS}m z1UvKXgrTDRf}&6%F`N7Qd<+wnYxKt%b$5Rp@x?q62g**Tuarh?FOC;fk= zL_&X*`7}3D0|Yx45{bjsKT;wQcOyU9Hi7~EYu+I>(Kl)Sz}N)=)F|K}e{C?_LH&w6 zUAT2o7+dW8O4X+EeSeBW8)b<`$a~IDpqAi^Et0Cvt$X6!Bq|Tq@_f^2%A6=Kuywt2 zS|_{?=v<#nt62nI;T^U2RJ2hafLoX{2b2@)bqoqNe4h7O7KY>Q)kAg>8Q4%{UhKyi zhFpIUHVL5{8YIAvW=zg+J5{BmFZJvcs3y*Xs(qVH8zUFOm?!L-`#8)afGA|3{^qN`KazY0n`j2Lv`qfz9Pm+JVs=+Cc^XKlz} z*>>o!IsG^@!&cMi@(vb@5WnGy7q#%dt3AF`bqcWfYGFg^y ze5lrk1XTO=AQsZk(n^5(4!!T!Iq#9du3Yd*VA%S1ot#}vom||Ua3z4?^ivoKJG-K$ H63PDnMe_kK delta 21694 zcmZ6y1C%6Pv$j2L+qP}nwr$(2=Co~2yQgj2Hm7adJ*|J{dEe)p^Zm7|a_@a-M8@7( zwN_@vRnd9|)OiY2YX}NpCuAg4h9>;y#b8dyM8`7+XU#|55+8yu5@As$Py}gba#C7G{KGgiL>9{24f#*}D=l{p-sBkTWy2G7@(1 zB-Ht%F%q&fG7_?~F%s(iV=m%g@A}8yg^=}s93-3_+#CscdHXu>rI7QZE~aiyU;Y{Wue$))c@mK&z+{ z_@P}3_5)#b)5F{yrFZT5_gmNVO2@A@7MJM^7K>>vm!rv#&~HS8;=iF37GGt!y#eqB zeqT<)8|zTJ<}(I!rSw)kM?k{9l7m%xt1e54S=_nFP`5Q?gYScCQ8fmbfV>McQGVg@ z^Q>wxyd3tgE)SFfbz)m#4B!j@Pxp&wBY^KpV)uFV$K)N>6NiaFU8|2j1kv@9;TO(@ z2mNVlwFD*71SFJRpkI^4ub|vO2tY__Er5imU%WrzBP6_d4?B5THb}c90>vw~MO#!~ zP;u#V@1q`uTrRQD`Jo@#LlVC2Fuv13)A6iZ=H+8AGpSQ4LOWTmB$iInL2ugX?L%8DK(E4d_-w zY$j7hRSG@CG>$==J0JJh9r@~c7-E=zbo2KD9S`Bk>Sf@4Z$AUy#|{I*<_F`pqj!|k z2It^GnT$UXXsCChM5W}MWhfIxED&3NnZOASao^gnNKjM? z2b*Z0$^EJ9Uz2ax(-0rNlmIS*#{2viUqj=>a?4v?ato)o&bCXJWAst_pu2WKBjHj| zh&aUQ*aP)w5Na8lt-CdvC;&Fn;Lh$e-uBM0MSh-4>CTow*UYF&%8z2F2Jx)Kfnj7DP zUP+IXQ7^#5ajX*SHyaeImuz9ztK%{=5C@-`VCQEXPMbr04$XGX8Gvw~J0>Ad>Y%cB zM)ooBZ9>U@knq(3nZ*lu`nCv>$JraroTi$BK4TVT7WJ@zhJos9Q)bLA)FCRPSI;2| zr$*!)Z^s-)|LtmTA22Lci}!VLz8jUs(9MBS*)`n!x^h#y>0AXNyaNWMYVzYY)@%aA z26dm$-*IP{;5}gNihY9sd~I5y#F`5oHG}hAx}<{0e#|uzesA+EY{{Q!%# z;8L`xyqsFwF%?A2Y|+mW*RxQi&9ZDSMsafjM--oQi~PXG4zPqG%w`sqdY^Keg3yhilrD_|hq6Le(!{SqNz%cXtmYehzHb zw$pLYb5Iij29SSw{dpB_vo~>QYa`4R9@bwC*cN03Ep5PNv4PnRpGi_*3UQm z_MOQV7YNO>7I^KnAYi##`_QGv&x4cGnf;FB z9LKH@fr#7)I>I%T>Xh2KVY@ZJF_<%sxwW5+D+F`vp9tY_*hl9zn+Sn%*tg5;Nse%Z zbBcZpgfR({2qAH|BZzjtW6=d7dVt`mQrGKJLbv%g#mVw>!zn8Ues>#?=%}anL_IveU;9Gl zcML<7RVDt_CY&DqR;^FdoHd2M*YNTlxp^q;(&n4EB8$8JUOR^{@}Bobv~+blptKQa z1z0NFswkgU_Fiiaca1qs5XD?N6rp*l2S85tbUdM0GqD4y%k#N3G z>3fDtl`3I3!Iam}2SwA3TVyz$*FQ}sZI%%vEiICI2ZC}}uAZe63I-igP8qC9WtU`-lqfDD zB_1ULLmJJh${ql1$MBF>Lp`dKjE4ugycaha(8>?wCxP@1i8D?cr$3zh4w!&|Z1Wpe zbqs4A_-;00oe^+Pz7 z@qHMl)YIs$OylI8&UJh59l#&Wq{61Qt4IH=S+Smp=Z*wb7%k$Q8HD*E$G;~xl&%_Y zRMUW)2tEvx$7>61;ZnaKcAsKNzuxd5*h?wG!32niCz_Pvf9?>edHSpm-#>Z!@O-T= zUh$vx3*P_Tnf49Ado(0l^o_ptBYCYY73SL4ex_k~csu&4@Xd+946rAkTOlTtsH=`K z1LICJHrWW>xtFAidu$!I-$xzEd_^}YuIFiMr;;ZMq?RA4EnPU7fhRUS)uS0dQY`3g z;l)SZfHuD_W67G!j6b;9^vv3dzxJ&nx+8jFxXST9<|CW=J)`w#wp;Gn9O3-E&riG{ zj4?wedLrbPXKJV<5a4u{{H?*RIj{HFSQGY6EEFh=qm;keGbi@_ck96ovWv;PPrtBH zXs-oKF)#(RejtiuUTL^>x2rvMG55$f3)4~|ph92x3EWb_AuJurxq2uxBHjjGV1=qa zV(OKzJX_7Y zFV-Rp)`A zNfcfzyQ-fAU2Hh>u9XOb!5FDw_V>agYC1}J8JYR1RnGyx^4yh}%B_PV6(0I1V5M1f zn@2R#AxFYFjY(Yo5V>3Y4YDr&e6GAvqu!~PY-NI15EDMQ&;E;P>`;vNUR0w8LYgIpkkuJH>Jvxo{>z3XE zVU4RIfJ#Q(WaDs>ntbT-_LgYRCm;80B2X+=ITeu7KuYEo=2DDs)F3R$twEWk2zE&O zl;JYpW<+gKdrynS(l+HDOJ<;wkGs5CL%@IYYltpwsu4R3~a0^v)?QB{R2_ zv}r)O4qxyYVI+Gij9tJfhQ)GXhNq}AI-a3~fGgIx>1aY91qj8!w&{!L_ZCx6t?v|OFQKr{qmEIq6K0x}B2lnS(f*yV`dFY? z?ee2Hr3pZB6N1fSjvP3FEV@)8!LUor$@KuFGOXCK8wvIuGA#G8TD|Wv%p1JI+oU2f zuqc>aNpl1i$XzVLUoKeeV*xaped_TA5)s7Ao~Ddeo6NHB=cDlk6E`8S2+Ur}53NC% zB<^G1LYySJ+hNx+x0AK8#nZhp3)3*uT$jkE2QY+rX_ zS&n##yYj%c#dxvZ>8Q;1NlLZ`?>Ae^i89Q&+<6*^+sPrVSo>_`bML#aZ(=B9Vg77# zDcj8Qs`}?^oqGi%99$<1A`hbPXGc{mj953fz@dmC~fQ#u8W^oKCaNP77_ zR=;@WCET+pCAveeH{gxyL_8?3od0OpqXmCq~lyW@n%wMQA zgiU}HVr;_nSs-tJbbwLz6RuuT;8!pmbOmvk5&d`rh|$EDHz(YR2&6|+k`aUGmOEm1 z@CvGAA_?UtxpRnZ5Vo`E#bE>Dt;5E|fyFzvy#U}>7_-4}m4s$ur;L6W@=vm^1E@Pf zli8^IDuFDL_UD-~^SfXc#apI*EvG3*sDvbgB#N;HS&)jcZ&K0l+~9z~7*9kCqbpwH zg=2~MBkoV@E2&tVZU`Qago;|%UZNuX8HKlHXLrm^ag6W-{N$au*>I;TuRBbAiRL6P zQwqQ{yA>IL3;|jn7|U8PI#?6xw|kCZeLNse0h_%t=Nj79k5wQAMEw|e zxj>lNa-!E9y!oNiafa|f4s3bn&YA+jc`#9`bSP!n-SrGE#;%DiP?1t!l~bi9%my=3 z)`He;!F6*|vpnkO$$`6Dt>bkso`}4&S%KggtcSbqT**Bir+YvR9%idt8qCAHV@xh! zfWlw*6}XF0SET|&vAQ8s{5ZqSTpGYT;hb&cwgcLC7}E2;rz{T**BE5}e37s}gla4re!NO!@k73B>0vq3c23Zs^S0 z^`N;xySd3(DtGK`Y^?zye4~K|F9#cPpp^L?5JLVFT@~J%6Scs))lZT{!x$iO?O-eO zsOD^Dh%pxlv&Ws5m)=E!=|D5|5N9YJM$vD%t~F%`(jL#!z@i(4-P0v<-Gkh9Zf%`j ztI`6%gA9ext>-S|oVHX;wl+T|R>L6qNs894{DGex_kho(}%E(mY zb@96m%IS6_y#TYK(wrL@l?1@jYM}|+))LPuY6(6Sbw4~76x&t_j-^B&_iB!B_e|+r zxy~Act)-6W)TrH}T^{*CF9S{yPcK5w6n_4IRL)A#nvSl@79hi#!B%qyaq@dXeh(9(&9`q-ZZAb;pZB>>Ft!DO5NATiU9lf$< zp@3+`LQL;&?T$QhIf=JpMzk^>xf+~Z^8&pCmM(h4$S`FEC897(Ha@1Qyd*NK^F*3j!w9BKQnHm9V2;YtD_Vd7PvMgHo zLM5u(n~Uu*AT}OE-FzK+_ttWiq2@4?3l*V@@KW3utk|Kk*+_W7GE9oORLClRbhgJ|1l1;RSVfeYs(!YeNFSe^j)f&-1 zbyBbcg1Z&h>@XYX-J9KvXfNe9=>P&U%}`dnhp7sl4WI@XXKf&!7l?|Jjw%d)IW+!p1wtvx&Jezp882G06 z8k~^1X=1$VjJqvf)O}c~RhX{H*pP`Gc2BX3BW*#7=b;8XUv}@pD=GT7T1SN*8 zFm;0c8Zr45o0bx$EXFSvZo~2cfNZ7~U`;@G2HTS4>?}q#iMmO(yQ`r$7)D0(zs-y4YuYw^WxF09% z2AUr6$M1Kko0fYfS)J+Qrr1OfVK1DKe}+MfW3&3kGHx)r#UnI}z0=Fs$p|}bbixm| zMs*@kcRSdO@~$y59y~PCjK9AYCwx2iRJbY8g8y;z^W|7!apwnr2J-uU!p4q-LL54~p%o#(fi5={X#ZqI$G1Y0mDbmpA$sAiembgah9WpUpoc;geB1lJ zbLV)f~#2q7g%kgQe=KJVsoQ; z6T;nPRIA~7x^@J&W6i@U*w-vRGDobucF|-PYHQ9337fpuE8$(!(c*TxJV(a^IyNWo zvhmIMU@lkUjo>hj*ZV^0N`q%Sm3JO2MlduU94pr+x50-)(${ev35tgkgwxJZxm9Zv z?0JlrVEEzfxB;|=hZUP{qUVU`agKo3_e9Vil4Zi-LgRRKsQJ70!FcRCP!QpU&#@C? z>-5zHVPy&A79Ud@e?k2*kGrF$3PvA*LjxBJ1h7$G+_sLrt0HfdfryoBOaou9%jM{x z_Y|Y$;Iw#h>-$v)6MUttQ4zk3CjGi>kB{q%W-#M%U*T2Ph1Gb}*`L15OOIn<54gxI zx#rm18su31fnnR5sP@BPX6PJa$wg#mKGK-0FQcx~P>&WhJ22{#{JV6({TVQzs$1Ei z0yb9NqO1_@{zIK`YzJ|)16aK>TN`OvNFt?hvu=%u6L%=qTcX7zcwaq5&irw&KelGOPHmtF8Tk6}QoKVKO> zj|_vI{Q`=~e#3*jt!3j}9Ne+Nst`e0?DNy(6CH!3!0fsNjc|p&?ZcBYs^Tmq6v*?_ z)8k{pV*FN+zOvwU_&47qeerPC5R2*jYJP{)I+N|Cbv53szVfhzLS~F#kfB|GKcS zF#V4QYf?8MJ-C*rn6!dK(jDP4(qA^C{C_|cd(s3y_rC^M#4AHblGIndQ0-y&M{+VEz%@Y^Tde0V^S29;@8jt+) zE57Fqml?h{FyIhGIw<5!M;HfNNMFZ$^^o)+K!~CnjDyu~$yfdzgfK0^UaOsx;aH3f zHas?zlr5=f6JQLk0t3L+GZl*Jj%@7PVWlKeXueMwhzJo?3nd3Bxhm_>R|he2D#y`W z37u@s#%bT@v=2vv&W?EJh7-Fz;It3}R{`vb^MOT#J5Q)p|mH7kuf|?1(@$LuwvnvsDn&_BQ7SxYMy~Y=O2Fn*Ky^4;^^M^E#^MA9_lQ&GF5#@AUj~IbM-?r3&{IBQT{N!<_$fcb{yvjOqhS?4?+K^9)gGgWZ-R@9Y`5+N%k~tI7*)m8JkTFGB zVW@Oxl|e!g`z>PWHiL^XcAP+kJGsiq7QObbE4Z-sm$yF3%SFDaY-Z0#_2Dz5sMB~_2R z_hD;46V|CpX<<1PGsPX+0P_kG07Z(p1=_wrsZ5%flwaSh=yFV&EI%oHBrP5kot#Rl zaAlqM4{H(e@x|7u!SlYDwF-mh0lh@54c>;w=Q&v0E$KsF3iH)wFZKWI)V!8}NK0m6 zN^mYw5)U35!m1)U1gYq~8$HV6AhWZB6!3ZK4`tXXJbuLUr~zm~ZS#Gh4!_^kApw&R zV(l3X_R83vw2eITCS_4z0U{$KbqDAM&2Rfx+G7__6o~b;3W>og>Jw!kB z_tHHyrsQ8m$3!(LSQ)Gd+>~^VG=wm6Syn(ddgH}_n5L4mVwY2SpRB?o`=!lU`8`p( z$PP^VPg6|Zp`15{(G@5)rgm5V^S z>_cNC3*ER2nWpzyuqE3^By9P-W-}o0ybvdY1@T|*dy^pAm>suLxC!aCz&Nnvz_^t7 zBqOm(!hd#{7L1_*W&R-jN+7q$oucNM5v{Pz>4X4Z z9Xj>kL*Q(^Z>Cd?0jB3|u#c%z%)-vl`S*O>F1EUJu!ij`<_c+tgyN%fYeLos^x~q!GAptH7;TL!;ehx;MW1RdfT)1Z*;p^zdtiF1c&Ce3* zn8Ag5lhI}mJs#YwdZ&*XBUsJHd3EZqP+bXKakr}>SElWNh_NW&ljFjm!x*hP3R>2Pz0`01T9Hv(4rQX51}y5=4-DqyHK@ zyx+mP(^JQuahdu2cthKm&JlQEN@D0lj}d!VIXNFI5~5_aV%1Kplxr?cw;lXWZp%V? z(lof113h#m1T zy7bwTU`!a?2mE7TGpD*IW1$&svl%dbx}#%YegOLHsZjm4!d!u5ViR^!ELnW_!&SC) z5$JA1l}rn)g$wZ^pl~7-^|#3x< zQ(=CIo8c-YimF%Lzbt~U?cL5bPxl!j!O6^VI%neD1?Ly>CkyupjwN~=Wms}Kjn3!x z7Jc44X1LtEzA_kep8t8=X6JvYh*=E3vDjtkgA%9a_;bN2x^AHm@O@TwV7=N7eFQ8U zFkMk3c^gGaRmBDuwHBdvND*~J()E=S{r)PhKwL4)1s0cS98ZMq7xl+wQ)b7nL^ZmL z;|tB;Wt4eBD}UXqmgyG$pty@}%k~F3%n(I~F(d6WRxejQ0_FZ?FZu9-o7$A>K4Ar8 zQilnf&722I?UUjs4W@*h?C-T2{0)%uU7fl~o3^}>Jq&Lao1W~PSYx_#SL?TO9qYzw z;UQ7_`qgdA8G`nkyZ;)X`T7(=8U9?I-VQ<8>iUs=6Y*>H-g4NSG05atq`lotqJa5F z*ps7vU1;>spQl6xWfjjX<^0-WLpK)sEyLy%jls9QwBd!{o;w{R7NNkPZLfe|@5Z_E zJ4Cq=$EQW4)mtsZXuf5N8IfX&)Rw4aQm|EJ&#x(rGEl|Riu2}`e#c@Tzd|W0(tg(~ ztjOr|JLoURMG6m(%>@rJT3h1g2=H%ui7{0gPKPOc|Ket3JZD_#xW2h9NYVM2xv-vu zpi#6Ks5ann7pV7h_|=|PW8VcZ8;hoGJ?FUY--~W)sqX5seJYaG8GdMMnc}?}XK2q3 z6bnStSdkx_=cy-OsT6J7pct~7B%{5BQK=L!l&WGbOH8awEs~?A$q9%xK9T#LYl@h( zTlpSKDq_G43&fK|6z>~KH(ttthN@>DT}L>D0JA;MCkCcMA>dT{G8qV9`dn*5xZaMl z-)^_v&E1AOQ0&mC#$@2}nsc8s)P0xuh5B~mFKCXOtj49#4&=B$#C zIkf&*gSU%&s#H%X1Va;}`F5_VkpP1=7|BTeVu`%l2Nr?SCw8un<9W*Y-dO_9j;qnA zf6)DvdzJJj2eGvvpFagSnN@KKHLcK&TVGSzV{9B^g4MY;ne0Q|A(eQCsOuMI^3~=} zaUjH;lBEXOU4UF}sOBx;Y~p(#FMA%S;fJF!`b|aYd&IGOJ(CObBdnEWJ0QsWqwwal zi&*!KX)=iC2aLcM(@#R4cm0?38;{+6M)l@n@&h}zyHwF57_deF(Ok*0r7ie_gdZ{K zY(xZRFcTt+sGTB=bzHm1X4#e6bJ@}@EC21dw?St{#|8w$BmMIL0RY%OU{AqtgAJnE`^PKMyZTGmjo^#gXzSHZ0{zby&~n3= z){2`aH`5b)CxJ77vIse3S)L;0QgY(*Fp$bMh@zpIO5C$tXhc-fOJ_1Ccw16`)&g}Y z3=M6*97YenG@#Fy|7L1APy?+h4heYz#8qIeC8j9hSHD`BXz2n4_2anfmgy1_a@gL^ zL2^88aY(EyWKl2)9`WgH8a&@=9SvO@Ee8UJ%>gk9dI2fmhg(dvnD%HKcxL{VyDkO|vY$WU z)qX@OmsyenMgxC<#73DL!ZvbzoYZs6ZXXeSz^|Q$M7&P>mNJ!pk|74z^n^TQK)`H! zzh^mO${q`UEDX4mtqHz_k`RRYG^^AM;A%FxJ%^}G3|xox2vvO8$$7Q1WOv@lBT;k1r+*z)nPA2o2hNUxymQl0Xb`SAZ}b2*2wuA!Mn zLT(bwlmvuEQoy?5`AZ;z2)mlnMF_I7_EKwtASl)-5jbb)(sN|2pUOsw7WE(`$Un#m zOmAZWvN-mA`)1z-90GiNA>(>YYS94PCIU6PGfbS{?|=W${itD&dT&k~BnlVy&pPzL z%~jqTd_Y*CzMac&*-;~~p&$Y8>?;xCl8O<(w^&}mDdLGT6;~Rqu$Ddfdbrd?_rd2R3M2Xqh2&1UDp`~|#-0>g$~$#Wy1P#x>l3f! zVX58BHaC6;JGsY7>Y2&&>KrAmu7Y=LfQOcq)a{b+Tw__N3OR;sCMVHX`9`*rNrObp zBLN@%eG3$vNl#_OhcCjiWD;Q}ON}mDwJXZGHW^dXp+@VQB8*_7+tF1fM^;48C z2*qC5*ulhv8R9mfv%mc0puzxStpO^c%Pm^EFsce>KPZTXBy=&Ir#!FHxXY#@K;v-e zvqX7oB*Nn0v&s3#STY0@rz5vh=w!+uod$WkgB5hjYl6f{I?KEluLIHSM-bzhQgz^Y@HMiGvqT{t~V=oNa5)}CdzDz4wU1#PZxLi$u% zYJ?W@(zuq$o2vdAvZ&biz+>TUWL|i{QN;ui98);+zztnOJFyyZKW&YQwL8Z+H>&e} zHqQY9o2x_+jOTQ3So_|-NFXyR$E|0AsK>!;jQgBljRNaSmSH`EyX$i$KzdG>ZG}b| z%f7k{i0DOUkcZ+|&M*<~j6fDSB|0j+^y+EyU~QcaA&es-ctz(NlLY$eGcpmy1ckD{ zRf&b;ur|#UD1^oL)Hkn1JF_MZQf4f)7+)-DUz1==-GJpLS!GxUL^nVS&ujeHDB8h` zsV;n5xux!SM5HV#ay520;M=9RHZt=l$mVw84;J?g!UfAcXirj-Ija4M_quJJOij|> z{Je>Vg71p^Xram%kZ?>fDQ<^HJCWc5)V5Y-m9k7>2;u16o{}BNI%YBzU3ZDA%yY`b z$Y1@d)J&zaPc+rE2MjUwgmZfpona9)^DAt*4c9ISw}TqgMC;PCfN|yt;jQ7Jf&m8y z(H@;iuSA=sANn z$DQxQTO%A!1v|IW7D}(e7O6lJHX|7l*JXfcbinuy#-(=cAG_WTfr_ zra=uK7&F^w3Xe61q$lPC>7SKU_*pUeRjpBB&E@n!Yvt#iVReuPYu24URC&_xL*3P` z+4uNIkVvMKWI|+ltUE=4qE_hei_G00wStn0xn`yTy!xwfqTkCPj0Dl$Rg}_++LcN= z69Uu@`t8ZN0VJ@-ERZS^MnYKRE%d|@va{5nO?gOrWuW6*V91sX!nq^R`XHtTyrcnL zFd9>deVd3E^N>%J;)qNklwo#FY0cGD_g7IAt+M|6;zPL0)Mt~;Q0Vfxye!`UR(wtTnXD6 zEaZw&RRnx<7XD+X{1KB&onW+yu2CxN%rm=1+{EQ@&!u!CJ=vSn2>#n^q{P|pGaR0H zr@c`~)e9ad#;w7K`>>~g&%4t^L-aDDmHnzvK);S0{1fCO=};bC-;%tlHxdvfVG|?W z6CF5^jRS;ki~)pyDI$if_zUU#9xsw0*`~tbGrM63-Ym;EAMgO!0QV!X9s1suT%}l- zE2Y>k=QyWqHw&(_x0NTA-=7&Dj~|yGKEKhNegN3}Z%+v};Xeq3UJBQbUW#TeIe&Sy z18REW(0}WqZ1~wP-#%u9@58+xn7Q_ zoUS39hKqc~?&C@OLpjvCKaZvD4&@4u$mGh7)Wk(Qi^2z&^3390lys2r2^Xy=AG_mH zbF<#hMvB-V*zq(T$4Bx(26pB640Sd70~lO!zukU2K?u<9sj)Qp;^(Gx4^O6i3hyG$ zX(+WsSZb5m5PQ5h+Y~#zpk0%)^MUtvf4Issw|CfrdM#^{oY=s5{noCu2iOD#4CFJ+ubi)(VRi9K2&nW3Z&wRc#hs{$L7uNH z&_GU-A~psBx91mDlyK8j)a#)2NlP7O&X!17_m-2Tw@J3W89aOSV^e;oepG5-TP{vPQ6?q&Q3P5kxvi!uJ# zCS3{QQZW6e?&p8&U6}uR{r?ur{I}jE5)LPw6aGb2@;Imb3e+`eR7e4^O)!9#JHQ3y zV$Q~Xx^MsiIaVpExr+-*dNhYn*M+>E;bL##IK!T_9HWM-puWxSU``S^odr3D9eDZJ z!m+h5Ieo-#^oH3SdcEE!NUC)}$xB1-y67-xlze`kZ3|QHLVQj3g!nmMr-!B8|Z>EjWwaK6_B1iSR33KMC` z63>^grU+PaZRplLm?;zJgd)?s?q+Q%<7+$!+>9e(3v~25K9R5m!xANz3~s{i zOORC7rQe&aPXrZ=L0aj&nX|}m^FV)#6o)Dy)+GUPm~q2%{035Ug%C&LdN+YjjD7h%x3;h8AfZP zONhBtR&jt)avRX)?4vcxyxqVblfK;%^vC%HKN5Hg^6hqIDh})W?!bkZ{-rL0?oDsR zw0yRV+Pbk&CK)n>6iG_rjHz4ugZgqt`l$6oB?Uk)$(*@(NkZ1@sg7KQSju%-0@-d* zi9m_mpm@*E+N&v1HKhXs!F+eV@6^yGAtR`XP=iS_a#(ah6>9GJ6KQkd)AB*{!J*4% zhx}(i#vIICFd*b8sm2b#*`8pte((5jZt`hy!05zvO5K9V=NsqQRSDuMn zCT_EM?#j)moRi|l*#s|H1g}4E?;GC}1pvfvPug@J>>_@V&#>P1(gbl?UfB`xA#mK! zM{DJGc!uBP_03I!;X}Xv;@l_6or;d@>gGlDj^2(~I1T1arXp#070yM!Jm@*fnBqmf zH+hBZ3AgaV_utti`IO1ch`RjRvC6Oi*)AQ#!qu1nVR=XdeuF7T#A49Tq8+wbEdn4~ z(TZ98xpY1WIx9|t3QOruTn+PK4c-|0==SK={hEF?$cjf7y4K&0z*EPzjjiZrsqCb; z@skqp>WL!C4e8J-3GzhBO69(#1qsA{-9;<+yY{E=UWp&q8ud>7qojH zFL%?L=M$PqjjrnTa*FveY6Sdqo)JJ-81n_`iX;_x)v_d&L({KBN#y!oYcW3)VW*~N z91OKW8uP12voX;HJyQfq@=ydwdsZh{qh)`*F2iLz{`U~K>-|~o&tDquXyRrg0`fHy2=Z8UcS@_WS@aI!PoiJqd;V*hg98D`T*A_9nXcSj5ocwa`E1LDI{5$=Oxs5! zkR_T=ZNr^CU8jgeoTQm@4=}Vmy*7GJxs#%rUQ|B(4}R?KLq<)QU4PByON}{-wq_Iv zfv&|jsFh_OI@HZB*}gzAo%!(J(xk5k>ZjV{rrHbh3x3=+E90|*KveNCm^5ge+|u(4 zlioj~?#TTu%|coft%*y7Qjh>iNqPJtp&pNyu?Bdg1ocRJ7ovB1PGKQJVv%}OKy!nG z8=PttWV?RU(CLB1g%GfD>4EG(%c9Sy{8Zg_?iKNU(bYXZ?e-nj4Qs5Ai7Pb*gR{Xf>@ zEPodq#sycIoR;IXhyfJskg+10{UOi7bd^Q1i-L@Xiw^qC)iFs)aeg5UksGLmEM>^r z>zN^^qbW&AJ9o^%Iq86#fw=+|1Iz_f)N7Xt)J(0--(^Nk(_n zW@w3(e-~+DwiIl8exm`X+7vLQ0Y#oXwQI|(G&uY^>&gp@Zg|ou+3s5Hau_X3bMGmI(1(Xr5h-5-&fFz55x&ezc7Cb3&fUZb+gQg`JiT@oX}O=VGE zw>4tc%v9qwa~*ofgcuodEh+&%G=gN6OtbZF=(A;O! zQ4(D0Tx^rt}#=ka)St3zEd_@Jf*&)BY6-ZL!L64=J}h==&x(~s)< zaV*eqAq7HAsE^g}O6wGp)oB>_R{i0mBjCPpA!Z}T#&ZPVZR~6F{jv6@d!{$JS3E#N zj(>UKM{WOho$dYn{$@UM0jRuhn%=?_jam5|z8b_NiZC^4Ov;kNvjVrS&!=u#`)Q{b zJ9Q=sQ~j+;ltxv&5k1j&Cvx`;9HxD)fCmqr5&d0VD#w9M^2 zIvo^Mv1=)&H7%f8IZUK-?XOq>Wk+ zuKKZB9;0Q8tV6!_XS=-IPh!>U%wywl8AmC4?P!?`w?0PP^$Tmax}0i~>IHi3J?&#- zTKW^GJ|qlU4?7R00Y}aOo*c|d9NLqCLIdMJOnuH&<5 zU4R_m#NSw~fmM-SOC@Y`sliL>E`~`O#flSCw#UBi)mWl$I72Who29<`cA6-Htt8<2 zMPHliqm{-RHU@@#nV%D;t)J`eh07Q3Zl$_?=%u@E_LXMjc4-F3A>jmlA03euzh5JZ0Pja7!!wYsO!T7s5*>^_(R+88B&T%iu=2Al$M2i zG3hD|2|kP_EfjjcW;*zHW6mi@+=#HK9@i_*r24z4M9k|H8B`xq!))}In;a<}46%;5 z&W`-!*PI%hcGea-oo2VYJ%%s|gU|Oj1+5 z)M1%mj>TI{zv~kzADJ)x5&whx!~H|wFHYYi4|}@OEJx6qSP&doLhNzsc`9zGrYc&0HRL|=H0*S`;R+kI3v+_L^Ci9c0M@?KcEUoWn z)8;|}z|j?N(ptyyOQcwB1>qhrqszC?F+#A3nZE97zu0OFFj{DDyY@%FKcslhv2YMI z*owIo<(6C4dMxp*Ov^Nn3G?b4D_cQnaE?zWdkuv|FfQq?HBdr_e*qJ7U;Ugw!vEd$ zT+W2g@^P12Va#hT2irpbK*vxZznFkK1|CufJ95&*og*}%2(JoYMi&Q+Y`TasQML;@ zRc4gNi7R|w{cH`Q;t71#ec6TZ)~>_2|2=I5dB=*RkWIp+yJ((=xopiexUi=7Bje-H zLky&r7~L}=3EaMRZ)HwzC*-^>uApK@Y(_M$8zHKP6aAxof1pJl;@E`1C96=P@GE9+ zvjAa7d&kFcV5DaaW78dQ)1~m#yQmQdDEV#MP$1KGjaX{hXe6fWO)lrcZMOzGeh6C+ z6`nR_&(113Imi~EcabyAK`6|G$_87ZMhf{scbGx{mR5VITDOVe_n~LAFqyEjKD@+t zC#vl5Fv+mzN7Vk==rh+j2=RIxtDTS7_~$PX4}Z!~-^`QqNs*HQ%y#fyzbqMXF2>x` zjD;{sn$~U2`TB8x|7iVXty_#4q-i@pBf6OR6zxBPJwTfLJ{KTtZ%mu7luIQ1@jdTt zA0u;$$`Em92dB&Cn6HtFd*~sr+O6&f-dab+U0zn&Psh`f3Z3C3|8?n^JY1(inhYfk z=l`#eD*=b9ecu(z5|NN4#%`>$&k%({#y&!XvJ9FmV;NiI*g_H`ODTzvvXosU6-Blv zLM16{F?=Z$#s3WI`^xXS{xjD*@0t6&=XuWiyw7>BxvqQeM@We&)lACm2T$!gV5ulR z`Bas(gT@cjaci~yfY$p}0OxIs(@9)}+HiKW+~?>Wil z%W2tdsx3NCM2p7QzT_6cnh%B6AN|(iarski=r~sbb@E`(E0t3!9#f~Ux*x8#3M|kG zzHbUqL~d=TI%&n7Rg^t##&q6$mm&~#zJoY^Slo86Ov0Y5FRowbp|hXzcnfp3E?;b3 zrVkuf?_Qnd9TvUTc*jA(qv^QYkH9@Z%kZW6M-L2}SN-2A5Mm47zdY=6)ARn){PlbZ zDbB8mM+lY*!IIA^(!%4U4-enqBqtRl{mCUHBuNqN$=)V@*pizQ0STWt(&!e^0860y;| zJnVb|Z;hQlWm7d7dVC80Sv=r~^f!)r0!Ct{=EEY( z4=-YNM@`<=a!FMo#X3X{ks9?#{QOLkJ(0AEC+EI zD!#P7nA$OK-a0v@XTQHZNCzOkM>oZOkL7A@7fijH26YY6Xyqz;;$eUi`l?nE;|yhz zGNt9@*xj&+L=VWNpF5u;9;_Wamsu_zJmyrY#V^Dn$bqrldB%b^AjiQOnR#A!#BzAM zO=EHefH99hpSoQyQg^Se9@I82@`p)=9JQ&p5Z2e9dX`7)z@5I9B7Xp5>nY{bH1$MK z-ip0iyH=#$-6SNK=>x019{!fKT02ZCi7J?2jdn7A#&hK!#0_GG#_|iTqC&%5PZo6&kYp@%6UK=P1nTFMsml$ap?%cw+Yz zAqsnYCC@~4&)mX7ED0zwea6>5I{3ux!spS7QT4K7y9H>GmBp^!jK`wtXJ1#*+$4vFqL;}C)maSl8_wUbyHN4IBVqj!$! zdzlM1OV=$Qvl~l~Dv~VEcjQV)c$_3!e49qM>}f=#@a-8#NiYMe20F8>HT`O^8*}gc zd8IgZ?|@Fv%ag2MDkTdJ&}@qu%c?b9?=JGR&YpW}6!pR{??H=A(bvj_#hpeJ?&KOi z^!tE}PN6qa>5@;5pxa82W_4b~y^=>sJ%L_#2D~cRpRB~GS!otGV>V>gSt6xVu@mwVKK?T}{KLCN9gk~0IDe6-J-e`iOjeJvl@7@XP z;C|S_XD2)~V>*=%XF!*qxV{ok6db-CY4v^Za;cj10pozwwme7I(DZ4QBeQ8{Lqlf# zcr6>lVs8m%sos9%h?N$iwgeT8ru*k`+)#jMZeV>E2!29L-I zt#Izg@i^B#VJ@EOToo)JUBLHfAJTv|>dMz7J=CaOPNd_rrTf01MJ3CCDU{tml@NYjDu3ugyvh# z7oA*Ggl~n%6P#m4b508PNH?|>DzjPj8F7l0XU|sqng@w78IhYV=^v3F5`|ZZF(YrBNXq$KEsF6y7oI>I5_N$o24-WYg%#c`}3S3U5H4bUN~mpuv;)tKrZ_+hs$C5Mw8RgMwF z$BYB-jny~aVwFz|Xcxd+yu0e7F>a6{Qi4sY(^rQdoEeD5yd3$Ky4=(zHa27&)8$VN z;@;;Pr%ITNN=P(;gG>0mu zU!7c?yDRYE*!On|Y4H*kDTGOdle9wq53+Y9@5jz)*G5kr%H~ev4tFx=$Q>EmOY6S4 z_;B9(RB3%Wvx8LQA&0X+^Y}XM&R{dHi~qboYn;+^`xZ1#yebjiY|2*SK@FT>V_Q<6*Juyf>(})-YbncLBP0YjBH;Eymhn|p5L{;a7e4b#(zpLhb$aE9W|(koEkH+ zNheQb&4*lirZFe#`rT`CP@iwJ@CVOcU$(EB#kEa6W%bVXhmP;4A&s7VNDC1Ay|>2 z?ur~rl-_IFWxi=U**ngO#rw-c)PsVYu!I893DawrXKSIjK_q9_em+AWq@g4o zFH`otGC-6Csg%R}QN1qH_lhLsh}f+I#FB0gV4fD4t$k1FrT)FF!UlQ0!a7$Sv%mqW?TG{5@SbwGVqbDBQK(db4RlFwOBBp zQ+JMvl#SMZ7amOEHem^gG;-(u286nm@O4zBhx#dP-&Ljga0Gtw(o|=AXYs3PtSPDs zc?Wq7v6xXM%{m3|31*hrJ+`_7HyJZ;pK>y^A>PYVOzryXHZr0(wc=#z9SgK&UvVmd z-b;S39A1^o=iWzyllgqZjIF)&&sjV*woXnpP_R-Y+p7!ml5M)n9db+w0Cy4h@pD}T zRs!F8>78*I9gNjFs^4XWB`Nz+Qg87hSvD8H6{!^Vwn2HoH6=MJlTfPbz7UHJZrZ8E9F7_ zmh$Wb%i(QEc@S$-o{DMGR<>>6@vJ95KjMF|iimYqQP+!Wt000=es6<~M4-{Xqrc8F zn&Qp$EzMH0DI*}u_{Gs+`)>e}GVcxBhF~y_u#_V{1bNWR1{bKxSkm=32FC~>28n~9 zQH(i*vqNFP1yBGGpySQ6{F^5HCF=UKv$q%gT$4C?N-!