OptimizationSNOPT.jl

OptimizationSNOPT.jl connects SNOPT, a commercial solver for large, sparse, constrained nonlinear optimization problems, to Optimization.jl. A problem defined as an OptimizationProblem is solved with SNOPT by passing SnoptOptimizer() to solve, and the derivatives come from any automatic differentiation backend that Optimization.jl supports.

With OptimizationSNOPT.jl, the same problem can be solved with SNOPT and with the other Optimization.jl solvers. The package is built on SNOPT.jl, which wraps the three SNOPT interfaces snOptA, snOptB and snOptC; use SNOPT.jl directly for linear constraints given separately, finite-difference derivatives estimated by SNOPT, or warm and hot starts.

Installation

OptimizationSNOPT.jl needs Julia 1.11 or later and a licensed SNOPT 7.7 library, libsnopt7, that includes the snopt-interface C functions. OptimizationSNOPT.jl does not include SNOPT; the library and the license are available from the SNOPT distributors.

using Pkg
Pkg.add("OptimizationSNOPT")

Before starting Julia, set SNOPTDIR to the directory that contains the library:

export SNOPTDIR=/path/to/snopt/lib

The installation guide gives the Windows commands and the setup check.

Example

Minimize the Rosenbrock function, with the gradient from ForwardDiff.jl:

julia> using OptimizationSNOPT

julia> rosenbrock(x, p) = (p[1] - x[1])^2 + p[2] * (x[2] - x[1]^2)^2;

julia> f = OptimizationFunction(rosenbrock, AutoForwardDiff());

julia> prob = OptimizationProblem(f, [0.0, 0.0], [1.0, 100.0]);

julia> sol = solve(prob, SnoptOptimizer());

julia> sol.retcode
ReturnCode.Success = 1

julia> round.(sol.u; digits = 6)
2-element Vector{Float64}:
 1.0
 1.0

For bounds and nonlinear constraints, pass lb, ub, lcons and ucons to OptimizationProblem and cons to OptimizationFunction, as the getting started tutorial shows.

Documentation

The documentation starts with a tutorial on bounds, nonlinear constraints and sparse Jacobians. It then gives how-to guides for solver options, monitoring and repeated solves, explains how SNOPT results map to Optimization.jl, and ends with the API reference.

Limits

• One Julia process runs one SNOPT solve at a time. Use several processes for parallel solves.
• The constraint Jacobian is dense unless the backend is sparse, for example AutoSparse(AutoForwardDiff()), or cons_jac_prototype gives its pattern.
• Integer variables are relaxed to continuous variables, with a warning.

Solver behaviour lists all limits.

License

OptimizationSNOPT.jl is released under the MIT License. SNOPT is a separate commercial product with its own license, and its binaries are not distributed with this package.