Multistability study results: gene network vs voltage-gated channels - #6
Merged
Merged
Conversation
… study Add mneme.analysis.steady_state for the question the project is meant to answer: does a system have more than one stable state? - assess_steady_state decides whether a relaxing run has settled. A low rate of change is not accepted on its own: the remaining distance is estimated from the fitted relaxation time. - count_distinct_states groups end states against a stated threshold and reports how clean the grouping is. Tested on exponential relaxations with known time constants and on a double-well system with two known stable states. studies/convergence holds the configuration generator, runner and analysis for a BETSE experiment that starts one tissue from six ion concentrations, with and without gap-junction coupling. Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
Thirteen BETSE runs of one 42-cell tissue, from six starting ion concentrations, with and without gap-junction coupling, plus an exact repeat. Runs that started up to 48.8 mV apart ended within 0.32 mV of each other: one stable state, no multistability. The runs did not meet the settling criterion fixed beforehand (estimated remaining drift 0.20 to 0.32 mV against a limit of 0.1 mV). The results say so. Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
Audit of every document not touched by the correctness work: the 11-module course, the demo notebook, the setup, structure, pipeline, API-design and testing guides, CONTRIBUTING, the API reference pages, the default config and the setup script. - Removed or renamed APIs updated (reconstructor names, method names, removed Lyapunov functions, CLI behaviour). - Claims of validated results removed; experimental components marked as such with pointers to docs/SCOPE.md; Lyapunov material points to the operating-range document and states the 39-surrogate and 4,000-point requirements. - Stale numbers (test count, coverage, Python version, install commands) corrected. Every mkdocstrings directive verified to import. - Historical documents get a one-line note and are otherwise untouched. - New API page for mneme.analysis.steady_state. - studies/multistability: config generator for the two multistability candidates (gene network, voltage-gated channels). Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01PPeuxmAc47j5SKMb65CMRT
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01PPeuxmAc47j5SKMb65CMRT
Gene network coupled to voltage (2018 patterns configuration, half-size world, tissue cut disabled): two distinct stable voltage patterns under one parameter set. Pattern A from three different gradient starts (identical to 3e-4 mV), pattern B from two near-uniform starts (identical to 0.01 mV); the two differ by 31.9 mV; the perfectly uniform state is an unstable saddle. All runs met the settling criterion. Kir2.1 plus a Na+ leak (2016 configuration): one state at each of three leak strengths; runs did not meet the settling criterion; the strongest leak from the polarised start was numerically unstable. Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01PPeuxmAc47j5SKMb65CMRT
README gets a 'Withdrawn results' section naming both withdrawn results, when and why; the withdrawal notice returns to the site navigation under 'Record'; the docs index warning links to it. Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01PPeuxmAc47j5SKMb65CMRT
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Summary
Results of the two multistability experiments. Stacked on #5; retarget to
mainonce #5 is merged.Findings
GRN. Three different gradient starts converged to one pattern (identical to 3e-4 mV). Two near-uniform starts (1% and 0.1% gradients) left the uniform state within 600 s and converged to a different pattern (identical to 0.01 mV). The two patterns differ by 31.9 mV in the worst cell and depolarise different regions of the tissue. The perfectly uniform state is an unstable saddle. Replicate noise 0 mV.
VGC. The Na⁺ leak depolarised the tissue in proportion to its strength; the two starts ended within 0.1 mV of each other at each leak. No second state. Runs missed the pre-set settling criterion (0.2 to 1.4 mV left), and the strongest leak from the polarised start was numerically unstable.
Limits stated in the write-up
README.mdgives the steps to regenerate it.Files
studies/multistability/RESULTS.md,results_grn.json,results_vgc.jsonstudies/multistability/make_configs.pygains the perturbation modeproject_plan.md,CLAUDE.md,CHANGELOG.mdstatus lines🤖 Generated with Claude Code
https://claude.ai/code/session_01PPeuxmAc47j5SKMb65CMRT