=XG85IEkBh^AO zeE#AY&3q{&e`gn8h@3OvNcN>b;L31}G7KVjf=c!FQ&UlKW}t4qj@~E8&VI@iUpFXp z%@rQZSSrOAqOPZ_rw4oP+hMtxOSgpikFTf)deD_ zqlSPX;BYwjfkCUnF$yraEDQ!+iw>|}QW3d1z<5WhqX)%pZQhLfihdgv760XKgM3|F z*}?opu!Fb0@q%E`Xe1iq3R%YxI4l}WP{s@5wSmFla8*#uwvGY6K^Ws1TL*t(Fc?Fq z_9q4hFaI0EVHw%|$2d4z6$>WX9~c6KL;YZ!`HVDJP z7(%^2$Kg;6PWx{RwP|vyaK;S+e+-195y-!tfEea)gK#uv)9gU(->_~(5RO4HHjF=f zAW$&$zYzpB)(Hi}{v8?w3ccxOKn(qR?KYes&`3DAdHm@Fi`f({@P7cCqJ;o(Fx21S z2Oka0-{D8#aGS!4P=&+(b1l`^k?i5(yH*zTF!_`VBQe*CZBi&y$XbH0)gVrLN^NXb_EMmzLHw)norZ^Gk)Q diff --git a/table3.tex b/table3.tex index b9956c2..8f1a5b6 100644 --- a/table3.tex +++ b/table3.tex @@ -6,22 +6,23 @@ \hline {\bf Keyword} & {\bf Definition} & {\bf MOC1.1} & {\bf MOC2.0} \\ \hline - MOCVERS & The version of the MOC encoding standard. If it is following this document (TMOC, STMOC and STMOC), it {\bf must} be '2.0'. If not defined it is assumed to be 1.1. & NA & mandatory \\ - MOCDIM & Physical(s) dimension(s). Either 'SPACE' for SMOC, 'TIME' for TMOC or 'TIME.SPACE' for STMOC. If omitted, 'SPACE' is assumed for backward compatibility. & NA & mandatory \\ + MOCVERS & The version of the MOC encoding standard. If it is following \add{the version 2.0 of} this document (TMOC, STMOC and STMOC), it {\bf must} be '2.0'. \add{If it is following this version of this document (FMOC, SFMOC), it {\bf must} be '2.1'.} If not defined it is assumed to be 1.1. & NA & mandatory \\ + MOCDIM & Physical(s) dimension(s). Either 'SPACE' for SMOC, 'TIME' for TMOC \add{, 'FREQUENCY' for FMOC} \replace{or}{,} 'TIME.SPACE' for STMOC \add{, 'FREQUENCY.SPACE' for SFMOC} \addb{or 'TIME.FREQUENCY' for FTMOC}. If omitted, 'SPACE' is assumed for backward compatibility. & NA & mandatory \\ ORDERING & The packaging method used. It is either NUNIQ (V1.1 or V2.0 SMOC) or RANGE (V2.0). & mandatory & mandatory \\ COORDSYS & The coordinate system in use. The value {\bf must} be 'C' for SMOC (ICRS). & mandatory & mandatory \\ - TIMESYS & The time system in use. The value {\bf must} be 'TCB' for TMOC. & NA & mandatory \\ + TIMESYS & The time system in use. The value {\bf must} be 'TCB' for TMOC. & NA & [mandatory] \\ MOCID & Original data identifier. For MOCs that are coverages of VO resources, in particulary those used in VODataService 1.2 coverage elements \citep{2021ivoa.spec.1102D}, MOCID can contain the IVOA id of the VO resource described. & optional & optional \\ MOCTOOL & The name of the MOC software generator. It is also recommended to add the software version number to its name. & optional & optional \\ MOCTYPE & Provenance data type. Either 'IMAGE', or 'CATALOG'. In the first case for areas computed from existing images and/or footprint or even STC strings, in the second case for areas computed from a collection of point sources using a unique and/or derived area. & optional & optional \\ - MOCORD\_S & Best resolution of the space dimension, expressed as the order. & NA & mandatory\\ - MOCORD\_T & Best resolution of the time dimension, expressed as the order. & NA & mandatory \\ + MOCORD\_S & Best resolution of the space dimension, expressed as the order. & NA & [mandatory]\\ + MOCORD\_T & Best resolution of the time dimension, expressed as the order. & NA & [mandatory] \\ + \add{MOCORD\_F} & \add{Best resolution of the frequency dimension, expressed as the order.} & NA & [mandatory] \\ MOCORDER & Best resolution of the space dimension, expressed as the order. & mandatory & NA \\ PIXTYPE & 'HEALPIX' & mandatory & NA \\ \hline \end{tabular} } \end{center} -\caption{FITS Keywords for MOC.} +\caption[FITS Keywords for MOC]{FITS Keywords for MOC. \addb{'[]' means mandatory if the MOC contains the keyword related quantity.}} \label{table:fits_stmoc} \end{table} diff --git a/tablef.tex b/tablef.tex new file mode 100644 index 0000000..d25e19b --- /dev/null +++ b/tablef.tex @@ -0,0 +1,41 @@ +\begin{table}[!htbp] +\smallskip +\begin{center} + {\scriptsize + \begin{tabular} {|l | l| l| l| l| l| l|} + \hline + Order & Significant digits (n) & Relative size $\frac{\Delta f}{f}$ & Order & Significant digits (n) & Relative size $\frac{\Delta f}{f}$ \\ + \hline +0 & 0.3 & 1.00e28 & 26 & 8.1 & 9.61e-7 \\ +1 & 0.6 & 1.00e14 & 27 & 8.4 & 4.80e-7 \\ +2 & 0.9 & 1.00e7 & 28 & 8.7 & 2.40e-7 \\ +3 & 1.2 & 3.16e3 & \textbf{29} & \textbf{9.0} & 1.20e-7 \\ +4 & 1.5 & 5.52e1 & 30 & 9.3 & 6.00e-8 \\ +5 & 1.8 & 6.50e0 & 31 & 9.6 & 3.00e-8 \\ +6 & 2.1 & 1.74e0 & 32 & 9.9 & 1.50e-8 \\ +7 & 2.4 & 6.55e-1 & 33 & 10.2 & 7.51e-9 \\ +8 & 2.7 & 2.86e-1 & 34 & 10.5 & 3.75e-9 \\ +\textbf{9} & \textbf{3.0} & 1.34e-1 & 35 & 10.8 & 1.88e-9 \\ +10 & 3.3 & 6.50e-2 & 36 & 11.1 & 9.38e-10 \\ +11 & 3.6 & 3.20e-2 & 37 & 11.4 & 4.69e-10 \\ +12 & 3.9 & 1.59e-2 & 38 & 11.7 & 2.35e-10 \\ +13 & 4.2 & 7.90e-3 & \textbf{39} & \textbf{12.0} & 1.17e-10 \\ +14 & 4.5 & 3.94e-3 & 40 & 12.3 & 5.86e-11 \\ +15 & 4.8 & 1.97e-3 & 41 & 12.6 & 2.93e-11 \\ +16 & 5.1 & 9.84e-4 & 42 & 12.9 & 1.47e-11 \\ +17 & 5.4 & 4.92e-4 & 43 & 13.2 & 7.33e-12 \\ +18 & 5.7 & 2.46e-4 & 44 & 13.5 & 3.66e-12 \\ +\textbf{19} & \textbf{6.0} & 1.23e-4 & 45 & 13.8 & 1.83e-12 \\ +20 & 6.3 & 6.15e-5 & 46 & 14.1 & 9.16e-13 \\ +21 & 6.6 & 3.07e-5 & 47 & 14.4 & 4.58e-13 \\ +22 & 6.9 & 1.54e-5 & 48 & 14.8 & 2.29e-13 \\ +23 & 7.2 & 7.69e-6 & 49 & 15.1 & 1.15e-13 \\ +24 & 7.5 & 3.84e-6 & 50 & 15.4 & 5.73e-14 \\ +25 & 7.8 & 1.92e-6 & 51 & 15.7 & 2.86e-14 \\ + \hline + \end{tabular} + } +\caption[FMOC order and cell resolutions in number fo significant digits]{FMOC cell resolutions in number of significant digits for each orders.}\label{table:fmocorders} +\end{center} +\end{table} + From de5c2753ad82a5e706819bcdf788bb39eb3eb643 Mon Sep 17 00:00:00 2001 From: fxpineau Date: Mon, 1 Jun 2026 13:43:28 +0200 Subject: [PATCH 02/10] Clean text removing replace/add/remove and replaceb/addb/removeb macros, removing commented parts.x --- appendix_custom.tex | 14 +- appendix_fmoc1.tex | 23 +- appendix_json.tex | 4 +- appendix_packing.tex | 33 +- moc.tex | 827 +++++++++++++++---------------------------- table3.tex | 10 +- 6 files changed, 318 insertions(+), 593 deletions(-) diff --git a/appendix_custom.tex b/appendix_custom.tex index bd7850f..71f699d 100644 --- a/appendix_custom.tex +++ b/appendix_custom.tex @@ -1,17 +1,17 @@ -\section{\add{User define quantity MOC conventions}\label{sec:udmoc1}} +\section{User define quantity MOC conventions\label{sec:udmoc1}} -\add{The conventions adopted for FMOCs are generic and could be applied to any quantity +The conventions adopted for FMOCs are generic and could be applied to any quantity $q$ numerically represented as a 64-bit floating-point value. A user wishing to define a custom QMOC (where $Q$ stands for Quantity) would only need to specify: \begin{itemize} \item the validity range of the quantity: $[q_{min}, q_{max}[$; \item the mapping function: $g_q: [q_{min}, q_{max}[ \rightarrow [0,1[$. -\end{itemize}} +\end{itemize} -\add{However, one of the advantages of MOCs is the use of standardized quantities expressed +However, one of the advantages of MOCs is the use of standardized quantities expressed in well-defined units and reference systems. In order to perform efficient and meaningful set operations between two MOCs, both MOCs must -be defined using the same quantity, expressed in the same units, and using -the same validity interval and mapping function.} +be defined using the same quantity, expressed in the same units, and using +the same validity interval and mapping function. -\add{Poorly constrained or insufficiently documented user-defined quantities may therefore compromise the interoperability of MOCs.} +Poorly constrained or insufficiently documented user-defined quantities may therefore compromise the interoperability of MOCs. diff --git a/appendix_fmoc1.tex b/appendix_fmoc1.tex index 394e7b0..bfdf365 100644 --- a/appendix_fmoc1.tex +++ b/appendix_fmoc1.tex @@ -1,13 +1,13 @@ -\section{\add{Previous frequency indexing convention}\label{sec:apx:fmoc1}} +\section{Previous frequency indexing convention\label{sec:apx:fmoc1}} -\add{This section describes the frequency axis indexing scheme that was initially explored before +This section describes the frequency axis indexing scheme that was initially explored before the current standardized version. Documenting it here serves two purposes: \begin{itemize} \item to document earlier versions of FMOC and SFMOC files that may still exist; \item to record an already explored solution and thus avoid attempting to reintroduce it in the future. -\end{itemize}} +\end{itemize} -\add{The original idea was to (almost) directly use, as the index at the deepest level, +The original idea was to (almost) directly use, as the index at the deepest level, the integer representation of the 64-bit floating-point value of a frequency expressed in Hz. Performing MOC operations (except \textit{degrade}) on floating-point ranges is equivalent to performing them on the corresponding integer representation ranges because: @@ -15,9 +15,9 @@ \section{\add{Previous frequency indexing convention}\label{sec:apx:fmoc1}} \item frequencies are strictly positive values; \item for positive floating-point numbers, the numerical ordering is identical to the ordering of their integer bit representations. -\end{itemize}} +\end{itemize} -\add{The only difference was that the exponent range of a 64-bit floating-point value was considered too +The only difference was that the exponent range of a 64-bit floating-point value was considered too large for the intended use. Therefore, it was restricted to raw exponent values $\in [929, 1184]$ instead of the full IEEE~754 range $\in [0, 2047]$. The selected interval contains 255 distinct values and therefore requires 8 bits @@ -27,7 +27,7 @@ \section{\add{Previous frequency indexing convention}\label{sec:apx:fmoc1}} $f \in [5.0487097934144756 \times 10^{-29},\; 5.846006549323611 \times 10^{48}[$. The following pseudo-code describes the transformation, which replaces the 11-bit exponent $\in [929,1184]$ of a floating-point value with an 8-bit exponent -$\in [0,255]$.} +$\in [0,255]$. \lstdefinelanguage{Pseudocode}{ keywords={input, output, return, datatype, function, in, if, else, foreach, while, begin, end }, @@ -77,21 +77,20 @@ \section{\add{Previous frequency indexing convention}\label{sec:apx:fmoc1}} end \end{lstlisting} - -\add{One advantage of this approach was the existence of a perfect bijection between the +One advantage of this approach was the existence of a perfect bijection between the 64-bit floating-point representation of a frequency (in Hz) within the valid range and its index at the deepest resolution. The conversions were also extremely fast, requiring only a few bitwise operations. In addition, the number of orders (60) was not limited to -the number of mantissa bits.} +the number of mantissa bits. -\add{However, a major issue arose from the representation of the axis using WCS, as used in HiPS 2.0. +However, a major issue arose from the representation of the axis using WCS, as used in HiPS 2.0. Because the representation allocates 8 bits to the exponent and 52 bits to the mantissa, the resulting WCS cannot be expressed purely as either linear or logarithmic. For orders 0 to 7, the WCS behaves logarithmically. For orders 8 to 59, inside a cell or order at least 7, it behaves linearly. For ranges spanning order 7, however, the WCS becomes a mixture of logarithmic and linear components, which cannot be represented easily. -This limitation was the primary motivation for abandoning this approach.} +This limitation was the primary motivation for abandoning this approach. diff --git a/appendix_json.tex b/appendix_json.tex index 8e259bb..e912b7b 100644 --- a/appendix_json.tex +++ b/appendix_json.tex @@ -11,8 +11,8 @@ \section{JSON encoding} of the MOC (MOCORDER) is greater than the greatest stored order, the MOCORDER should be provided with an empty index list. -\addb{Although we expect index arrays to be sorted, we cannot impose -the same constraint on orders, since JSON object members are considered unordered.} +Although we expect index arrays to be sorted, we cannot impose +the same constraint on orders, since JSON object members are considered unordered. \paragraph{Example of a JSON SMOC or TMOC:} \par\noindent diff --git a/appendix_packing.tex b/appendix_packing.tex index 7f0cf3b..1697b0b 100644 --- a/appendix_packing.tex +++ b/appendix_packing.tex @@ -1,42 +1,42 @@ -\section{\add{SUNIQ and ZUNIQ packaging}} +\section{SUNIQ and ZUNIQ packaging} -\add{Although they are not used in this standard, the SUNIQ and ZUNIQ packing schemes +Although they are not used in this standard, the SUNIQ and ZUNIQ packing schemes are implemented in some MOC libraries and are therefore briefly described here. -They are valid with all quantities.} +They are valid with all quantities. -\subsection{\add{SUNIQ packaging}} +\subsection{SUNIQ packaging} -\add{SUNIQ provides a generalized alternative to the UNIQ packing, which is specific to HEALPix indices. +SUNIQ provides a generalized alternative to the UNIQ packing, which is specific to HEALPix indices. Its main drawback is that it requires one additional bit. SUNIQ relies on a sentinel bit (hence the “S” in SUNIQ). -Given a cell index $index$, the SUNIQ value is obtained by setting a sentinel bit immediately to the +Given a cell index $index$, the SUNIQ value is obtained by setting a sentinel bit immediately to the left of the most significant bit (MSB) of the index. The resulting expressions for HEALPix and TMOC/FMOC indices are: \begin{eqnarray*} suniq_{hpx} & = & 4^{order + 2} + index \\ suniq_{t/f} & = & 2^{order + 1} + index \end{eqnarray*} -} -\add{The inverse operation uses the {\textit leading\_zeros} function, +The inverse operation uses the {\textit leading\_zeros} function, which returns the number of leading zeros in the binary representation of a 64-bit (unsigned) integer. \begin{eqnarray*} order_{hpx} & = & \frac{59 - leading\_zeros(suniq)}{2} \\ index_{hpx} & = & suniq - 4^{order + 2} \\ order_{t/f} & = & 62 - leading\_zeros(suniq) \\ index_{t/f} & = & suniq - 2^{order + 1} -\end{eqnarray*}} +\end{eqnarray*} -\add{Sorting UNIQ or SUNIQ values produces the same ordering of cells: cells of lower order appear first, and within a given order, cells are ordered by increasing index.} +Sorting UNIQ or SUNIQ values produces the same ordering of cells: cells of lower order appear first, +and within a given order, cells are ordered by increasing index. -\subsection{\add{ZUNIQ packaging}} +\subsection{ZUNIQ packaging} -\add{ZUNIQ encodes both the cell index and its order in a single integer while preserving the +ZUNIQ encodes both the cell index and its order in a single integer while preserving the ordering of cells according to their ranges at the maximum resolution. -HEALPix indices correspond to the concatenation of 12 Z-order curves and therefore follow the +HEALPix indices correspond to the concatenation of 12 Z-order curves and therefore follow the natural ordering of a multi-resoltuion Morton (Z-order) curve, which motivates the name ZUNIQ. -ZUNIQ is constructed from the starting index of the cell at the maximum resolution, left-shifted by one bit, +ZUNIQ is constructed from the starting index of the cell at the maximum resolution, left-shifted by one bit, with the bit at the right of the least significant bit used as a sentinel. This scheme is similar to the numbering used by \textit{S2CellId} in the Google \textit{S2Geometry} library. @@ -50,9 +50,8 @@ \subsection{\add{ZUNIQ packaging}} zuniq_{hpx} & = & 2 \times index_{order_{max}} + 4^{order_{max}-order} \\ zuniq_{t/f} & = & 2 \times index_{order_{max}} + 2^{order_{max}-order} \end{eqnarray*} -} -\add{The inverse transformation uses the {\textit trailing\_zeros} function, +The inverse transformation uses the {\textit trailing\_zeros} function, which returns the number of trailing zeros in the binary representation of a 64-bit unsigned integer: \begin{eqnarray*} order_{hpx} & = & 29 - \frac{trailing\_zeros(zuniq)}{2} \\ @@ -62,4 +61,4 @@ \subsection{\add{ZUNIQ packaging}} \end{eqnarray*} Sorting cells according to their ZUNIQ values enables efficient transformations between lists of cells at mixed resolutions and lists of ranges. -More generally, such orderings allow efficient streamed operations.