cyhighs¶

cyhighs is a Python package that provides Cython bindings to HiGHS, the high performance solver for linear programming and mixed integer programming. It compiles HiGHS from source and links it statically, so an installed wheel carries the solver with it and needs no external system libraries.

The package exposes three interfaces at increasing levels of convenience. You can hand the solver raw arrays, you can pass SciPy sparse matrices, or you can call a drop in replacement for scipy.optimize.linprog. All three run the same underlying solver, so you can pick the one that fits your code and move between them as your needs change.

Here is the shortest way to solve a problem. It maximizes x + 2y subject to x + y <= 4 and x + 3y <= 6, with both variables between 0 and 10.

from cyhighs import linprog

result = linprog(
    c=[-1.0, -2.0],
    A_ub=[[1.0, 1.0], [1.0, 3.0]],
    b_ub=[4.0, 6.0],
    bounds=(0, 10),
)

assert result.success
print(result.x, result.fun)

The objective is negated because linprog minimizes, which is the SciPy convention. The solver returns x = [3, 1] with an objective of -5, so the maximum of x + 2y is 5.

Where to go next¶

• What is cyhighs explains the problem the package solves and how its three interfaces relate.
• How HiGHS is bundled describes the build and why wheels need no extra dependencies.
• Installation covers installing from a wheel and building from source.
• The API Reference documents every public function and enumeration.