Knitro

Knitro is a commercial optimization software library designed for solving nonlinear programming (NLP) and mixed-integer nonlinear programming (MINLP) problems. It is built to handle large-scale, real-world optimization tasks with an emphasis on robustness, speed, and scalable performance. Knitro provides solvers for continuous NLP as well as MINLP, including features for exact Hessian or sparse derivative evaluation, warm starts, and callback-based user-defined function handling.

History and overview: Knitro originated with Ziena Optimization and, after organizational changes in the optimization software ecosystem, has been developed and maintained by Artelys. It runs on major operating systems including Windows, Linux, and macOS, and is distributed under a proprietary license.

Algorithms and capabilities: The NLP solver in Knitro uses interior-point methods complemented by other robust strategies, including active-set options and a filter-based convergence framework to improve reliability on difficult problems. For MINLP, Knitro incorporates a branch-and-bound framework that leverages the NLP subproblem solver to explore discrete decision variables, enabling effective handling of mixed-integer nonlinear problems. The software is optimized for large, sparse problems and supports advanced features such as multi-threading and sensitivity analysis.

Interfaces and modeling support: Knitro offers APIs for several programming languages, including C, C++, Fortran, Java, Python, MATLAB, R, and Julia. It also provides interfaces to modeling environments, notably AMPL, enabling users to formulate problems in a high-level modeling language while solving with Knitro’s engines.

Applications and use cases: It is employed across engineering, energy, finance, and research domains for tasks such as design optimization, parameter estimation, and operations planning where nonlinear constraints and variables are present.

Licensing and availability: Knitro is a proprietary software product licensed through Artelys and its distribution partners.