} +More generally, such orderings allow efficient streamed operations. diff --git a/moc.tex b/moc.tex index 9fd4fd6..61b1450 100644 --- a/moc.tex +++ b/moc.tex @@ -42,24 +42,17 @@ \customcss{custom.css} \newcommand\ada[1]{\textcolor{red}{\textbf{#1}}} -\newcommand\daniel[1]{\textcolor{blue}{\textbf{#1}}} -\newcommand\pierre[1]{\textcolor{orange}{\textbf{#1}}} - +\newcommand\daniel[1]{\textcolor{blue}{\textbf{#1}}} +\newcommand\pierre[1]{\textcolor{orange}{\textbf{#1}}} \newcommand\fxp[1]{\textcolor{teal}{\textbf{#1}}} -%\newcommand\add[1]{\fxp{#1}} -% Switch the comment between the following lines to see/hide previous versions of the text -%\newcommand\remove[1]{\xout{#1}} -%\newcommand\replace[2]{\remove{#1} \add{#2}} -\newcommand\add[1]{#1} -\newcommand\remove[1]{} -\newcommand\replace[2]{\add{#2}} - -%\newcommand\addb[1]{\textcolor{purple}{\textbf{#1}}} -%\newcommand\removeb[1]{\textcolor{purple}{\xout{#1}}} -%\newcommand\replaceb[2]{\removeb{#1} \addb{#2}} -\newcommand\addb[1]{#1} -\newcommand\removeb[1]{} -\newcommand\replaceb[2]{\addb{#2}} +%% Switch the comment between the following lines to see/hide previous versions of the text +%%\newcommand\add[1]{\fxp{#1}} +%%\newcommand\remove[1]{\xout{#1}} +%%\newcommand\replace[2]{\remove{#1} \add{#2}} +%\newcommand\add[1]{#1} +%\newcommand\remove[1]{} +%\newcommand\replace[2]{\add{#2}} + \title{MOC: Multi-Order Coverage map} @@ -74,13 +67,13 @@ \author{Baptiste Cecconi (LESIA, Observatoire de Meudon)} \author{Giuseppe Greco (EGO-Virgo)} \author{Tom Donaldson (STScI/NASA)} -\author{\addb{Thomas Robitaille}} +\author{Thomas Robitaille} \author{Mark Taylor (University of Bristol)} \author{Wil O'Mullane (Vera C. Rubin Observatory)} \author{Martin Reinecke (Max Planck)} \author{S\'{e}bastien Derri\`{e}re (CDS)} -\editor{\add{Francois-Xavier Pineau}} +\editor{Francois-Xavier Pineau} \editor{Pierre Fernique} \editor{Ada Nebot} \editor{Daniel Durand} @@ -92,26 +85,26 @@ \begin{document} \begin{abstract} -This document \replace{describes}{defines version 2.1 of} the Multi-Order Coverage map \add{(MOC)} method \remove{(MOC)} -\remove{version 2.0} \replace{to specify}{, which enables the specification of} arbitrary coverages \replace{for}{across} sky regions -\replace{and/or time coverages and potentially other dimensions}{, time intervals and frequency domains}. +This document defines version 2.1 of the Multi-Order Coverage map (MOC) method, +which enables the specification of arbitrary coverages across sky regions, +time intervals and frequency domains. The goal is to be able to provide a very fast comparison mechanism between coverages. The -mechanism \replace{is based}{relies} on \replace{a discretization of}{discretizing} \replace{space}{the unit sphere} and \replace{time -dimensions}{one-dimensional quantities such as time or frequency}. -\add{Specifically, the discretization uses the Z-order curve for 2-dimensional sky regions and on hierarchical dyadic subdivisions for 1-dimensional quantities.} +mechanism relies on discretizing the unit sphere and one-dimensional quantities +such as time or frequency. +Specifically, the discretization uses the Z-order curve for 2-dimensional sky +regions and on hierarchical dyadic subdivisions for 1-dimensional quantities. The system is based on the definition of a specific storage of the map coverage using predefined cells hierarchically grouped which makes it easy to produce and use for exploring astronomical -collections. \remove{There are already a few applications and libraries which -are taking advantage of this major evolution of the MOC standard.} -\add{This version 2.1 extends the use of MOCs to the three physical quantities: -space, time, frequency, as well as their pairwise combinations.} +collections. +This version 2.1 extends the use of MOCs to the three physical quantities: +space, time, frequency, as well as their pairwise combinations. \end{abstract} \listoffigures -\listoftables +\listoftables %\todototoc %\listoftodos[Issues which need attention] @@ -145,25 +138,22 @@ \section*{Conformance-related definitions} \end{figure} \section{Introduction} -This document is a \replace{major}{minor} release of -the already existing encoding method recommendation \emph{Multi-Order - Coverage} map \citep[MOC~\replace{1.1}{2.0},][]{2022ivoa.spec.0727F}. -We generalize the MOC -originally limited to \replace{s}{S}pace \add{, Time, and Space-Time} dimension\add{s} - (Space MOC, SMOC\add{; Time MOC, TMOC; and Space-Time -MOC, STMOC}) to \replace{the time -dimension (Time MOC, TMOC), and space and time dimensions (Space-Time -MOC, STMOC)}{the Frequency, Space-Frequency and Frequency-Time dimensions -(Frequency MOC, FMOC; Space-Frequency, SFMOC; Frequency-Time, FTMOC)}. -Figure~\ref{fig:ivoadiagram} illustrates the role MOC\replace{2.0}{ 2.1} +This document is a minor release of the already existing encoding method +recommendation \emph{Multi-Order Coverage} map +\citep[MOC~2.0,][]{2022ivoa.spec.0727F}. +We generalize the MOC originally limited to Space, Time, and Space-Time +dimensions (Space MOC, SMOC; Time MOC, TMOC; and Space-Time MOC, STMOC) +to the Frequency, Space-Frequency and Frequency-Time dimensions +(Frequency MOC, FMOC; Space-Frequency, SFMOC; Frequency-Time, FTMOC). +Figure~\ref{fig:ivoadiagram} illustrates the role MOC 2.1 plays within the IVOA architecture \citep{2021ivoa.spec.1101D}. The encoding method described in this document allows one to define -and manipulate space \replace{and}{,} time \add{and frequency} coverage in such a way that basic -operations like union, intersection\replace{,}{or} equality test can be performed +and manipulate space, time and frequency coverage in such a way that basic +operations like union, intersection or equality test can be performed very efficiently. This methodology allows VO applications and data servers to build efficient -procedures to perform such operations on observations and catalogs. +procedures to perform such operations on observations and catalogs. In the next sections we will describe the different MOCs and their encoding standards. @@ -174,12 +164,10 @@ \section{The rationale} equality operations between them. In order to accomplish this task, we based the system on a regular and hierarchical discretization as exposed below. -\replace{The standard MOC1.0 was limited to space, but for a -multitude of use cases in astronomy we need the notion of time to be -properly integrated}{The evolution of the MOC standard -- which was initially -limited to spatial quantities in version 1.0, then expanded to include temporal quantities in version 2.0, -and finally supplemented with frequency quantities in version 2.1 --- enables it to cover a wide range of astronomical use cases}, e.~g.: +The evolution of the MOC standard -- which was initially limited to spatial +quantities in version 1.0, then expanded to include temporal quantities in +version 2.0, and finally supplemented with frequency quantities in version +2.1 -- enables it to cover a wide range of astronomical use cases, e.~g.: \begin{itemize} \item What are the space and time coverages of the 2MASS observations and are there any observations which are coincidental with the HST @@ -195,7 +183,8 @@ \section{The rationale} by Galex? By both missions simultaneously? If so, are there detections within the source catalogues? \item Has Neptune been observed by DSS? -\item \add{Is a radio source contaminating my observations in given frequency ranges?} +\item Is a radio source contaminating my observations in given frequency + ranges? \end{itemize} It was possible to answer those questions with MOC 1.0 standard and @@ -203,9 +192,9 @@ \section{The rationale} quite a big hurdle for the researchers. With MOC 2.0 it is possible to answer these questions in a few milli-seconds. Another example of usage would be the visual inspection of the spatio-temporal -coverage of PanSTARRS observations (see Figure~\ref{fig:panstarrs}). +coverage of PanSTARRS observations (see Figure~\ref{fig:panstarrs}). The choice of a temporal resolution and a spatial resolution makes it -possible to obtain a MOC of a desired data volume. +possible to obtain a MOC of a desired data volume. \begin{figure}[!ht] \begin{center} @@ -225,7 +214,7 @@ \subsection{Comparing the coverage of multiple data sets} possible to compute the intersection of Saturn's ephemeris and the spatio-temporal coverage of HST ACS observations, and subsequently, query a database for retrieving images for which time and position -fall within this intersection (see Fig.~\ref{fig:stmoc-saturn}). +fall within this intersection (see Fig.~\ref{fig:stmoc-saturn}). \begin{figure}[h] \begin{center} @@ -260,53 +249,19 @@ \subsection{Gravitational Wave localisations} Every single level of probability can be used as a regular MOC even in the case in which the sky localization is irregularly shaped with disjoint regions. This coding technique allows for an extra fast integration in the existing -Virtual Observatory structures and tools\footnote{\url{https://emfollow.docs.ligo.org/userguide/}}. -The 2D contours of a GW sky localization can be visualised +Virtual Observatory structures and +tools\footnote{\url{https://emfollow.docs.ligo.org/userguide/}}. +The 2D contours of a GW sky localization can be visualised and manipulated using Aladin Desktop, allowing one to compare them with existing surveys, overlap sky map generations with increasing accuracy -and computational cost and query the VizieR database. +and computational cost and query the VizieR database. These sets of tasks can also be performed via Python using the astropy affiliated package mocpy\footnote{\url{https://cds-astro.github.io/mocpy/}}, efficiently displayed in javascript applications with Aladin Lite, and integrated within Jupyter notebooks through the -ipyaladin\footnote{\url{https://cds-astro.github.io/ipyaladin/}} widget\footnote{\url{https://pos.sissa.it/357/031/pdf}}. - -\remove{Adding temporal information in a space MOC encoding can provide a -systematic approach in extracting information from follow-up campaigns -involving tens of ground and space-based observatories in searching -for gravitational-wave counterparts. -For illustrative purposes only, a potential use of the time-space MOC -is provided when a kilo/macronova emission is a credible counterpart of a -gravitational-wave event. -Figure~\ref{fig:gw} shows a mock electromagnetic -follow-up campaign of a gravitational-wave sky localization over a period -of time (left panel), a schematic kilonova light-curve and -temporal coverage of observations (top right panel) and associated -spatial coverage (bottom right panel). This approach permits us to depict -the approximate timeline of the instruments involved in the observational -campaign and place constraints on the emission properties during the -source evolution.} - -%\begin{figure}[!htbp] -%\begin{center} -%\includegraphics[width=\textwidth]{kilo.png} -%\end{center} -%\caption{A mock electromagnetic follow-up campaign -% of a gravitational-wave sky localization over a time period (left). -% A schematic kilonova light-curve with the observations temporal -% coverage (top right) and associated spatial coverage (bottom right).} -%\label{fig:gw} -%\end{figure} - -%\begin{figure}[!htbp] -%\begin{center} -%\includegraphics[width=0.5\textwidth]{kilo1.png} -%\includegraphics[width=0.45\textwidth]{kilo4.png} -%\caption{A schematic kilonova light-curve, with the -% associated coverage in time (left panel) and space (right panel).} -%\label{fig:kilonova} -%\end{center} -%\end{figure} +ipyaladin\footnote{\url{https://cds-astro.github.io/ipyaladin/}} +widget\footnote{\url{https://pos.sissa.it/357/031/pdf}}. + \subsection{Space and Time MOC: Einstein Telescope and Early Warning Alerts} The space and time MOC provides us with an effective way to develop new @@ -319,7 +274,7 @@ \subsection{Space and Time MOC: Einstein Telescope and Early Warning Alerts} neutron star mergers per year \citep{2020JCAP...03..050M}. For fast and real time data access, the user can query by a specific time range the gravitational-wave sky localizations encoded as a space -and time MOC. +and time MOC. In addition, the ET sensitivity at low frequencies enables enough signal-to-noise ratio to accumulate before the merger, making possible @@ -336,32 +291,30 @@ \subsection{Space and Time MOC: Einstein Telescope and Early Warning Alerts} Early warning alerts are also planned in the LIGO-Virgo-KAGRA O4 run with an experimental capability to produce and distribute early warning gravitational-wave alerts up to tens of seconds before -merger\footnote{\url{https://emfollow.docs.ligo.org/userguide/early_warning.html}}. +merger\footnote{\url{https://emfollow.docs.ligo.org/userguide/early_warning.html}}. - -\subsection{\replace{Multi-site positional and temporal search}{Multi-site search in space, time and/or frequency}} +\subsection{Multi-site search in space, time and/or frequency} Often, a typical query from a virtual observatory (VO) user is to request all possible records from the VO at a given sky position -and/or at a given time \add{, and/or at a given frequency}. -While this is in principle possible by -dispatching one narrow positional query to every registered Cone -Search, SSA or SIA service and filtering by time, in practice, the -number of queries required leads to an unacceptable load on both -clients and services. Moreover, most of these queries will deliver no -results since most services often lack coverage in the queried +and/or at a given time, and/or at a given frequency. +While this is in principle possible by dispatching one narrow positional +query to every registered Cone Search, SSA or SIA service and filtering +by time, in practice, the number of queries required leads to an unacceptable +load on both clients and services. Moreover, most of these queries will +deliver no results since most services often lack coverage in the queried region. If basic footprint/coverage information was available for all registered services, for instance using VODataService 1.2 \citep{2021ivoa.spec.1102D} only those with coverage in the region -and time\add{/frequency} of interest would then be queried. This would provide a great -reduction in the number of services to be queried optimizing the +and time/frequency of interest would then be queried. This would provide a +great reduction in the number of services to be queried optimizing the response time. Using the MOC offers the opportunity to provide this coverage information in a uniform way. The MOC could be stored locally for a given service or centrally where the coverage for a number of services would be supported. -\subsection{\add{Low frequency radio astronomy}} +\subsection{Low frequency radio astronomy} -\add{Ground-based low-frequency radio astronomy instruments typically span over +Ground-based low-frequency radio astronomy instruments typically span over a wide spectral range, such as from 10 MHz to 85 MHz, with NenuFAR \citep{zarka_2028}, which covers a factor of 8.5 in spectral domain. The main beam of the instrument has a field of view that varies by almost one @@ -374,9 +327,9 @@ \subsection{\add{Low frequency radio astronomy}} assessment by cross-matching a series of spatial MOCs (for the grating lobes and the interfering bright radio sources) with a temporal and spectral sampling. Merging the spatial and spectral dimensions into an spatio-frequency MOC -for a series of observation times would simplify and speed up this task.} +for a series of observation times would simplify and speed up this task. -\add{The low-frequency radio sources observed in our solar system, +The low-frequency radio sources observed in our solar system, such as the Sun or the magnetised planets, but also those of active stars and some exoplanetary systems, are showing very dynamical structures in the temporal-spectral domain, at time scales spanning from milliseconds to hours. @@ -387,68 +340,48 @@ \subsection{\add{Low frequency radio astronomy}} those features, cross-matching the catalogues is still a difficult process. Transforming each of the catalogues into an frequency-time MOC would allow very efficient identification of possibly simultaneous events listed -in different catalogues.} +in different catalogues. \section{MOC principles} -\add{Irrespective of the quantity,} the MOC standard is defined using four basic building blocks: -discretization, unique reference system, hierarchization and efficient -encoding: - +Irrespective of the quantity, the MOC standard is defined using four basic +building blocks: discretization, unique reference system, hierarchization +and efficient encoding: \begin{enumerate} \item Determine a proper tessellation/discretization methodology for - each \replaceb{dimension axis}{quantity} (space, time, \add{frequency}, ...); -\item Fix a unique referential system for each \replaceb{dimension}{quantity}, to avoid + each quantity (space, time, frequency, ...); +\item Fix a unique referential system for each quantity, to avoid reference conversions and thus allowing to easily compare different - data collections; + data collections; \item Use an hierarchical procedure and a unique representation - (canonical form) for compacting and quickly manipulating \removeb{each axis} - coverage\addb{s of a given quantity} at any level of accuracy; + (canonical form) for compacting and quickly manipulating + coverages of a given quantity at any level of accuracy; \item Implement at least one serialization in a binary encoding format - (other serializations are possible, e.g. ASCII). -\end{enumerate} + (other serializations are possible, e.g. ASCII). +\end{enumerate} With these principles, a MOC consists of a list of numbers which -represent the indices of the cells mapping the coverage of the spatial -\replace{or}{, } temporal \add {or frequency} axis. +represent the indices of the cells mapping the coverage of the spatial, +temporal or frequency axis. As soon as the consecutive cells are used at order n, they will be hierarchically grouped in their parent cell at order n-1, and this recursively. This introduces the notion of orders and associated cell index boundary alignment implied by the hierarchical structure -facilitates the combination of cells at different orders. - -\remove{To work efficiently on existing hardware, we encode of any pair -(order, index) as a long integer (64 bits), and we reserve the two -most significant bits to encode the type of MOC (spatial, temporal, -or future usages). -The earlier MOC standard was limited to spatial coverage. -We are reusing these principles to manipulate temporal coverages, -as well as space-time coverages where we can manipulate the two -physical dimensions simultaneously.} - - -%To work efficiently on existing hardware, we encode of any pair -%(order, index) as a \replace{long}{64-bits} integer \rm{(64 bits)}, and we reserve the two -%most significant bits to encode the type of MOC (spatial, temporal, -%\add{frequency} or \replace{future usages}{custom}). -%The earlier MOC standard was limited to spatial\add{, time, and space-time} coverage\add{s}. -%We are reusing these principles to manipulate \replace{temporal}{frequency} coverages, -%as well as \replace{space-time}{space-frequency and frequency-time} coverages where we can manipulate the two -%physical dimensions simultaneously. +facilitates the combination of cells at different orders. \medskip \par\noindent -We will now explain the conventions chosen for \rm{the spatial, temporal and frequency axis} -\add{single quantity MOCs, namely space, time and frequency MOCs} -\replaceb{designed}{designated} \add{by the acronyms SMOC, TMOC and FMOC. +We will now explain the conventions chosen for single quantity MOCs, +namely space, time and frequency MOCs designated by the acronyms SMOC, +TMOC and FMOC. Then, we also explain the conventions chosen for two-dimensional MOCs, -namely space-time, space-frequency and frequency-time MOCs} -\replaceb{designed}{designated} \add{by the acronyms STMOC, SFMOC and FTMOC.} +namely space-time, space-frequency and frequency-time MOCs +designated by the acronyms STMOC, SFMOC and FTMOC. -\subsection{\add{Single quantity MOCs}} +\subsection{Single quantity MOCs} -\add{In this section, we describe the convention relative to single quantity MOCs.} +In this section, we describe the convention relative to single quantity MOCs. \subsubsection{Space MOC conventions} Defining a sky region by a subset of regular sky tessellation or tiles @@ -472,13 +405,13 @@ \subsubsection{Space MOC conventions} \item Accuracy: HEALPix provides libraries which allow the calculation up to accuracy of 0.4 mas (order 29) (\url{http://sourceforge.net/projects/healpix/}); \item Standard: Existence of many HEALPix libraries: C++, Java, - Fortran, \replace{IDL}{Python, Rust, Web Assembly, }... + Fortran, Python, Rust, Web Assembly, ... Also, HEALPix was selected for several all sky - missions such as WMAP, Planck and Gaia. The HEALPix main web site \replaceb{is}{, previsously} - located at Jet Propulsion Lab\addb{oratory}\addb{, is now on sourceforge} (\url{http://healpix.sourceforge.io/}). + missions such as WMAP, Planck and Gaia. The HEALPix main web site, previsously + located at Jet Propulsion Laboratory, is now on sourceforge (\url{http://healpix.sourceforge.io/}). \end{itemize} -The HEALPix \citep{2005ApJ...622..759G} \add{NESTED} tessellation technique divides +The HEALPix \citep{2005ApJ...622..759G} NESTED tessellation technique divides the sphere into 12 cells, each of them sub-divided into 4 cells recursively (see Fig.~\ref{fig:healpix}). Thus the sphere at order 1 will consist of 48 cells, @@ -493,17 +426,15 @@ \subsubsection{Space MOC conventions} \label{fig:healpix} \end{figure} -% MOVED HERE FROM v2.1-v2 -\add{In practice, the sub-division of each base cell -- -which are squares in the Euclidean HEALPix projection plane -- consists in -iterations of a Z-order curve, and bit interleaving allows for -a non-recursive algorithm.} -\addb{When encoding a "NESTED" HEALPix index, the four Most Significant Bits (MSB) are used to code + +In practice, the sub-division of each base cell -- which are squares in the +Euclidean HEALPix projection plane -- consists in iterations of a Z-order +curve, and bit interleaving allows for a non-recursive algorithm. +When encoding a "NESTED" HEALPix index, the four Most Significant Bits (MSB) are used to code the base cell index ($ \in [0, 12[)$ and the remaining LSBs are the 2-dimensional Morton code of the -coordinates inside the base cell.} +coordinates inside the base cell. -% MOVED HERE FROM v2.1-v2 -\replace{This}{Thus, the "NESTED"} numbering consists of enumerating all cells in a +Thus, the "NESTED" numbering consists of enumerating all cells in a specific order. For instance, at order 1, there are 48 cells (12x4) enumerated from 0 to 47. In this scheme, the 4 sub-cells of cell M have the indices: (M$\times4)+3$, (M$\times4)+1$, (M$\times4)+2$, @@ -520,16 +451,6 @@ \subsubsection{Space MOC conventions} \caption{HEALPix numbering principle} \end{figure} -\remove{HEALPix allows three coordinate systems: galactic, equatorial and -ecliptic. Allowing various coordinate systems would limit the -possibility to compare efficiently SMOCs. There is indeed no -equivalence between an HEALPix cell described in a given coordinate -system and a cell, or a list of sub-cells expressed in a different -coordinate system. Consequently, the SMOC definition is expressed in -equatorial coordinate using the ICRS reference. This choice has been -motivated by looking at most catalogs and realizing that most of them -are using equatorial coordinates.} - To support the encoding based on 64-bit longs, the best resolution available is provided at order 29 and according to the HEALPix equations, corresponds approximately to 0.4mas. The SMOC resolution @@ -543,7 +464,7 @@ \subsubsection{Space MOC conventions} each HEALPix order. In Figure~\ref{fig:smoc-view} we show the MOC creation, from images to their coverage, -and indicating their corresponding HEALPix numbers. +and indicating their corresponding HEALPix numbers. \begin{figure}[h] \begin{center} @@ -557,20 +478,13 @@ \subsubsection{Space MOC conventions} \input{table1.tex} \input{table2.tex} -\replaceb{The mechanism used to determine which HEALPix cell contains a given -The formula computing a NESTED HEALPix index provided a given $order$ -and}{The formula computing a cell index from a} sky location is described in +The formula computing a cell index from a sky location is described in the main article defining the HEALPix system \citep{2005ApJ...622..759G}. -\replaceb{Several support libraries supporting the most important set of primitives are -already available. -These libraries are required if one wants to generate SMOCs, and also -wants to compare them with sky coordinates. Though please -note that these libraries do not have built-in support to performing -basic SMOC arithmetics.}{Several support libraries implementing the most +Several support libraries implementing the most important primitives are already available. These libraries are required for generating SMOCs and for comparing them with sky coordinates. However, they do not provide built-in support for performing basic -SMOC arithmetic operations.} +SMOC arithmetic operations. @@ -591,7 +505,7 @@ \subsubsection{Time MOC conventions} 1$\mu$s resolution, enough for most astronomical time events. At the deepest order (61) the TMOC cell number is the number of $\mu$s since JD=0. -\addb{Hence, ab integer time value $t$ in $\mu$s since JD=0 is valid if +Hence, ab integer time value $t$ in $\mu$s since JD=0 is valid if $t \in [0, 2^{62}[$. For a given $order$, its cell $index$ is: \begin{equation} index = \frac{t}{2^{order}} @@ -604,40 +518,34 @@ \subsubsection{Time MOC conventions} t_{max} &= (index + 1) \times 2^{order} \label{eq:tmoc:t} \end{align}. -} -% MOVED HERE fom v2.1-v2 -\addb{We note that two consecutive cells at order N with the indices +We note that two consecutive cells at order N with the indices (M$\times$2)+1, (M$\times$2) will be coded at order N-1 with the index M and thus recursively. The order N-1 cell duration is 2 times more -than the N cell duration (61: 1 $\mu$s, 60: 2 $\mu$s, 59: 4 $\mu$s, etc...).} +than the N cell duration (61: 1 $\mu$s, 60: 2 $\mu$s, 59: 4 $\mu$s, etc...). -\remove{The time is a relative observation, and depends on the position of the +The time is a relative observation, and depends on the position of the observer. There are many time scales for measuring time: Terrestrial Time (TT), Barycentric Coordinate Time (TCB), Geocentric Coordinate Time (TCG), Ephemeris Time (ET), Barycentric Dynamic Time (TDB), International Atomic Time (AI), etc. We opt for the TCB -reference} (see \cite{2015A&A...574A..36R} for details). -\remove{Our choice is motivated by the fact that this system is linear by -construction and has been adopted by numerous missions such as Gaia.} +reference (see \cite{2015A&A...574A..36R} for details). -% MOVED HERE fom v2.1-v2 -TMOC {\bf must} be based on \removeb{JD system,} the time -scale TCB (see section~\ref{sec:tcb}) and the Solar System Barycenter as the reference -position \citep[see also][]{2019ASPC..523..497F} and section~\ref{sec:tmoc1}. -%The best resolution supported by TMOC is $1 \mu$s. +TMOC {\bf must} be based on the time scale TCB (see section~\ref{sec:tcb}) +and the Solar System Barycenter as the reference position +\citep[see also][]{2019ASPC..523..497F} and section~\ref{sec:tmoc1}. It may be necessary to convert the temporal events to the chosen scale. If the ephemeris required for this conversion are not available, opt to degrade the accuracy of the time measurement (typically around 20 minutes for observations within the Earth orbit environment to cover all possible observer positions). -Table~\ref{table:tmocorders} -\replaceb{is showing some time values at a given order}{shows the time resolutions of the cells for each possible order}. +Table~\ref{table:tmocorders} shows the time resolutions of the cells +for each possible order. In Figure~\ref{fig:tmoc-view} we show the creation of TMOC, from a time series to the list of numbers based on time discretization. - + \begin{figure}[h!] \begin{center} \includegraphics[width=\textwidth]{tmoc_view.png} @@ -648,16 +556,16 @@ \subsubsection{Time MOC conventions} \label{fig:tmoc-view} \end{figure} -\subsubsection{\add{Frequency MOC conventions}\label{sec:fmoc1}} +\subsubsection{Frequency MOC conventions\label{sec:fmoc1}} -\add{As for Time, representing a frequency coverage requires defining \emph{a priori} +As for Time, representing a frequency coverage requires defining \emph{a priori} the total frequency range to be described by the notation. However, unlike Time -- which is expressed as milliseconds since a reference epoch and therefore has a constant absolute resolution (1 millisecond) -- frequency values are represented as floating-point numbers with an approximately constant \emph{relative} resolution (i.e. a fixed number of significant digits), -rather than a constant absolute resolution.} +rather than a constant absolute resolution. -\add{The adopted\footnote{To handle floating-point quantities, two approaches have been proposed. +The adopted\footnote{To handle floating-point quantities, two approaches have been proposed. The first implemented solution is described in Appendix~\ref{sec:apx:fmoc1}. It was later replaced by the more WCS-compliant solution presented above.} approach is fully generic. First, we define the global valid frequency range $[f_{min}, f_{max}[$. @@ -666,195 +574,160 @@ \subsubsection{\add{Frequency MOC conventions}\label{sec:fmoc1}} \[ g: [f_{min}, f_{max}[ \rightarrow [0,1[ \] -such that any frequency $f \in [f_{min}, f_{max}[$ is associated with the dyadic interval containing $g(f)$.} +such that any frequency $f \in [f_{min}, f_{max}[$ is associated with the dyadic interval containing $g(f)$. -\add{For a given $order$, the interval $[0,1[$ is subdivided into $2^{order+1}$ dyadic intervals (the MOC cells), +For a given $order$, the interval $[0,1[$ is subdivided into $2^{order+1}$ dyadic intervals (the MOC cells), indexed from $0$ to $2^{order+1}-1$. -The index corresponding to a frequency $f$ at order $order$ is therefore:} +The index corresponding to a frequency $f$ at order $order$ is therefore: -\add{\begin{equation} +\begin{equation} index = \lfloor g(f) \times 2^{order+1} \rfloor \label{eq:fmoc:i} -\end{equation}} +\end{equation} -\add{The adopted bounds of the valid frequency range are: +The adopted bounds of the valid frequency range are: \begin{itemize} \item $f_{min} = 10^{-18}$ Hz, corresponding approximately to the inverse of the age of the Universe. \item $f_{max} = 10^{+38}$ Hz, a somewhat arbitrary upper limit chosen so that the full range spans 56 decades in logarithmic scale. -\end{itemize}} +\end{itemize} -\add{The mapping function $g$ is defined as a linear interpolation in logarithmic space: +The mapping function $g$ is defined as a linear interpolation in logarithmic space: \begin{equation} - g(f) = \frac{\log_{10}(f) - \log_{10}(f_{min})}{\log_{10}(f_{max}) - \log_{10}(f_{min})} - %= \frac{\ln(f) - \ln(f_{min})}{\ln(f_{max}) - \ln(f_{min})} + g(f) = \frac{\log_{10}(f) - \log_{10}(f_{min})}{\log_{10}(f_{max}) - \log_{10}(f_{min})} \label{eq:fmoc:g} -\end{equation}} +\end{equation} -\add{Combining Eq.~(\ref{eq:fmoc:i}) and Eq.~(\ref{eq:fmoc:g}) with the selected frequency range, +Combining Eq.~(\ref{eq:fmoc:i}) and Eq.