Minotaur Toolkit

Minotaur is an open-source software for solving Mixed-Integer Nonlinear Optimization (MINLO) problems. It includes solvers for both convex and nonconvex problems. It can be extended or customized for specific problem types.

Features

Native support for nonlinear functions including automatic differentiation
Multithreaded solvers (OA and QG)
Uses BQPD, CBC, Clp, CPLEX, FilterSQP, IPOPT for LP, QP and MILP subproblems
Compatible with AMPL and Pyomo

Solvers Available

mbnb: nonlinear optimization based branch-and-bound for convex MINLPs
mqg: LP/NLP based branch-and-cut for convex MINLP
mglob: Global optimization solver for quadratically constrained problems.
mmultistart: Multistart heuristic for nonconvex MINLP

Also available when compiled with CPLEX:

moa: Outer Approximation for convex MINLP
mlstoa: Outer Approximation for convex MINLP with lazy callbacks