Finding 1
Defined benchmark comparison
The Forgetting Engine was compared with a stated vehicle-routing baseline under a fixed experimental configuration.
Benchmark Case Study
Vehicle Routing
A computational test of strategic candidate elimination in a constrained routing problem. The results are evidence about the tested benchmark, not a claim of universal superiority in real fleet operations.
Finding 2
Scale-dependent results
The reported relative gap varied by problem size, with the largest stated improvement reaching 89.3% at the largest tested scale.
Finding 3
Feasibility and objective both matter
Route quality should be interpreted alongside constraint satisfaction, search budget, instance generation, and baseline tuning.
What is supported
Under the documented test conditions, the Forgetting Engine produced better routing-objective results than the selected baseline, with the relative difference changing across tested scales.
What remains open
Production value would require stronger and tuned baselines, standard public datasets, sensitivity analysis, implementation cost measurement, and independent replication on operational fleet constraints.
Read the result inside its benchmark.
The Evidence page places the routing result alongside the other optimization studies and explains why cross-domain percentages are not directly interchangeable.
Review the Optimization Evidence