~(\ref{eq:fmoc:g}) with the selected frequency range, we obtain, for a given $order$: \begin{equation} - index = \left\lfloor + index = \left\lfloor \frac{\log_{10}(f) + 18}{56} \times 2^{order+1} \right\rfloor - %= \left\lfloor - %\frac{\ln(f \times 10^{18})}{56 \ln(10)} \times 2^{order+1} - %\right\rfloor -\end{equation}} +\end{equation} -\add{The number of consecutive integers that can be exactly represented by an IEEE~754 floating-point -is limited by the number of bits in the mantissa. -For double-precision floating-points, this number is 52.} +The number of consecutive integers that can be exactly represented by an +IEEE~754 floating-point is limited by the number of bits in the mantissa. +For double-precision floating-points, this number is 52. -\add{Notably, for a fixed exponent, the fractional part of a floating-point number corresponds to -a dyadic rational. Consequently, there is no benefit in defining a MOC $order$ with more cells -than the number of dyadic rationals representable with 52 bits. Beyond this limit, the cells -would be smaller than the intrinsic precision of double-precision floating-point numbers.} +Notably, for a fixed exponent, the fractional part of a floating-point number +corresponds to a dyadic rational. Consequently, there is no benefit in defining +a MOC $order$ with more cells than the number of dyadic rationals representable +with 52 bits. +Beyond this limit, the cells would be smaller than the intrinsic precision of +double-precision floating-point numbers. -\add{For this reason, the maximum order is limited to 52 levels, ranging from 0 to +For this reason, the maximum order is limited to 52 levels, ranging from 0 to $order_{max} = 51$. -At the maximum resolution, the index corresponding to a frequency $f$ is therefore computed as: +At the maximum resolution, the index corresponding to a frequency $f$ is +therefore computed as: \begin{align} order_{max} &= 51 \\ - index_{51} &= \left\lfloor + index_{51} &= \left\lfloor \frac{\log_{10}(f) + 18}{56} \times 2^{52} \right\rfloor -\end{align}} - -\add{Conversely, the frequency range $[f_{min}, f_{max}[$ corresponding to a given index $index$ at -a given order $order$ can be obtained using the inverse transformation: +\end{align} + +Conversely, the frequency range $[f_{min}, f_{max}[$ corresponding to a given +index $index$ at a given order $order$ can be obtained using the inverse +transformation: \begin{eqnarray*} f_{min} & = & g^{-1}(\frac{index}{2^{order + 1}}) = 10^{ 56 \times \frac{index}{2^{order + 1}} - 18} \\ f_{max} & = & g^{-1}(\frac{index + 1}{2^{order + 1}}) = 10^{ 56 \times \frac{index + 1}{2^{order + 1}} - 18} -\end{eqnarray*}} - -\add{Choosing a given order is equivalent to selecting the number of mantissa bits used to represent the mapped value -in the interval $\in [0, 1[$. -In practice, this corresponds to choosing a level of numerical precision, analogous to selecting a -number of significant bits (or digits), noting that approximately one decimal digit of precision is -lost for every 3 to 4 mantissa bits removed. -The approximate number of significant decimal digits corresponding to a given order can be estimated using -the following relation: +\end{eqnarray*} + +Choosing a given order is equivalent to selecting the number of mantissa bits +used to represent the mapped value in the interval $\in [0, 1[$. +In practice, this corresponds to choosing a level of numerical precision, +analogous to selecting a number of significant bits (or digits), noting that +approximately one decimal digit of precision is lost for every 3 to 4 mantissa +bits removed. +The approximate number of significant decimal digits corresponding to a given +order can be estimated using the following relation: \begin{equation} n = \frac{(order + 1)}{\log_2(10)} \end{equation} This expression follows from the conversion between binary and decimal precision, since one decimal digit corresponds to approximately $\log_2(10) \approx 3.32$ bits. -For example, a frequency stored using a 32-bit floating-point number provides about +For example, a frequency stored using a 32-bit floating-point number provides about 24 bits of significand precision, corresponding to approximately 7 decimal digits. To obtain a comparable precision in the hierarchical representation, an order of 29 can be chosen. Additional values are provided in Table~\ref{table:fmocorders}. -} -\add{We can also express, for a given order $order$ the relative size of cell using $g^{-1}$: +We can also express, for a given order $order$ the relative size of cell using $g^{-1}$: \begin{equation} \frac{\Delta f}{f} = \frac{g^{-1}(\frac{index + 1}{2^{order + 1}}) - g^{-1}(\frac{index}{2^{order + 1}})}{g^{-1}(\frac{index}{2^{order + 1}})} = 10^\frac{56}{2^{order + 1}} - 1 \end{equation} -} -\input{tablef.tex} +\input{tablef.tex} -\addb{Unlike time, the finest resolution is not constant across the full frequency range. +Unlike time, the finest resolution is not constant across the full frequency range. Because frequencies are represented using 64-bit floating-point values, the achievable precision is \textit{relative} rather than \textit{absolute}. At the highest resolution (order 51), the relative precision is approximately $10^{-15}$, which is close to the number of significant decimal digits provided by double-precision -floating-point numbers.} +floating-point numbers. -\addb{Like for time, two consecutive cells at order $N$, with indices +Like for time, two consecutive cells at order $N$, with indices $(M\times 2)$ and $(M\times 2)+1$, are represented at order $N-1$ by a single cell with index $M$. -This hierarchical structure applies recursively across all orders.} +This hierarchical structure applies recursively across all orders. -\addb{Considering the dyadic structure of the intervals, a cell at order +Considering the dyadic structure of the intervals, a cell at order $N-1$ spans twice the frequency range of a cell at order $N$ Consequently, the frequency resolution at order -$N-1$ is half that at order $N$.} +$N-1$ is half that at order $N$. -%\paragraph{\add{Remark}} -% -%\add{To handle floating-point quantities, two approaches have been proposed. -%The first implemented solution is described in Appendix~\ref{sec:apx:fmoc1}. -%It was later replaced by the more WCS-compliant solution presented above.} +\subsection{Two dimensional MOCs} -\subsection{\add{Two dimensional MOCs}} - -\add{In this section, we describe the convention adopted\footnote{The use of +In this section, we describe the convention adopted\footnote{The use of a single index obtained by interleaving the indices of two quantities, -similarly to a 2D Z-order curve, was considered but ultimately discarded. -First, the resolutions of both quantities would have been coupled, making it impossible, for example, -to describe an observational coverage with low temporal resolution and high spatial resolution. -Second, encoding two 64-bit indices into a single 64-bit index would have reduced the maximum -achievable resolution for each quantity.} -for two-dimensional MOCs, -i.e. MOCs describing the coverage of two quantities at the same time.} - -\add{To respond to the different use cases presented at the beginning of +similarly to a 2D Z-order curve, was considered but ultimately discarded. +First, the resolutions of both quantities would have been coupled, making it +impossible, for example, to describe an observational coverage with low +temporal resolution and high spatial resolution. +Second, encoding two 64-bit indices into a single 64-bit index would have +reduced the maximum achievable resolution for each quantity.} +for two-dimensional MOCs, i.e. MOCs describing the coverage of two quantities +at the same time. + +To respond to the different use cases presented at the beginning of the document, independent single quantity MOCs are not enough. We need to link two quantities in a global mechanism. -For example, we need to be able to select a SMOC using a time or a frequency window. +For example, we need to be able to select a SMOC using a time or a frequency +window. We also need to select a TMOC or a FMOC using a spatial constraint. Implementing this linkage would allow the potential users to select and interact with the astronomical collections which support space, time and/or frequency and use logical -operators between them.} +operators between them. -\add{Our approach combines two quantities -- time and space, frequency and space, or time and frequency -- -by associating sub-MOCs of the first quantity with MOCs of the second quantity. +Our approach combines two quantities -- time and space, frequency and space, +or time and frequency -- by associating sub-MOCs of the first quantity with +MOCs of the second quantity. For the first quantity, the range of deepest-order indices bounding a sub-MOC must not overlap with the one of another sub-MOC. For the second quantity, MOCs are independent and may overlap. -In the various serializations, a 2D-MOC is represented as a set of tuples, -each consisting of a sub-MOC of the first quantity and a MOC of the second quantity. -Although the resolutions of the first and second quantities may differ, the resolution of each quantity remains uniform across the entire 2D-MOC.} +In the various serializations, a 2D-MOC is represented as a set of tuples, each +consisting of a sub-MOC of the first quantity and a MOC of the second quantity. +Although the resolutions of the first and second quantities may differ, +the resolution of each quantity remains uniform across the entire 2D-MOC. -\addb{In the remaining of the document we adopt the following convention: -a $YX$MOC is a 2D-MOC in which $X$ is the first quantity and $Y$ the second one.} +In the remaining of the document we adopt the following convention: +a $YX$MOC is a 2D-MOC in which $X$ is the first quantity and $Y$ the second one. -%\add{Remark: the use of a single index obtained by interleaving the indices of two quantities, -%similarly to a 2D Z-order curve, was considered but ultimately discarded. -%First, the resolutions of both quantities would have been coupled, making it impossible, for example, -%to describe an observational coverage with low temporal resolution and high spatial resolution. -%Second, encoding two 64-bit indices into a single 64-bit index would have reduced the maximum -%achievable resolution for each quantity.} \subsubsection{Space and Time MOC conventions\label{sec:stmoc}} -\remove{To respond to the different use cases presented at the beginning of -the document, the SMOC and TMOC independently are not enough. We need -to link the two dimensions in a global mechanism. In other words, we -need to be able to select the SMOC using a time window or to select a -TMOC using a spatial constraint. Implementing this linkage would allow -the potential users to select and interact with the astronomical -collections which support space and time and use logical -operators between them.} - -\remove{Our approach is to combine these two dimensions - time and space - by -associating each time period (coded according to the TMOC -convention) with its spatial region (coded according to the SMOC -convention). For that, we interleave the information of time coverage -with the information of space coverage for this period.} - -\remove{This two-dimensional interleaving approach has the advantage of making -the resolutions chosen for time and for space independent. For -instance, it is possible to describe observation coverage with a low -resolution for time while using a high spatial -resolution. -A single coding for indexing space and time simultaneously would imply -at best very low resolution MOCs due to the 64-bit coding constraint. -We thus proposed the interleaving algorithm which allows us to define and -manipulate high resolution STMOCs of reasonable sizes for fast algorithms -(see Appendix~\ref{app:perf} for STMOC performance). -} - -\add{In Space-Time MOCs (STMOCs), the first quantity is time and the second quantity is space. -Consecutive time ranges associated with the same spatial coverage -are grouped into a sub-TMOC linked to their common SMOC. -The STMOC is represented as a sorted set of (sub-TMOC, SMOC) pairs.} +In Space-Time MOCs (STMOCs), the first quantity is time and the second +quantity is space. +Consecutive time ranges associated with the same spatial coverage are +grouped into a sub-TMOC linked to their common SMOC. +The STMOC is represented as a sorted set of (sub-TMOC, SMOC) pairs. In Figure~\ref{fig:stmoc-view} we show the visual representation of an STMOC, -in which at a given \replace{TMOC range}{sub-TMOC} we obtain the corresponding SMOC. +in which at a given sub-TMOC we obtain the corresponding SMOC. \begin{figure}[h] @@ -866,153 +739,41 @@ \subsubsection{Space and Time MOC conventions\label{sec:stmoc}} \label{fig:stmoc-view} \end{figure} -\subsubsection{\add{Space and Frequency MOC conventions}} +\subsubsection{Space and Frequency MOC conventions} -\add{In some use cases, it is desirable to combine the spatial and frequency dimensions. +In some use cases, it is desirable to combine the spatial and frequency dimensions. Similarly to STMOCs -- see section~\ref{sec:stmoc} -- SFMOCs allow the selection of a SMOC within a given frequency window -- or more generally, within a FMOC. Conversely, they can also be used to derive a FMOC subject to spatial constraints. The mechanism is identical to that described in the previous section; -the only difference lies in the nature of the quantities involved.} -\addb{Thus, in Space-Frequency MOCs (SFMOCs), the first quantity is the frequency -and the second quantity is space.} -\add{The type of quantity is specified by a prefix in the ASCII representation or -by a keyword in the FITS serialization.} +the only difference lies in the nature of the quantities involved. +Thus, in Space-Frequency MOCs (SFMOCs), the first quantity is the frequency +and the second quantity is space. +The type of quantity is specified by a prefix in the ASCII representation or +by a keyword in the FITS serialization. -\subsubsection{\add{Frequency and Time MOC conventions}} +\subsubsection{Frequency and Time MOC conventions} -\add{FTMOCs are used to combine the frequency and time dimensions. +FTMOCs are used to combine the frequency and time dimensions. They use the same convention as STMOCs and SFMOCs, with the appropriate quantities prefix -in the ASCII serialization.} -\addb{Thus, in Frequency-Time MOCs, the first quantity is the time -and the second quantity is the Frequency.} - - -\section{\replace{SMOC and TMOC encoding}{MOC serializations}} - -\removeb{The encoding described in this section guarantees backward compatibility -with MOCs corresponding to previous versions of this standard.} -\addb{Two complementary serialization formats are defined for all MOCs: -a string serialization based on ASCII and a binary format based on FITS.} - -\subsection{\replace{SMOC and TMOC encoding}{MOC serializations for a single quantity}} - -%\subsection{Space MOC \remove{or SMOC}} -% -%\remove{As introduced above, the SMOC {\bf should} be based on the -%HEALPix tessellation of the sphere, expressed in the ICRS coordinate -%reference system for celestial coverages. This document does not describe -%the use of SMOC outside celestial scope. However, it is possible to use -%SMOC for other coverages, such as planetary coverages. The definition of -%the unique reference for each body will have to be defined.} -%\remiveb{Two complementary encoding formats are defined: a string serialization based -%on ASCII and a binary format based on FITS.} - -%\subsubsection{Numbering} - -%The numbering scheme used in SMOC for specifying the cell indices {\bf -% must} follow the "NESTED" HEALPix numbering schemes -%\citep{2005ApJ...622..759G}. - -%\add{In a "NESTED" HEALPix index, the four Most Significant Bits (MSB) are used to code the base cell index -%($ \in [0, 12[)$ and the remaining LSBs are the 2-dimensional Morton code of the coordinates inside the base cell. -%Said differently, an HEALPix number comes from the concatenation of 12 Z-order curves, one curve per base cell.