CVXPY

CVXPY is a Python-embedded modeling language for convex optimization. It lets users express convex optimization problems in a natural, readable syntax that resembles the mathematical formulation. It's designed to be independent of a specific solver, delegating solving to external backends.

The library centers on variables, parameters, expressions, constraints, and objectives. Problems are declared as Problem(objective, constraints). It supports Disciplined Convex Programming (DCP), a rule-based system that automatically verifies convexity of expressions and ensures that the problem is solvable as convex optimization.

CVXPY supports major convex problem classes including linear programs, quadratic programs, second-order cone programs, semidefinite programs, and exponential cone programs. It provides Parameter objects to represent data that can change between solves without reconstructing the entire problem, enabling parametric or scenario-based analyses.

For solving, CVXPY provides interfaces to multiple solvers such as ECOS, SCS, MOSEK, OSQP, GUROBI, GLPK, and CPLEX, among others. It is cross-platform, widely used in academia and industry for prototyping and research. It integrates with NumPy and SciPy for numerical data handling and supports several extensions and tooling.

CVXPY is open-source software released under the BSD 3-Clause license. It is maintained by a community of developers with founding contributions from Steven Diamond and Stephen Boyd, and is hosted on its project website and repository. It is used for education, research, and practical optimization tasks in finance, engineering, and data science.