An Evidence-Driven Solver Router: Regret 0.046 vs 0.190 for the Best Single Backend
Rules Decided on a Training Split, Checked Out of Sample, Every Answer Verified; on the GPU 93/166/11 vs the CPU.
0.046mean regret vs the oracle, 276 instances
214/276oracle hits
93/166/11router on cuda vs cpu
0planted wrong answers accepted
- Router regret 0.046 (train 0.049, test 0.042) vs 0.190 best single per kind
- Rules HOLD out of sample on 14 of 15
- GPU vs CPU routing: 93/166/11
More garmo solvers benchmark results
- Proof-Carrying Max-Cut on a GH200: 67 Biq Mac Optima Proved vs HiGHS's 49
- Weighted MaxSAT on a GH200: From 14/33/50 to 33/38/26 Against NuWLS-c-IBR
- Large-Instance TSP on 16 Arm Cores: 0.692 % → 0.165 % Mean Gap, Still Behind LKH-3
- A Native QCDCL Solver vs DepQBF: From 0 vs 4 to a 15–15 Tie, Then 15 vs 18 on Fresh Instances
- Vehicle Routing vs PyVRP: a C++ Iterated Local Search Wins 37/26/21 at Equal Budget
- ISPD98 Bisection: a Multilevel FM + Flow Engine Beats KaHyPar and mt-KaHyPar at Equal Budget
- A Verified MIP Race: HiGHS, SCIP and GPU Primal Heuristics Against cuOpt on MIPLIB 2017
- Lazy Clause Generation in C++: 39 of 40 Schedules Proved Optimal vs CP-SAT's 20
- Certified GPU Numerics: Every Answer Carries a Checkable Certificate: and Where It Costs
- LABS on a GH200: Exponent 1.24 to the Proven Optimum, Beside the QAOA Exponents
- UNISOLVE 0.16.1: One Solver Platform, 13 Problem Domains, Every Answer Checked
- Four Certified Builds: Toroidal Proofs, Unroutability Certificates, CertBQP and a 235× Batch Service