} - -%\replace{This}{Thus, the "NESTED"} numbering consists of enumerating all cells in a -%specific order. For instance, at order 1, there are 48 cells (12x4) -%enumerated from 0 to 47. In this scheme, the 4 sub-cells of cell M -%have the indices: (M$\times4)+3$, (M$\times4)+1$, (M$\times4)+2$, -%(M$\times4$) in reading -%order. And reciprocally, the parent index of cell N is -%N/4. Each SMOC cell is coded by a pair of numbers: -%(order, index) which are the HEALPix order and the HEALPix index in this -%order. -% -%\begin{figure}[h] -%\begin{center} -%\includegraphics[width=\textwidth]{nested_healpix.jpg} -%\end{center} -%\caption{HEALPix numbering principle} -%\end{figure} -% -%{\bf Note:} The order 0 is a special case, it contains only 12 -%cells enumerated from 0 to 11. - - -% \subsubsection{Sky coordinates} -% The mechanism used to determine which HEALPix cell contains a given -% sky location is described in the main article defining the HEALPix -% system \citep{2005ApJ...622..759G}. Several support libraries -% supporting the most important set of primitives are already available. -% These libraries are required if one wants to generate SMOCs, and also -% wants to compare them with sky coordinates. Though please -% note that these libraries do not have built-in support to performing -% basic SMOC arithmetics. - - -%\subsection{Time MOC \remove{or TMOC}} -%\remove{As introduced above, }TMOC {\bf must} be based on JD system, the time -%scale TCB (see section~\ref{sec:tcb}) and the Solar System Barycenter as the reference -%position \citep[see also][]{2019ASPC..523..497F} and section~\ref{sec:tmoc1}. -%The best resolution supported by TMOC is $1 \mu$s. -% -%\removeb{In the case that the time scale and the time reference position are unknown, -%we recommend to set the time resolution of the generated TMOC to order -%31, e.g. about 1000 seconds (see Table~\ref{table:tmocorders}) corresponding -%to about twice the light travel time correction between the Earth and -%the Solar Barycenter. Please refer to the VO note on TIMESYS for more -%information about this limitation} \citep{timesysnote}. -% -%\subsubsection{Numbering} -%The numbering scheme used in TMOC for specifying the time cell indices -%{\bf must} reuse a similar hierarchical principle as for the SMOC -%with the difference that the time line has only one dimension, so the -%hierarchical progression uses a factor of 2 instead of 4, and there is -%no need to use a HEALPix mapping. -% -%TMOC has 62 orders, and at the best -%resolution (order 61), a time event will be coded by the integer value -%representing the number of $\mu$s of this event since JD=0. -% -%Two consecutive cells at order N with the indices -% (M$\times$2)+1, (M$\times$2) will be coded at order N-1 with the index -% M and thus recursively. The order N-1 cell duration is 2 times more -% than the N cell duration (61: 1 $\mu$s, 60: 2 $\mu$s, 59: 4 $\mu$s, etc...). - -%\subsection{\add{Frequency MOC}} -% -%\add{As introduced in section~\ref{sec:fmoc1}, the quantity manipulated by a FMOC -%{\bf must} be a frequency, expressed in $Hz$.} -% -%\add{FMOCs supports 52 orders, conresponding to the number of explicitly stored -%mantissa bits in an IEEE 754 double-precision binary floating-point number.} -% -%\add{Unlike time, the finest resolution is not constant across the full frequency range. -%Because frequencies are represented using 64-bit floating-point values, -%the achievable precision is \textit{relative} rather than \textit{absolute}. -%At the highest resolution (order 51), the relative precision is approximately $10^{-15}$, -%which is close to the number of significant decimal digits provided by double-precision -%floating-point numbers.} -% -%\add{Two consecutive cells at order $N$, with indices -%$(M\times 2)$ and $(M\times 2)+1$, are represented at order $N-1$ -%by a single cell with index $M$. -%This hierarchical structure applies recursively across all orders.} -% -%\add{Considering the dyadic structure of the intervals, a cell at order -%$N-1$ spans twice the frequency range of a cell at order $N$ -%Consequently, the frequency resolution at order -%$N-1$ is half that at order $N$.} - - -%\subsection{Serialization} +in the ASCII serialization. +Thus, in Frequency-Time MOCs, the first quantity is the time +and the second quantity is the Frequency. + + +\section{MOC serializations} + +Two complementary serialization formats are defined for all MOCs: +a string serialization based on ASCII and a binary format based on FITS. + +\subsection{MOC serializations for a single quantity} + A MOC can be manipulated and serialized either as a list of cell numbers for each order (hierarchical view), or as a list of intervals at the deepest order (range view). -\add{While} \replace{T}{t}hese two methods are used for -SMOCs, \replace{TMOC and STMOC serializations and are presented -below}{only the intervals approach is used for other MOCs}. -\add{The various possible MOC} serializations are presented below. +While these two methods are used for +SMOCs, only the intervals approach is used for other MOCs. +The various possible MOC serializations are presented below. \subsubsection{Binary serialization} To encode a MOC in a FITS file, each MOC pair (order, index) {\bf @@ -1053,21 +814,21 @@ \subsubsection{Binary serialization} be in ascending numerical order. The resulting list is stored in a single-column binary table extension. -\paragraph{\add{Remark on RANGE vs UNIQs packing}} +\paragraph{Remark on RANGE vs UNIQs packing} -\add{For one-dimensional quantities such as time or frequency, the RANGE representation +For one-dimensional quantities such as time or frequency, the RANGE representation is often\footnote{Except for MOCs that would be dominated by disjoint cells} more compact than any UNIQ-based representation. In one dimension, contiguous cells can always be merged into a single continuous range, -a property that does not generally extend to higher dimensions.} +a property that does not generally extend to higher dimensions. \paragraph{Backward compatibility} -RANGE packaging has been introduced for MOC\replace{2.0}{ 2.0}. +RANGE packaging has been introduced for MOC 2.0. This method is generally faster than the previous one for reading or writing a MOC because the internal representation of MOC in memory is often range oriented. However, we recommend to use the first method for SMOC for compatibility with existing libraries -not yet compatible with MOC\replace{2.0}{ 2.0}. +not yet compatible with MOC 2.0. \subsubsection{ASCII serialization} To encode a MOC as a string each MOC pair (order, index) {\bf must} be written @@ -1080,15 +841,14 @@ \subsubsection{ASCII serialization} The usage of a range operator is allowed in the list of indices using the dash ("-": decimal ASCII code 45) as a separator: lowindex-highindex. -The list of cells {\bf must} be well-formed\replaceb{,}{.} -\addb{Both the orders} and \removeb{the} values {\bf must} -be \addb{sorted} in ascending numerical order\addb{, comparing first the order and -then the cell value}. -\addb{This guarantees the same cell orderering as in the UNIQ FITS serialization, +The list of cells {\bf must} be well-formed. +Both the orders and values {\bf must} be sorted in ascending numerical order, +comparing first the order and then the cell value. +This guarantees the same cell orderering as in the UNIQ FITS serialization, enabling easy conversions in streaming mode. Although these rules are intentionally strict, implementations are encouraged to follow the robustness principle: -be conservative in what they generate and liberal in what they accept.} +be conservative in what they generate and liberal in what they accept. If the best resolution of the MOC (moc order) is greater than the @@ -1098,7 +858,7 @@ \subsubsection{ASCII serialization} cells underneath the explicit pair (order, cell) are implicitly covered up to order 8, the moc order, annotated followed by the terminator "/". Without the terminator we would only have the information of the explicit -pair (order, cell), and the assumed best resolution would be at order 2. +pair (order, cell), and the assumed best resolution would be at order 2. \paragraph{Example of an ASCII MOC:} @@ -1138,39 +898,30 @@ \subsubsection{ASCII serialization} Note that we use Extended BNF supporting regular expression syntax with the following rules: i) postfix * means "repeated 0 or more times”; ii) postfix + means "repeated 1 or more -times”; iii) postfix ? means "0 or 1 times". The first \replace{three}{six} +times”; iii) postfix ? means "0 or 1 times". The first six rules depend on the MOC type, i.e. SMOC for space, TMOC for -time, \add{FMOC for Frequency} \replace{and}{,} STMOC for space-time -\add{,SFMOC for space-frequency and FTMOC for frequency-time}. +time, FMOC for Frequency, STMOC for space-time ,SFMOC for +space-frequency and FTMOC for frequency-time. -\subsection{\replace{STMOC encoding}{MOC serialization for quantity pairs}} +\subsection{MOC serialization for quantity pairs} \label{sec:2dmoce} - -%\removeb{\add{In this section, we first describe the encoding of STMOC. -%The same principles apply to the other possible combinations of quantities.}} - -%\subsection{STMOC encoding} - -Coding \replaceb{STMOC}{a 2D-MOC} consists in the following: -for each \replaceb{element ofa temporal coverage}{sub-MOC of the first quantity} -we list the associated \replaceb{spatial coverage}{MOC of the second quantity} -using the natural packaging as defined in the previous section. +Coding a 2D-MOC consists in the following: +for each sub-MOC of the first quantity we list the associated MOC of the +second quantity using the natural packaging as defined in the previous section. \subsubsection{Binary Serialization} The binary serialization of a STMOC is a FITS binary table following the RANGE packaging presented previously, which interleaves time range(s) and their corresponding space coverage ranges. Following the binary RANGE serialization method described below, each range (time or space) -is coded as two 64-bit signed integers ([min..max[). To distinguish -\replaceb{time}{the first quantity (time)} -and \replaceb{space}{the second quantity (space)} indices, -the \replaceb{time}{first quantity} indices {\bf must} have the 64th bits forced +is coded as two 64-bit signed integers ([min..max[). To distinguish +the first quantity (time) and he second quantity (space) indices, +the first quantity indices {\bf must} have the 64th bits forced to 1. It is not a sign inversion (two's complement) but a mask affecting -only that \replaceb{last}{most significant} bit -without \replaceb{touching}{affecting} any other bits. -The order of dimensions is always \replaceb{time}{the first quantity (time)} first. +only that most significant bit without affecting any other bits. +The order of dimensions is always the first quantity (time) first. \paragraph{Illustration of STMOC interleaving method} @@ -1199,30 +950,12 @@ \subsubsection{Binary Serialization} \end{itemize} This is illustrated in Figure~\ref{fig:stmocbin-encoding}. -%\subsection{\add{SFMOC and FTMOC encoding}} -%\label{sec:sfmoc} - -%\add{The SFMOC and FTMOC serializations are the same as the STMOC serialization}, -%\removeb{using the prefix 'f' for the Frequency quantity.} -%\addb{except that: -%\begin{itemize} -% \item the first quantity of a SFMOC is the Frequency; -% \item the second quantity of a FTMOC is the Frequancy; -% \item the prefix 'f' is used for Frequency in the ASCII serialization. -%\end{itemize} -%} - \subsubsection{ASCII Serialization} -The ASCII serialization of a \replaceb{ST}{2D-}MOC is a string +The ASCII serialization of a 2D-MOC is a string following the ASCII MOC serialization presented below, which interleaves -\replaceb{time coverage}{first quantity sub-MOCs} \removeb{as a -excerpt of TMOC} and associated -\replaceb{space coverage as a SMOC}{second quantity MOCs}. -\removeb{Each TMOC -element {\bf must} be prefixed by the character 't', and each SMOC -element {\bf must} be prefixed by the character 's'.} -\addb{Each sub-MOC and MOC {\bf must} be prefixed by the appropriate character: -'t' for time, 's' for space and 'f' for frequency}. +first quantity sub-MOCs and associated second quantity MOCs. +Each sub-MOC and MOC {\bf must} be prefixed by the appropriate character: +'t' for time, 's' for space and 'f' for frequency. The character is thus omitted until the next dimension element is defined. @@ -1245,17 +978,17 @@ \subsubsection{ASCII Serialization} \section{FITS keywords} -\label{sec:fits-key-moc} +\label{sec:fits-key-moc} For the binary representations which are packaged in binary FITS table, we define a set of FITS keywords, their possible values and set when those fields are required, optional or recommended in Table~\ref{table:fits_stmoc} and show an example of FITS headers for a MOC. Since MOC~1.1 \citep{2019ivoa.spec.1007F} the {\tt PIXTYPE = "HEALPIX"} -keyword/value is no longer required, and should be omitted. +keyword/value is no longer required, and should be omitted. The keyword {\tt MOCORDER} is no longer required either, but it can be used for backwards compatibility if required. -Other FITS Keywords could be used to augment the information like DATES and ORIGIN. - +Other FITS Keywords could be used to augment the information like DATES and ORIGIN. + \input{table3.tex} \clearpage @@ -1296,9 +1029,9 @@ \section{FITS keywords} \section{MOC usage constraints} -\subsection{\add{Reference frames}} +\subsection{Reference frames} -\subsubsection{\add{SMOC and ICRS}} +\subsubsection{SMOC and ICRS} HEALPix supports three coordinate systems: galactic, equatorial, and ecliptic. Supporting multiple coordinate systems in SMOCs would prevent from performing @@ -1306,7 +1039,7 @@ \subsubsection{\add{SMOC and ICRS}} Indeed, a HEALPix cell defined in one coordinate system cannot be represented exactly in another coordinate system. For this reason, the SMOC specification adopts a single reference frame -\add{for celestial coverages}: +for celestial coverages: equatorial coordinates in the ICRS system. This choice reflects common astronomical practice, as the majority of astronomical catalogs are provided in equatorial coordinates. @@ -1316,7 +1049,7 @@ \subsubsection{\add{SMOC and ICRS}} body will have to be defined. -\subsubsection{\add{TMOC and TCB} \label{sec:tcb}} +\subsubsection{TMOC and TCB \label{sec:tcb}} The time is a relative observation, and depends on the position of the observer. There are many time scales for measuring time: @@ -1327,7 +1060,7 @@ \subsubsection{\add{TMOC and TCB} \label{sec:tcb}} choice is motivated by the fact that this system is linear by construction and has been adopted by numerous missions such as Gaia. -\subsubsection{\addb{TMOC of unknown time scale and reference}} +\subsubsection{TMOC of unknown time scale and reference} In the case that the time scale and the time reference position are unknown, we recommend to set the time resolution of the generated TMOC to order @@ -1342,12 +1075,12 @@ \subsection{Canonical form} The speed of MOC operations - creation, union, intersection, etc is directly dependent on the speed of the equality test. It is therefore essential to always express a MOC in a canonical way, ie -one unique representation for one coverage \add{, for both the possible UNIQ (SMOC) and RANGE alternatives}. +one unique representation for one coverage, for both the possible UNIQ (SMOC) and RANGE alternatives. Thus in the case of a hierarchical representation a MOC \textbf{must} be "well-formed", i.e. redundant cells are not allowed, the cells must be ascending sorted and the hierarchical encoding principle must be respected. Thus it is not allowed to encode sibling cells instead of their parent (4 -siblings for SMOC, 2 siblings for TMOC \add{and FMOC}). In the case of range +siblings for SMOC, 2 siblings for TMOC and FMOC). In the case of range representation, the list of ranges must be expressed without overlapping and sorted ascending. @@ -1356,11 +1089,11 @@ \subsection{Compromise of Volume VS. Resolution} In order to easily handle MOCs, it is recommended to adjust the maximum resolution, i.e. the deepest order, to obtain a representation of the desired data volume even if it means degrading the accuracy of the coverage -(see Appendix~\ref{app:perf}). In the case of STMOC \add{(or SFMOC)}, it is possible to adjust -the spatial order and/or on the temporal \add{(or frequency)} order independently. +(see Appendix~\ref{app:perf}). In the case of STMOC (or SFMOC), it is possible to adjust +the spatial order and/or on the temporal (or frequency) order independently. \subsection{Working resolution} -The MOC has been \replaceb{designed}{designated} to be able to efficiently handle observation +The MOC has been designated to be able to efficiently handle observation coverages (images, catalogs, ...). During the construction of the MOC, we must then ensure that at the chosen nominal resolution, any cell of the MOC contains at least one observation (no empty cell). To keep @@ -1381,10 +1114,7 @@ \subsection{Working resolution} \label{fig:operation} \end{figure} -%However, MOC can also be used to manipulate surfaces, unrelated to -%observations. When used this way, nothing prevents working at the best -%resolution and some operations such as oversampling, dilation, -%erosion may be applied. + Note that MOC usage can be diverted to also manipulate surfaces unrelated to observations. When it is used this way such operations (oversampling, surface dilatation or surface erosion) can be applied. @@ -1393,36 +1123,33 @@ \subsection{Working resolution} surfaces operations (commutativity, associativity, ...). - - \section{Summary and conclusion} -We have reviewed the standards for encoding the different MOC flavors\replaceb{,}{:} -\addb{in addition to} the space MOC \replaceb{(already described in version 1.1 of this document), -SMOC, the time MOC, TMOC and the space-time MOC, STMOC}{(SMOC), time MOC (TMOC) +We have reviewed the standards for encoding the different MOC flavors: +in addition to the space MOC (SMOC), time MOC (TMOC) and the Space-Time MOC (STMOC), which were already described in version 2.0 of this document, we introduced the Frequency MOC (FMOC), the Space-Frequency MOC (SFMOC) and the -Frequency-Time MOC (FTMOC)}. +Frequency-Time MOC (FTMOC). The conventions for space and time MOC are the following: \begin{itemize} -\item The SMOC is simply defined \addb{either} as a list of HEALPix indices (order, index) - \addb{or as a sorted list of non-overlapping indices ranges}. +\item The SMOC is simply defined either as a list of HEALPix indices (order, index) + or as a sorted list of non-overlapping indices ranges. \item According to HEALPix, the sphere is divided in cells, hierarchically grouped 4 by 4 with 30 orders and the space coverage for the deepest order is approximatively 0.4mas. \item The space reference system is ICRS. -\item The TMOC is \removeb{simply} a list of \addb{sorted and non-overlapping ranges of} time \removeb{interval} indices \replaceb{(order, index)}{at the deepest order}. +\item The TMOC is a list of sorted and non-overlapping ranges of time indices at the deepest order. \item The time scale is divided in intervals hierarchically grouped 2 by 2 with 62 orders and the time coverage for the deepest order is 1 $\mu$s. \item The time values are defined using JD = 0 as the origin, in Barycentric system. -\item \addb{The FMOC is also a list of sorted and non-overlapping ranges of frequency indices at the deepest order} -\item \addb{The conversion from Frequency, in Hz, to deepest order frequency index is provided by Eq. \ref{eq:fmoc:i}} -\item \addb{Similarly to time, frequency index intervals are grouped 2 by 2 with a total of 52 orders.} +\item The FMOC is also a list of sorted and non-overlapping ranges of frequency indices at the deepest order +\item The conversion from Frequency, in Hz, to deepest order frequency index is provided by Eq. \ref{eq:fmoc:i} +\item Similarly to time, frequency index intervals are grouped 2 by 2 with a total of 52 orders. \end{itemize} Once defined and encoded for a given astronomical collection, one -can easily combine the MOCs for \addb{two of} these \replaceb{two dimensions}{three quantities} -to create a merged \replaceb{STMOC}{2D-MOC} which can be used to navigate and access the collection -through their coverage for both \replaceb{time and space}{its quantities} simultaneously. +can easily combine the MOCs for two of these three quantities +to create a merged 2D-MOC which can be used to navigate and access the collection +through their coverage for both its quantities simultaneously. The possibilities are then very interesting and will be a very valuable astronomical tool. @@ -1432,15 +1159,15 @@ \section{Summary and conclusion} \appendix \section{Version History} -\subsection{\add{Changes between versions 2.0 and 2.1}} +\subsection{Changes between versions 2.0 and 2.1} -\add{The differences between version 2.1 of MOC and the preceding version 2.0 are: +The differences between version 2.1 of MOC and the preceding version 2.0 are: \begin{itemize} \item Introduction of the frequency dimension to define FMOCs, SFMOCs and FTMOCs; \item Introduction of the SUNIQ and ZUNIQ numbering in appendix. \end{itemize} Adding a new quantity (frequency) as an extension to the existing standard, -without modifying existing MOCs, led us to choose a minor version for the document.} +without modifying existing MOCs, led us to choose a minor version for the document. \subsection{Changes between versions 1.1 and 2.0} diff --git a/table3.tex b/table3.tex index 8f1a5b6..7d310dc 100644 --- a/table3.tex +++ b/table3.tex @@ -6,23 +6,23 @@ \hline {\bf Keyword} & {\bf Definition} & {\bf MOC1.1} & {\bf MOC2.0} \\ \hline - MOCVERS & The version of the MOC encoding standard. If it is following \add{the version 2.0 of} this document (TMOC, STMOC and STMOC), it {\bf must} be '2.0'. \add{If it is following this version of this document (FMOC, SFMOC), it {\bf must} be '2.1'.} If not defined it is assumed to be 1.1. & NA & mandatory \\ - MOCDIM & Physical(s) dimension(s). Either 'SPACE' for SMOC, 'TIME' for TMOC \add{, 'FREQUENCY' for FMOC} \replace{or}{,} 'TIME.SPACE' for STMOC \add{, 'FREQUENCY.SPACE' for SFMOC} \addb{or 'TIME.FREQUENCY' for FTMOC}. If omitted, 'SPACE' is assumed for backward compatibility. & NA & mandatory \\ + MOCVERS & The version of the MOC encoding standard. If it is following this version of this document, with the MOC containing the frequecy quantity (FMOC, SFMOC or FTMOC), it {\bf must} be '2.1'. If not defined it is assumed to be 1.1. & NA & mandatory \\ + MOCDIM & Physical(s) dimension(s). Either 'SPACE' for SMOC, 'TIME' for TMOC, 'FREQUENCY' for FMOC, 'TIME.SPACE' for STMOC, 'FREQUENCY.SPACE' for SFMOC or 'TIME.FREQUENCY' for FTMOC. If omitted, 'SPACE' is assumed for backward compatibility. & NA & mandatory \\ ORDERING & The packaging method used. It is either NUNIQ (V1.1 or V2.0 SMOC) or RANGE (V2.0). & mandatory & mandatory \\ - COORDSYS & The coordinate system in use. The value {\bf must} be 'C' for SMOC (ICRS). & mandatory & mandatory \\ + COORDSYS & The coordinate system in use. The value {\bf must} be 'C' for SMOC (ICRS). & mandatory & [mandatory] \\ TIMESYS & The time system in use. The value {\bf must} be 'TCB' for TMOC. & NA & [mandatory] \\ MOCID & Original data identifier. For MOCs that are coverages of VO resources, in particulary those used in VODataService 1.2 coverage elements \citep{2021ivoa.spec.1102D}, MOCID can contain the IVOA id of the VO resource described. & optional & optional \\ MOCTOOL & The name of the MOC software generator. It is also recommended to add the software version number to its name. & optional & optional \\ MOCTYPE & Provenance data type. Either 'IMAGE', or 'CATALOG'. In the first case for areas computed from existing images and/or footprint or even STC strings, in the second case for areas computed from a collection of point sources using a unique and/or derived area. & optional & optional \\ MOCORD\_S & Best resolution of the space dimension, expressed as the order. & NA & [mandatory]\\ MOCORD\_T & Best resolution of the time dimension, expressed as the order. & NA & [mandatory] \\ - \add{MOCORD\_F} & \add{Best resolution of the frequency dimension, expressed as the order.} & NA & [mandatory] \\ + MOCORD\_F & Best resolution of the frequency dimension, expressed as the order. & NA & [mandatory] \\ MOCORDER & Best resolution of the space dimension, expressed as the order. & mandatory & NA \\ PIXTYPE & 'HEALPIX' & mandatory & NA \\ \hline \end{tabular} } \end{center} -\caption[FITS Keywords for MOC]{FITS Keywords for MOC. \addb{'[]' means mandatory if the MOC contains the keyword related quantity.}} +\caption[FITS Keywords for MOC]{FITS Keywords for MOC. Symbol '[]' means mandatory if the MOC contains the keyword related quantity.} \label{table:fits_stmoc} \end{table} From e86e87eb086634dbdd50050e533636e3fb0b4793 Mon Sep 17 00:00:00 2001 From: fxpineau Date: Mon, 1 Jun 2026 16:54:59 +0200 Subject: [PATCH 03/10] Remove redundant text, minor corrections. --- moc.tex | 43 ++++++++++++++----------------------------- 1 file changed, 14 insertions(+), 29 deletions(-) diff --git a/moc.tex b/moc.tex index 61b1450..642cffb 100644 --- a/moc.tex +++ b/moc.tex @@ -89,12 +89,11 @@ which enables the specification of arbitrary coverages across sky regions, time intervals and frequency domains. -The goal is to be able to -provide a very fast comparison mechanism between coverages. The -mechanism relies on discretizing the unit sphere and one-dimensional quantities -such as time or frequency. +The goal is to be able to provide a very fast comparison mechanism between +coverages. The mechanism relies on discretizing the unit sphere and +one-dimensional quantities such as time or frequency. Specifically, the discretization uses the Z-order curve for 2-dimensional sky -regions and on hierarchical dyadic subdivisions for 1-dimensional quantities. +regions and hierarchical dyadic subdivisions for 1-dimensional quantities. The system is based on the definition of a specific storage of the map coverage using predefined cells hierarchically grouped which makes it easy to produce and use for exploring astronomical @@ -149,7 +148,7 @@ \section{Introduction} plays within the IVOA architecture \citep{2021ivoa.spec.1101D}. The encoding method described in this document allows one to define -and manipulate space, time and frequency coverage in such a way that basic +and manipulate space, time and frequency coverages in such a way that basic operations like union, intersection or equality test can be performed very efficiently. This methodology allows VO applications and data servers to build efficient @@ -189,7 +188,7 @@ \section{The rationale} It was possible to answer those questions with MOC 1.0 standard and other VO tools but the amount of manipulation and computation was -quite a big hurdle for the researchers. With MOC 2.0 it is possible +quite a big hurdle for the researchers. With MOC 2.0 and 2.1 it is possible to answer these questions in a few milli-seconds. Another example of usage would be the visual inspection of the spatio-temporal coverage of PanSTARRS observations (see Figure~\ref{fig:panstarrs}). @@ -461,10 +460,8 @@ \subsubsection{Space MOC conventions} the data set coverage. The quality of this approximation depends directly on the chosen SMOC resolution (MOCORD\_S). Table~\ref{table:orders} provides the HEALPix cell angular resolution for -each HEALPix order. - -In Figure~\ref{fig:smoc-view} we show the MOC creation, from images to their coverage, -and indicating their corresponding HEALPix numbers. +each HEALPix order, while Figure~\ref{fig:smoc-view} shows a MOC creation, +from images to their coverage, with their corresponding HEALPix numbers. \begin{figure}[h] \begin{center} @@ -505,7 +502,7 @@ \subsubsection{Time MOC conventions} 1$\mu$s resolution, enough for most astronomical time events. At the deepest order (61) the TMOC cell number is the number of $\mu$s since JD=0. -Hence, ab integer time value $t$ in $\mu$s since JD=0 is valid if +Hence, an integer time value $t$ in $\mu$s since JD=0 is valid if $t \in [0, 2^{62}[$. For a given $order$, its cell $index$ is: \begin{equation} index = \frac{t}{2^{order}} @@ -523,6 +520,8 @@ \subsubsection{Time MOC conventions} (M$\times$2)+1, (M$\times$2) will be coded at order N-1 with the index M and thus recursively. The order N-1 cell duration is 2 times more than the N cell duration (61: 1 $\mu$s, 60: 2 $\mu$s, 59: 4 $\mu$s, etc...). +See Table~\ref{table:tmocorders} for the list of time resolutions +of TMOC cells. The time is a relative observation, and depends on the position of the observer. There are many time scales for measuring time: @@ -531,18 +530,10 @@ \subsubsection{Time MOC conventions} (TDB), International Atomic Time (AI), etc. We opt for the TCB reference (see \cite{2015A&A...574A..36R} for details). -TMOC {\bf must} be based on the time scale TCB (see section~\ref{sec:tcb}) +Thus, TMOC {\bf must} be based on the time scale TCB (see section~\ref{sec:tcb}) and the Solar System Barycenter as the reference position \citep[see also][]{2019ASPC..523..497F} and section~\ref{sec:tmoc1}. -It may be necessary to convert the temporal events to the -chosen scale. If the ephemeris required for this conversion are -not available, opt to degrade the accuracy of the time measurement -(typically around 20 minutes for observations within the Earth orbit -environment to cover all possible observer positions). -Table~\ref{table:tmocorders} shows the time resolutions of the cells -for each possible order. - In Figure~\ref{fig:tmoc-view} we show the creation of TMOC, from a time series to the list of numbers based on time discretization. @@ -567,7 +558,8 @@ \subsubsection{Frequency MOC conventions\label{sec:fmoc1}} The adopted\footnote{To handle floating-point quantities, two approaches have been proposed. The first implemented solution is described in Appendix~\ref{sec:apx:fmoc1}. -It was later replaced by the more WCS-compliant solution presented above.} approach is fully generic. +It was later replaced by the more WCS-compliant solution presented in the core of the document.