AI Paper Atlas

Research map / Learning theory and algorithms / Combinatorial and search optimization

FMIP: Joint Continuous-Integer Flow For Mixed-Integer Linear Programming

Hongpei Li, Hui Yuan, Han Zhang, Jianghao Lin, Dongdong Ge, Mengdi Wang, Yinyu Ye

ICLR 2026 · Learning theory and algorithms › Combinatorial and search optimization · approach: Learning branch-and-cut for MILP

Read the paperCheck your idea against 137,217 papers

AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.

Measured effects reported in the paper

What was changedWhereMeasuredEffect
model joint distribution of integer and continuous variables (add continuous generation) (ablation)FMIP · Item Placement, ND 400sprimal gap: 0.69 -> 0.25Improved
add feasibility guidance (ablation)FMIP · Item Placement, ND 400sprimal gap: 1.01 -> 0.25Improved
add objective guidance (ablation)FMIP · Item Placement, PMVB 600sprimal gap: 8.58 -> 0.67Improved
add holistic guidance (both terms) at inference (ablation)FMIP · Item Placement, ND 400sprimal gap: 0.84 -> 0.25Improved
add holistic guidance at inference (ablation)FMIP · Item Placement, PS 600sprimal gap: 0.24 -> 0.18Improved
use FMIP instead of best learning baselinedownstream solvers (ND, PS, PMVB, Apollo) · eight MILP benchmarksprimal gap: 41.34% average relative improvementImproved
use FMIP on simple benchmarks where baselines already reach optimum (CA, GIS, MIS, FCMNF with PS/PMVB/Apollo)downstream solvers · CA/GIS/MIS/FCMNFprimal gap: 0.00% relative improvement in many cellsNo clear effect
swap graph backbone (Bi-GCN, Tri-GCN, GAT, ClusterGCN) under FMIP (ablation)FMIP · IP and LB, objective valueobjective value: IP ND: 14.11, 13.99, 13.95, 14.06No clear effect
increase number of sampling steps beyond an optimum (ablation)FMIP · downstream solversprimal gapHurt
add adaptive guidance balancing at 32 steps (ablation)FMIP · Item Placement, PSobjective value: 13.25 -> 11.92Improved
add adaptive guidance balancing at 12 steps (ablation)FMIP · Item Placement, PSobjective value: 11.79 -> 11.93 (alpha=1.1)No clear effect
apply FMIP to very large MIPLIB instances (70,000 to 300,000 variables)FMIP · MIPLIB 2017 size subsetsrelative primal gap: 25.38% -> 23.26%; gap stays largeMixed

Later papers that compared with this method

1 later accepted paper reports FMIP as a baseline in their results tables; 1 of them report a better number on most of the results they share with it.

Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.

Closest papers by meaning

Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy