Description

The PAVER reports used in the paper: “A Convexification-based Outer-Approximation Method for Convex and Nonconvex MINLP” by Z. Peng, K. Cao, K.C. Furman, C. Li, I.E. Grossmann, D.E. Bernal are available here.

Implementation

The implementation of the Outer-Approximation method is based on MindtPy, the Mixed-Integer Nonlinear Decomposition Toolbox in Pyomo. Furthermore, we integrate two implementations of the bound tightening and the convexification techniques for MINLP problems in MindtPy.

  1. The first implementation is based on a special version of BARON 19.4.4, the state-of-the-art commercial MINLP solver.
  2. The second implementation uses Coramin and the FBBT (C++) code in Pyomo, both of which are open-source and offer more flexibility.

MINLP Instances

All the MINLP instances benchmarked here are from MINLPLib, including

  1. 434 convex MINLP instances
  2. 181 nonconvex MINLP instances

Paver Reports

  1. C-OA method for convex instances

  2. C-LP/NLP-based B&B method for convex instances

  3. C-GOA method for nonconvex instances

  4. C-GLP/NLP-based B&B method for nonconvex instances