} +approach is fully generic. First, we define the global valid frequency range $[f_{min}, f_{max}[$. The frequency $f$ {\bf must} be expressed in $Hz$. Next, we introduce a bijective mapping @@ -660,13 +652,6 @@ \subsubsection{Frequency MOC conventions\label{sec:fmoc1}} \input{tablef.tex} -Unlike time, the finest resolution is not constant across the full frequency range. -Because frequencies are represented using 64-bit floating-point values, -the achievable precision is \textit{relative} rather than \textit{absolute}. -At the highest resolution (order 51), the relative precision is approximately $10^{-15}$, -which is close to the number of significant decimal digits provided by double-precision -floating-point numbers. - Like for time, two consecutive cells at order $N$, with indices $(M\times 2)$ and $(M\times 2)+1$, are represented at order $N-1$ by a single cell with index $M$. From 4f3a8004a81d84e1cb46fddb492928f681a60609 Mon Sep 17 00:00:00 2001 From: fxpineau Date: Tue, 2 Jun 2026 08:45:08 +0200 Subject: [PATCH 04/10] Add T. Robitaille affiliation --- moc.tex | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/moc.tex b/moc.tex index 642cffb..94ddfed 100644 --- a/moc.tex +++ b/moc.tex @@ -67,7 +67,7 @@ \author{Baptiste Cecconi (LESIA, Observatoire de Meudon)} \author{Giuseppe Greco (EGO-Virgo)} \author{Tom Donaldson (STScI/NASA)} -\author{Thomas Robitaille} +\author{Thomas Robitaille (Aperio Software)} \author{Mark Taylor (University of Bristol)} \author{Wil O'Mullane (Vera C. Rubin Observatory)} \author{Martin Reinecke (Max Planck)} From 4f0c3c101ba10227105ee178d04c3331dd35ad86 Mon Sep 17 00:00:00 2001 From: fxpineau Date: Tue, 2 Jun 2026 13:15:24 +0200 Subject: [PATCH 05/10] Add text to allow for 32-bits integer RANGE packaging --- moc.tex | 20 ++++++++++++-------- 1 file changed, 12 insertions(+), 8 deletions(-) diff --git a/moc.tex b/moc.tex index 94ddfed..032e9af 100644 --- a/moc.tex +++ b/moc.tex @@ -790,14 +790,18 @@ \subsubsection{Binary serialization} orders in a 64-bit signed integer (TFORM1='K'). \paragraph{RANGE packaging} -For the coding of RANGE alternative packaging, all the indices are -expressed at the maximum resolution, and it is the succession -of intervals that will be stored in the FITS table as -two 64-bit signed integers (TFORM1='K') : -the smallest index of the interval and the index strictly greater -than the largest value of the interval. The RANGE values {\bf must} -be in ascending numerical order. The resulting list is stored in a -single-column binary table extension. +For the coding of the RANGE alternative packaging, all indices {\bf shall} +be expressed at the maximum resolution. +The sequence of intervals {\bf shall} be stored in the FITS table as pairs +of signed integers. +For orders strictly less than 14, 32-bit signed integers (TFORM1='J') +{\bf may} be used. In this case, the maximum resolution {\bf shall} be 13. +Otherwise, 64-bit signed integers (TFORM1='K') {\bf must} be used, +with a maximum resolution of 29. +The two integers {\bf must} represent the smallest index value of the interval +(inclusive) and largest value of the interval (exclusive), respectively. +The RANGE values {\bf must} be stored in ascending numerical order. +The resulting list {\bf shall} be stored in a single-column binary table extension. \paragraph{Remark on RANGE vs UNIQs packing} From b614905c11d3598f3dbdfc64c71ac59504bfa285 Mon Sep 17 00:00:00 2001 From: fxpineau Date: Tue, 2 Jun 2026 16:24:56 +0200 Subject: [PATCH 06/10] Put 'designated' back to 'designed' in sec 3 --- moc.tex | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/moc.tex b/moc.tex index 032e9af..6ed7df9 100644 --- a/moc.tex +++ b/moc.tex @@ -1082,7 +1082,7 @@ \subsection{Compromise of Volume VS. Resolution} the spatial order and/or on the temporal (or frequency) order independently. \subsection{Working resolution} -The MOC has been designated to be able to efficiently handle observation +The MOC has been designed to be able to efficiently handle observation coverages (images, catalogs, ...). During the construction of the MOC, we must then ensure that at the chosen nominal resolution, any cell of the MOC contains at least one observation (no empty cell). To keep From 1f95deb65b3d8c93e39cc118156359461d18c223 Mon Sep 17 00:00:00 2001 From: fxpineau Date: Tue, 2 Jun 2026 23:54:34 +0200 Subject: [PATCH 07/10] Implement M. Taylor comments: fix Time equations, update RANGE serialization (more to come), ... --- appendix_json.tex | 2 +- moc.tex | 29 +++++++++++++++-------------- 2 files changed, 16 insertions(+), 15 deletions(-) diff --git a/appendix_json.tex b/appendix_json.tex index e912b7b..f312e55 100644 --- a/appendix_json.tex +++ b/appendix_json.tex @@ -14,7 +14,7 @@ \section{JSON encoding} Although we expect index arrays to be sorted, we cannot impose the same constraint on orders, since JSON object members are considered unordered. -\paragraph{Example of a JSON SMOC or TMOC:} +\paragraph{Example of a JSON SMOC:} \par\noindent \begin{lstlisting}[] {"1":[1,2,4], "2":[12,13,14,21,23,25], "8":[]} diff --git a/moc.tex b/moc.tex index 6ed7df9..e1ab6e1 100644 --- a/moc.tex +++ b/moc.tex @@ -505,14 +505,15 @@ \subsubsection{Time MOC conventions} Hence, an integer time value $t$ in $\mu$s since JD=0 is valid if $t \in [0, 2^{62}[$. For a given $order$, its cell $index$ is: \begin{equation} - index = \frac{t}{2^{order}} + index = \frac{t}{2^{61-order}} \label{eq:tmoc:i} \end{equation} +% = 2^{order_max - order} (useful with a different datatype in which order max is different) Conversly, the range of time values $[t_{min}, t_{max}[$, in $\mu$s since JD=0, of a given $index$ of order $order$ is given by \begin{align} - t_{min} &= index \times 2^{order} \\ - t_{max} &= (index + 1) \times 2^{order} + t_{min} &= index \times 2^{61-order} \\ + t_{max} &= (index + 1) \times 2^{61-order} \label{eq:tmoc:t} \end{align}. @@ -756,8 +757,6 @@ \subsection{MOC serializations for a single quantity} A MOC can be manipulated and serialized either as a list of cell numbers for each order (hierarchical view), or as a list of intervals at the deepest order (range view). -While these two methods are used for -SMOCs, only the intervals approach is used for other MOCs. The various possible MOC serializations are presented below. \subsubsection{Binary serialization} @@ -766,20 +765,22 @@ \subsubsection{Binary serialization} defined: either all MOC pairs (order, index) are stored individually thanks to NUNIQ packaging, or all ranges of indices at the deepest order are stored following the RANGE packaging. +While these two methods are used for +SMOCs, only the intervals approach is used for other MOCs. \paragraph{NUNIQ Packaging (valid for SMOC only)} The NUNIQ scheme defines an algorithm for packaging a MOC pair (order, index) into a single integer for compactness: -$$ +\begin{equation} \textit{uniq} = 4 \cdot (4\,^\textit{order}) + \textit{index} -$$ +\end{equation} \par\noindent The inverse operation is: -\begin{eqnarray*} +\begin{eqnarray} \textit{order} & = &\log_2(\textit{uniq}/4)/2\\ \textit{index} & = &\textit{uniq} - 4 \cdot (4\,^\textit{order}) -\end{eqnarray*} +\end{eqnarray} \par\noindent The list of cells {\bf must} be well-formed (see Section~\ref{sec:can}) allowing @@ -791,13 +792,13 @@ \subsubsection{Binary serialization} \paragraph{RANGE packaging} For the coding of the RANGE alternative packaging, all indices {\bf shall} -be expressed at the maximum resolution. +be expressed at the deepest order: 29 for SMOCs, 61 for TMOCs and 59 for FMOCs. The sequence of intervals {\bf shall} be stored in the FITS table as pairs of signed integers. -For orders strictly less than 14, 32-bit signed integers (TFORM1='J') -{\bf may} be used. In this case, the maximum resolution {\bf shall} be 13. -Otherwise, 64-bit signed integers (TFORM1='K') {\bf must} be used, -with a maximum resolution of 29. +For orders strictly less than 14 (SMOC), 30 (TMOC) or 28 (FMOC), +32-bit signed integers (TFORM1='J') {\bf may} be used. +In such a case, all indices {\bf shall} be expressed at order 13 (SMOC), 29 (TMOC) or 27 (FMOC). +Otherwise, 64-bit signed integers (TFORM1='K') {\bf must} be used. The two integers {\bf must} represent the smallest index value of the interval (inclusive) and largest value of the interval (exclusive), respectively. The RANGE values {\bf must} be stored in ascending numerical order. From 1067b74fa826086eb62b6105424ba2232516707c Mon Sep 17 00:00:00 2001 From: fxpineau Date: Tue, 4 Aug 2026 14:35:17 +0200 Subject: [PATCH 08/10] Fix human readable time values in table2 --- table2.tex | 20 ++++++++++---------- 1 file changed, 10 insertions(+), 10 deletions(-) diff --git a/table2.tex b/table2.tex index a05b1c6..937670b 100644 --- a/table2.tex +++ b/table2.tex @@ -7,16 +7,16 @@ \hline Order & Time Cell Resolution ($\mu$s) & Order & Time Cell Resolution ($\mu$s) \\ \hline -0 & 2305843009213693952 ($\simeq$73117.8y) & 31 & 1073741824 ($\simeq$17.9m) \\ -1 & 1152921504606846976 ($\simeq$36558.9y) & 32 & 536870912 ($\simeq$9m) \\ -2 & 576460752303423488 ($\simeq$18279.4y) & 33 & 268435456 ($\simeq$4.5m) \\ -3 & 288230376151711744 ($\simeq$9139.7y) & 34 & 134217728 ($\simeq$2.2m) \\ -4 & 144115188075855872 ($\simeq$4569.9y) & 35 & 67108864 ($\simeq$1.1m) \\ -5 & 72057594037927936 ($\simeq$2284.9y) & 36 & 33554432 ($\simeq$33s) \\ -6 & 36028797018963968 ($\simeq$1142.5y) & 37 & 16777216 ($\simeq$16s) \\ -7 & 18014398509481984 ($\simeq$571.2y) & 38 & 8388608 ($\simeq$8s) \\ -8 & 9007199254740992 ($\simeq$285.6y) & 39 & 4194304 ($\simeq$4s) \\ -9 & 4503599627370496 ($\simeq$142.8y) & 40 & 2097152 ($\simeq$2s) \\ +0 & 2305843009213693952 ($\simeq$73067.8y) & 31 & 1073741824 ($\simeq$17.9m) \\ +1 & 1152921504606846976 ($\simeq$36533.9y) & 32 & 536870912 ($\simeq$9m) \\ +2 & 576460752303423488 ($\simeq$18266.9y) & 33 & 268435456 ($\simeq$4.5m) \\ +3 & 288230376151711744 ($\simeq$9133.5y) & 34 & 134217728 ($\simeq$2.2m) \\ +4 & 144115188075855872 ($\simeq$4566.7y) & 35 & 67108864 ($\simeq$1.1m) \\ +5 & 72057594037927936 ($\simeq$2283.4y) & 36 & 33554432 ($\simeq$33s) \\ +6 & 36028797018963968 ($\simeq$1141.7y) & 37 & 16777216 ($\simeq$16s) \\ +7 & 18014398509481984 ($\simeq$570.8y) & 38 & 8388608 ($\simeq$8s) \\ +8 & 9007199254740992 ($\simeq$285.4y) & 39 & 4194304 ($\simeq$4s) \\ +9 & 4503599627370496 ($\simeq$142.7y) & 40 & 2097152 ($\simeq$2s) \\ 10 & 2251799813685248 ($\simeq$71.4y) & 41 & 1048576 ($\simeq$1s) \\ 11 & 1125899906842624 ($\simeq$35.7y) & 42 & 524288 ($\simeq$524ms) \\ 12 & 562949953421312 ($\simeq$17.8y) & 43 & 262144 \\ From 220d4501cdcd91abb277d0ab7c4474d9729d7d40 Mon Sep 17 00:00:00 2001 From: fxpineau Date: Thu, 29 Oct 2026 15:32:19 +0100 Subject: [PATCH 09/10] Mainly add RANGEORD (M. Taylor) and very slighlty relax COORDYS to reflect usages (P. Fernique) --- moc.tex | 22 ++++++++++++++++------ table2.tex | 2 +- table3.tex | 5 +++-- 3 files changed, 20 insertions(+), 9 deletions(-) diff --git a/moc.tex b/moc.tex index e1ab6e1..93e4a15 100644 --- a/moc.tex +++ b/moc.tex @@ -799,6 +799,11 @@ \subsubsection{Binary serialization} 32-bit signed integers (TFORM1='J') {\bf may} be used. In such a case, all indices {\bf shall} be expressed at order 13 (SMOC), 29 (TMOC) or 27 (FMOC). Otherwise, 64-bit signed integers (TFORM1='K') {\bf must} be used. +The FITS keyword /RANGEORD {\bf must} contain the order in which the ranges are expressed. +The value of the order is defined by the combination of the quantity and TFORM values, +as listed above. +Although the RANGEORD keyword is mandatory, reading tools {\bf should} accept +files in which this keyword is missing and to issue a warning. The two integers {\bf must} represent the smallest index value of the interval (inclusive) and largest value of the interval (exclusive), respectively. The RANGE values {\bf must} be stored in ascending numerical order. @@ -820,6 +825,7 @@ \subsubsection{Binary serialization} use the first method for SMOC for compatibility with existing libraries not yet compatible with MOC 2.0. + \subsubsection{ASCII serialization} To encode a MOC as a string each MOC pair (order, index) {\bf must} be written sequentially in an ASCII stream as two ASCII numbers separated @@ -887,8 +893,8 @@ \subsubsection{ASCII serialization} \end{lstlisting} Note that we use Extended BNF supporting regular expression syntax with the following rules: i) postfix * means "repeated -0 or more times”; ii) postfix + means "repeated 1 or more -times”; iii) postfix ? means "0 or 1 times". The first six +0 or more times"; ii) postfix + means "repeated 1 or more +times"; iii) postfix ? means "0 or 1 times". The first six rules depend on the MOC type, i.e. SMOC for space, TMOC for time, FMOC for Frequency, STMOC for space-time ,SFMOC for space-frequency and FTMOC for frequency-time. @@ -928,7 +934,7 @@ \subsubsection{Binary Serialization} This list of ranges will be coded in a list of 64bits integers (time indices with the 64th bit forced to 1) as: \begin{lstlisting}[] - tmin1 tmax1 smin1 smax1 tmin2 tmax2 tmin3 tmax3 smin2 smax2 smin3 smax3 +tmin1 tmax1 smin1 smax1 tmin2 tmax2 tmin3 tmax3 smin2 smax2 smin3 smax3 \end{lstlisting} STMOC encoding must conform to the following simple rules: @@ -979,6 +985,7 @@ \section{FITS keywords} backwards compatibility if required. Other FITS Keywords could be used to augment the information like DATES and ORIGIN. + \input{table3.tex} \clearpage @@ -1034,9 +1041,11 @@ \subsubsection{SMOC and ICRS} This choice reflects common astronomical practice, as the majority of astronomical catalogs are provided in equatorial coordinates. -However, it is possible to use SMOC for other coverages, -such as planetary coverages. The definition of the unique reference for each -body will have to be defined. +However, some astronomical data, such as planetary data, do not pertain to the sky. +Others, such as some legacy HiPS, were created using Galactic coordinates. +The use of \emph{COORDSYS} other than \emph{C} could be considered for these specific cases. +However, this is outside the scope of the current standard, and support for such values +is not required in current implementations. \subsubsection{TMOC and TCB \label{sec:tcb}} @@ -1154,6 +1163,7 @@ \subsection{Changes between versions 2.0 and 2.1} The differences between version 2.1 of MOC and the preceding version 2.0 are: \begin{itemize} \item Introduction of the frequency dimension to define FMOCs, SFMOCs and FTMOCs; + \item Introduction of the 32-bit version of the RANGE packaging serialization, with the RANGEORD keyword; \item Introduction of the SUNIQ and ZUNIQ numbering in appendix. \end{itemize} Adding a new quantity (frequency) as an extension to the existing standard, diff --git a/table2.tex b/table2.tex index 937670b..496f5fe 100644 --- a/table2.tex +++ b/table2.tex @@ -41,6 +41,6 @@ \hline \end{tabular} } -\caption[TMOC order and cell resolutions]{TMOC cell resolutions for each orders.}\label{table:tmocorders} + \caption[TMOC order and cell resolutions]{TMOC cell resolutions for each orders. Julian years (365.25 days of 84\,000 seconds each) are used in this table.}\label{table:tmocorders} \end{center} \end{table} diff --git a/table3.tex b/table3.tex index 7d310dc..90e2f07 100644 --- a/table3.tex +++ b/table3.tex @@ -1,7 +1,7 @@ \begin{table}[!htbp] \smallskip \begin{center} - {\small + {\footnotesize \begin{tabular} {|l | p{6cm}| c| c|} \hline {\bf Keyword} & {\bf Definition} & {\bf MOC1.1} & {\bf MOC2.0} \\ @@ -19,10 +19,11 @@ MOCORD\_F & Best resolution of the frequency dimension, expressed as the order. & NA & [mandatory] \\ MOCORDER & Best resolution of the space dimension, expressed as the order. & mandatory & NA \\ PIXTYPE & 'HEALPIX' & mandatory & NA \\ + RANGEORD & ORDER of the stored ranges. The permitted values are defined by the standard and depend on both MOCDIM and TFORM. & NA & [mandatory] \\ \hline \end{tabular} } \end{center} -\caption[FITS Keywords for MOC]{FITS Keywords for MOC. Symbol '[]' means mandatory if the MOC contains the keyword related quantity.} +\caption[FITS Keywords for MOC]{FITS Keywords for MOC. Symbol '[]' means mandatory if the MOC contains the keyword related quantity, or if ORDERING=RANGE for RANGEORD.} \label{table:fits_stmoc} \end{table} From 98b1f7800b78a90335dbb55701a5f72889025eed Mon Sep 17 00:00:00 2001 From: fxpineau Date: Thu, 29 Oct 2026 16:56:03 +0100 Subject: [PATCH 10/10] Replace the 2 'algin' env by 'eqnarray' --- moc.tex | 16 ++++++++-------- 1 file changed, 8 insertions(+), 8 deletions(-) diff --git a/moc.tex b/moc.tex index 93e4a15..4b385d6 100644 --- a/moc.tex +++ b/moc.tex @@ -511,11 +511,11 @@ \subsubsection{Time MOC conventions} % = 2^{order_max - order} (useful with a different datatype in which order max is different) Conversly, the range of time values $[t_{min}, t_{max}[$, in $\mu$s since JD=0, of a given $index$ of order $order$ is given by -\begin{align} - t_{min} &= index \times 2^{61-order} \\ - t_{max} &= (index + 1) \times 2^{61-order} +\begin{eqnarray*} + t_{min} &= &index \times 2^{61-order} \\ + t_{max} &=& (index + 1) \times 2^{61-order} \label{eq:tmoc:t} -\end{align}. +\end{eqnarray*}. We note that two consecutive cells at order N with the indices (M$\times$2)+1, (M$\times$2) will be coded at order N-1 with the index @@ -613,12 +613,12 @@ \subsubsection{Frequency MOC conventions\label{sec:fmoc1}} $order_{max} = 51$. At the maximum resolution, the index corresponding to a frequency $f$ is therefore computed as: -\begin{align} - order_{max} &= 51 \\ - index_{51} &= \left\lfloor +\begin{eqnarray*} + order_{max} &=& 51 \\ + index_{51} &=& \left\lfloor \frac{\log_{10}(f) + 18}{56} \times 2^{52} \right\rfloor -\end{align} +\end{eqnarray*} Conversely, the frequency range $[f_{min}, f_{max}[$ corresponding to a given index $index$ at a given order $order$ can be obtained using the inverse