Quadratic Multiple Knapsack Problem on QMKP 10k
0.03Optimality Gap (%)Neural QP
Evaluation results
| Method | Links | |
|---|---|---|
| Neural QPTrain=Mini, Base solver=Gurobi2026.03 | 0.03 | |
| GNNQPTrain=Mini, Base solver=Gurobi2026.03 | 0.03 | |
| HNNTrain=Mini, Base solver=Gurobi2026.03 | 0.03 | |
| HNNTrain=2k, Base solver=Gurobi2026.03 | 0.03 | |
| Neural QPTrain=1k, Base solver=Gurobi2026.03 | 0.04 | |
| Neural QPTrain=2k, Base solver=Gurobi2026.03 | 0.04 | |
| TriGNNTrain=Mini, Base solver=Gurobi2026.03 | 0.04 | |
| TriGNNTrain=1k, Base solver=Gurobi2026.03 | 0.04 | |
| TriGNNTrain=2k, Base solver=Gurobi2026.03 | 0.04 | |
| HNNTrain=1k, Base solver=Gurobi2026.03 | 0.04 | |
| HNNTrain=1k, Base solver=SCIP2026.03 | 16.79 | |
| TriGNNTrain=Mini, Base solver=SCIP2026.03 | 17.25 | |
| Neural QPTrain=2k, Base solver=SCIP2026.03 | 17.41 | |
| HNNTrain=2k, Base solver=SCIP2026.03 | 17.45 | |
| Neural QPTrain=Mini, Base solver=SCIP2026.03 | 17.9 | |
| GNNQPTrain=Mini, Base solver=SCIP2026.03 | 17.9 | |
| HNNTrain=Mini, Base solver=SCIP2026.03 | 17.93 | |
| TriGNNTrain=2k, Base solver=SCIP2026.03 | 18.26 | |
| TriGNNTrain=1k, Base solver=SCIP2026.03 | 18.77 | |
| Neural QPTrain=1k, Base solver=SCIP2026.03 | 19.44 | |
| Exact solverBase solver=Gurobi2026.03 | 26.69 | |
| Exact solverBase solver=SCIP2026.03 | 27.61 | |
| GNNQPTrain=2k, Base solver=SCIP2026.03 | 31.12 | |
| GNNQPTrain=1k, Base solver=SCIP2026.03 | 38.09 | |
| GNNQPTrain=1k, Base solver=Gurobi2026.03 | 38.95 | |
| GNNQPTrain=2k, Base solver=Gurobi2026.03 | 39.05 |