Knitro

The NEOS Server offers Knitro for the solution of:

• general (convex or nonconvex) nonlinear problems (NLP),
• linear problems (LP),
• convex or nonconvex quadratic problems (QP/QCQP/SOCP),
• linear or nonlinear least-squares problems,
• mathematical programs with complementarity constraints (MPEC),
• mixed-integer nonlinear problems (MIP/MINLP), and
• derivative-free optimization problems.

Knitro natively offers several other interfaces in a variety of languages: (i) object-oriented interfaces for C++, C#, Java and Python; (ii) matrix-oriented interfaces for C, Fortran, Matlab, and R; (iii) links to modelling languages AIMS, AMPL, GAMS, and MPL; (iv) and links to Excel through Frontline Solvers.

Key features of Knitro include: (i) a large set of well-documented user options (see the Knitro documentation) and automatic tuner; (ii) (parallel) multi-start for global optimization; (iii) derivatives approximation and checker; and (iv) internal presolver.

Knitro was originally developed by Richard Byrd, Mary Beth Hribar, Jorge Nocedal and Richard Waltz with additional help from Guanghui Liu, Marcelo Marazzi, Todd Plantenga, and Jose Luis Morales. Since July 2015, Knitro has been developed by Artelys with Richard Waltz as lead developer.

Free trial and evaluations of Knitro are available on the Download page of the Artelys website. For more information on Knitro, see the Artelys Knitro Overview or directly contact Artelys.

Using the NEOS Server for Knitro/AMPL

The user must submit a model in AMPL format. Examples are provided in the examples section of the AMPL website.

The problem must be specified in a model file. A data file and commands files may also be provided. If the commands file is specified, it must contain the AMPL solve command; however, it must not contain the model or data commands. The model and data files are renamed internally by NEOS.

The commands file may include option settings for the solver. To specify solver options, add

  option knitro_options 'OPTIONS';

Knitro offers both interior-point and active-set methods. The user may specify the algorithm by setting the Knitro/AMPL option "alg" to one of the following values:

• 0: automatic algorithm selection (default)
• 1: Interior/Direct algorithm
• 2: Interior/Conjugate-gradient algorithm
• 3: Active-set algorithm
• 4: Sequential quadratic programming (SQP) algorithm
• 5: Run all algorithms