A Frank-Wolfe-based primal heuristic for quadratic mixed-integer optimization
Gioni Mexi, Deborah Hendrych, Sébastien Designolle, Mathieu Besançon, Sebastian Pokutta · 2025
Read A Frank-Wolfe-based primal heuristic for quadratic mixed-integer optimization by Gioni Mexi, Deborah Hendrych, Sébastien Designolle, Mathieu Besançon, Sebastian Pokutta (2025) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2508-01299.txt.
Description
Gioni Mexi and colleagues built a fast heuristic for hard quadratic optimization problems in a 2025 paper. Quadratic mixed-integer problems, with both discrete choices and curved objectives or constraints, model everything from portfolio design to process engineering, and exact solving is slow. Their heuristic extends the Boscia framework, a solver using Frank–Wolfe iterations inside branch-and-bound for convex problems, into the nonconvex quadratic world, reformulating nonlinear constraints and adding preprocessing.
A suite of tricks exploits the integer-feasible points the Frank–Wolfe iterations generate along the way: rounding strategies, gradient-guided variable selection, and large neighborhood search. On benchmark QPLIB instances the method improved results within minutes, and it took first place in the Land–Doig MIP Computational Competition 2025, per the authors. Good feasible answers, fast, are often worth more than proofs of optimality.
Catalog record
| CREATOR | Gioni Mexi, Deborah Hendrych, Sébastien Designolle, Mathieu Besançon, Sebastian Pokutta |
|---|---|
| YEAR | 2025 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 38 KB |
| VERSION | 1.0 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2508-01299