Version 12 expands the scope of optimization solvers in the Wolfram Language to include optimization of convex functions over convex constraints. Convex optimization is a class of problems for which there are fast and robust optimization algorithms, both in theory and in practice. Just as advances in linear optimization opened up many industrial applications, ever-wider classes of problems are being identified to be convex in a wide variety of domains, such as statistics, finance, signal processing, geometry and many more.
- New set of functions for classes of convex optimization.
- Enhanced support for linear optimization. »
- Support for linear fractional optimization. »
- Support for quadratic optimization. »
- Support for second-order cone optimization. »
- Support for semidefinite optimization. »
- Support for conic optimization. »
- Support for primal and dual solution properties.
- Support for matrix- or formula-based modeling input.
- Vector inequalities for modeling with vector-valued variables.
- Automatic dimensional inference for vector-valued variables.
- Automatic detection of convex problems in existing general optimization functions.
- Automatic use of convex optimization for specific tasks.