CTSolvers.jl

CTSolvers.jl is the resolution layer of the control-toolbox ecosystem. It provides the strategies — NLP modelers, NLP solvers, ODE integrators — and the abstract problem types that upstream packages use to solve optimal control problems.

Who is this documentation for?

This is a reference for developers and experienced users of the ecosystem. To model and solve optimal control problems, start with OptimalControl.jl, which builds on CTSolvers and exposes its strategies through solve.

CTSolvers does not solve optimal control problems by itself. Two packages build on it:

• CTDirect.jl — direct methods: discretizes continuous-time OCPs (defined in CTModels.jl) into finite-dimensional NLPs, then uses CTSolvers' Modelers and Solvers to solve them.

• CTFlows.jl — flows for indirect methods: builds Hamiltonian flows from ODE systems and integrates them with CTSolvers' Integrators.

Qualified Module Access

CTSolvers does not export functions directly. All functions and types are accessed via qualified module paths:

using CTSolvers
CTBase.Strategies.id(CTSolvers.Solvers.Ipopt)   # ✓ Qualified
CTSolvers.Optimization.build_model(prob, x0, m)  # ✓ Qualified

Module overview

Module	Responsibility
Optimization	Abstract problem type (AbstractOptimizationProblem), BuiltModel, NoCache, generic build_model/build_solution
Modelers	Modelers.ADNLP, Modelers.Exa — NLP backend selection
DOCP	AbstractDiscretizer, DiscretizedModel — an OCP paired with its discretizer; discretize
Solvers	Solvers.Ipopt, Solvers.MadNLP, Solvers.MadNCL, Solvers.Knitro, Solvers.Uno — NLP solvers
Integrators	Integrators.SciML — ODE integrator, and integration results

How this documentation is organized

• Getting Started — installation, extension loading, options.

• Architecture — modules, type hierarchies, data flow, and design patterns.

• Using CTSolvers — for experienced users:

  • Strategies Catalogue — every solver, modeler and integrator, with its package and devices

  • Integration Results — reading results, saveat, regrid, multi-phase merge

  • GPU Execution — the GPU parameter and its requirements

• Developer Guides — implementing new components:

  • Implementing a Solver — tag dispatch, extension pattern, CommonSolve integration

  • Implementing an Integrator — SciML wrapper, integration result types

  • Implementing a Modeler — strategy options, build_model/build_solution dispatch

  • Implementing an Optimization Problem — AbstractOptimizationProblem contract, DiscretizedModel

  • Performance & Type Stability — hot path vs setup path, JET and allocation checks

  • Error Messages Reference — all exception types with examples and fixes

• API Reference — auto-generated documentation for all public and private symbols.

Ask DeepWiki

offers an interactive, AI-generated overview of this codebase. Answers may be inaccurate — use this reference documentation as the source of truth.