MathOptSolversCMake

This library includes:

• CMake wrappers for mathematical optimization solvers
• A mathematical programming modeler that supports:
  • Coninous and integer variables
  • Linear structures
  • Quadratic structures
  • Nonlinear structures
  • Black-box functions

The goal of the modeler are:

• Minimize the modeler's overhead
• Run multiple solvers while writing the model's code once and ensuring that the model passed to each solver is the same
• Keep access to all the direct API features of the solvers
• Minimize the quantity of code to integrate a new solver
• Provide some features to help model debugging

They are not designed to be as user-friendly as possible. And switching solver requires a bit more lines of code than changing a string.

Supported solvers:

• HiGHS (MILP) https://highs.dev/
• Cbc (MILP) https://github.com/coin-or/Cbc
• FICO Xpress (MILP) https://www.fico.com/en/products/fico-xpress-optimization
• Artelys Knitro (all) https://www.artelys.com/solvers/knitro/
• Dlib (box-constrained) https://dlib.net/
• ConicBundle (box-constrained) https://www-user.tu-chemnitz.de/~helmberg/ConicBundle/

Examples:

• MILP:
  • Multiple-choice knapsack
  • Set covering
  • Generalized assignment
  • Clique, stable
  • Knapsack with conflicts
  • Graph coloring
  • Shop scheduling, positional model, disjunctive model
• Box-constrained (Lagrangian relaxations):
  • Generalized assignment
  • Knapsack with conflicts

CMake integration example:

# Fetch fontanf/mathoptsolverscmake.
set(MATHOPTSOLVERSCMAKE_USE_CLP ON)
FetchContent_Declare(
    mathoptsolverscmake
    GIT_REPOSITORY https://github.com/fontanf/mathoptsolverscmake.git
    GIT_TAG ...)
    #SOURCE_DIR "${PROJECT_SOURCE_DIR}/../mathoptsolverscmake/")
FetchContent_MakeAvailable(mathoptsolverscmake)

...

target_link_libraries(MyProject_my_target PUBLIC
    MathOptSolversCMake::clp)