SolutionStrategyMultiTree
Iterative outer approximation strategy (multi-tree)
Iterative outer approximation strategy (multi-tree)
Classic outer approximation loop for convex MINLP.
Algorithm Pattern:
- Solve MIP relaxation to get candidate point
- Generate supporting hyperplanes at violated points
- Add cuts to MIP and resolve
- Repeat until convergence or termination
Task Flow (initializeStrategy):
- CreateDualProblem → SolveIteration → SelectHyperplanes
- AddHyperplanes → CheckTermination → loop
Advantages:
- No callback complexity
- Can use any MIP solver
- Easier debugging/logging
Disadvantages:
- Multiple MIP solves
- May regenerate same B&B tree work
Algorithm
Standard ESH outer approximation
See Also
- SolutionStrategySingleTree for callback-based variant
Source
Header file: `src/SolutionStrategy/SolutionStrategyMultiTree.h`