Installation

Install

Open Julia's interactive session (REPL) and use the package manager:

using Pkg
Pkg.add("OptimalControl")

Tip

If you are new to Julia, follow this guideline.

OptimalControl alone is enough to define a problem with @def and describe solve strategies. Everything below is optional — loaded only when the feature it backs is actually used.

You will also need a solver

solve needs an NLP solver backend loaded. NLPModelsIpopt is the default and the one used throughout this documentation:

using NLPModelsIpopt

Alternatives exist, each behind its own package:

Solver	Load
:ipopt (default)	using NLPModelsIpopt
:madnlp	using MadNLP (CPU) or using MadNLPGPU (GPU)
:uno	using UnoSolver
:madncl	using MadNCL and using MadNLP (both)
:knitro	using NLPModelsKnitro (commercial licence required)

See Choosing a method for how these combine with a discretizer, a modeler and a :cpu/:gpu parameter. Calling solve before the matching package is loaded raises an ExtensionError naming exactly which using statement to add — the same mechanism covers every optional piece on this page.

Optional: plotting

using Plots

unlocks plot(sol). Without it:

julia> using OptimalControl
julia> plot(sol)
ERROR: ExtensionError: missing dependencies to plot solutions
Missing  Plots
Hint     Run: using Plots

See Plotting.

Optional: flows

Building a Flow — indirect shooting, simulation, or inspecting a Hamiltonian vector field — needs an ODE integrator:

using OrdinaryDiffEqTsit5

The most common first failure

Every page in the Flows section opens with using OrdinaryDiffEqTsit5 for this reason: building a Flow before it is loaded is the single most common trap for newcomers to this part of the package. It fails cleanly rather than silently:

julia> using OptimalControl
julia> φ = Flow(ocp, u)
ERROR: ExtensionError: missing dependencies to access SciML options metadata
Missing  OrdinaryDiffEqTsit5
Hint     Run: using OrdinaryDiffEqTsit5

Optional: saving solutions

using JLD2   # format=:JLD (default)
using JSON3  # format=:JSON

unlock export_ocp_solution/import_ocp_solution. Without the matching one:

julia> export_ocp_solution(sol; format=:JLD)
ERROR: ExtensionError: missing dependencies to export solutions to JLD2 format
Missing  JLD2
Hint     Run: using JLD2

See Save & load.

Optional: GPU

using ExaModels
using MadNLPGPU
using CUDA

NVIDIA GPUs only; the problem's dynamics must be written coordinatewise. Missing any of the three raises the same ExtensionError as above, naming whichever is missing first. See GPU for the constraints and check CUDA.functional() before assuming a :gpu solve will actually run on the device.

Checking your setup

Loading everything and calling methods() is a quick way to confirm the install is sound — if this runs without error, OptimalControl and every optional piece above are wired in:

using OptimalControl
using NLPModelsIpopt
using Plots
using OrdinaryDiffEqTsit5
using JLD2
using JSON3

methods()

(:collocation, :adnlp, :ipopt, :cpu)
(:collocation, :adnlp, :madnlp, :cpu)
(:collocation, :adnlp, :uno, :cpu)
(:collocation, :adnlp, :madncl, :cpu)
(:collocation, :adnlp, :knitro, :cpu)
(:collocation, :exa, :ipopt, :cpu)
(:collocation, :exa, :madnlp, :cpu)
(:collocation, :exa, :uno, :cpu)
(:collocation, :exa, :madncl, :cpu)
(:collocation, :exa, :knitro, :cpu)
(:collocation, :exa, :madnlp, :gpu)
(:collocation, :exa, :madncl, :gpu)

See also

• Your first problem — model, solve and plot one, end to end.

• Choosing a method — the full discretizer/modeler/solver/parameter picture.