| |
|
|
stoaMINLP
stoaMINLP
solves convex or nonconvex Mixed-Integer NonLinear Programming (MINLP) problems.
Main features
-
stoaMINLP
is using a Single-search Tree Outer Approximation algorithm
to solve MINLP problems.
-
It is best suited for solving convex problems, but nonconvex problems might be solvable.
-
Default is to use the global optimization solver multiMin (included in Tomlab /Base) to call the
NLP subsolver many times with different initial point to try to find the global minimum for nonconvex problems.
The number of NLP trials at the root node and in all other nodes are possible to change by the user.
-
If the nonlinear subproblems are known to be convex,
setting an input parameter will make the solver will run much faster.
In this case only one call is made to the NLP solver in each subproblem.
-
stoaMINLP
needs a good NLP solver as sub solver, e.g. Tomlab /SNOPT or Tomlab /KNITRO.
-
stoaMINLP
needs an LP solver as sub solver, e.g. Tomlab /CPLEX, MINOS in Tomlab /SOL or milpSolve in Tomlab /BASE.
-
stoaMINLP
is integrated with the TOMLAB driver routines.
-
stoaMINLP
may be used as subproblem solver in the TOMLAB
environment.
|
|
