New presolve operations for integer variables and additional Mixed integer Rounding cuts are now available. These new cuts are automatically derived from nonlinear constraints. This feature significantly increases the number of problem classes for which cuts can be generated.
Those new cuts have been tested on a subset of instances derived from CMU-IBM library. As illustrated above, the additional cuts derived from nonlinear constraints (only these cuts are presented in the chart) significantly reduce the root node integrality gap.
This version also includes a completely updated MATLAB interface integrating dedicated solve functions for many optimization problems. These dedicated functions exploit as much as possible the problem structure, leading to significant performance improvements, especially for LP/QP/QCQP problems.
Additional features of Artelys Knitro 12.1:
- The Python interface now supports the SciPy format to solve NLP instances.
- Improved performance on MPEC problems thanks to new presolve operations.
- Exporting of linear and quadratic instances in MPS file format is now available.
- General performance on large-scale NLP and SOCP instances.
— The European Commission has published the METIS models and datasets developed by Artelys and its partners, enabling everyone to use them to conduct highly detailed analyses of the European energy system.read more
— We are pleased to announce the release of Artelys Knitro 12.2 which comes with outstanding performance improvements, in particular for general nonlinear problems.read more
subscribe to our newsletters