v1.1-beta1
Pre-releaseWhat's new in version 1.1?
Major features
-
SHOT no longer needs an external NLP solver in its primal strategy since SHOT can now call itself for solving fixed NLP problems. Activated with
Primal.FixedInteger.Solver=2
. For nonconvex problems it is still recommended to use an external NLP solver. -
Partition convex nonseparable quadratic functions as separate constraints using an eigenvalue decomposition-based reformulation. Activated with
Model.Reformulation.Quadratics.UseEigenValueDecomposition=true
. -
Support for performing an initial polyhedral approximation of the nonlinear feasible set before feasibility-based bound tightening. Activated with
Model.BoundTightening.InitialPOA.Use=true
. -
Support for problems containing semi-continuous and semi-integer variables.
-
Support for problems containing special ordered sets.
Minor improvements and bug fixes
-
Improved support for generating supporting hyperplanes for the entire nonlinear feasible set instead of the feasible sets for the individual constraint functions. Activated with
Dual.ESH.Rootsearch.UseMaxFunction=true
. -
Improved support for passing nonconvex quadratic functions directly to the MIP solver (if supported).
-
Bug fixes to the AMPL interface.