An open Gaussian-process surrogate benchmark for multi-objective electric machine design optimization (P1: interior PM traction machine)

A reproducible benchmark for constrained multi-objective design optimization of electric machines. Problem P1 is a 16-variable, two-objective, two-constraint interior permanent-magnet traction-machine design task, evaluated through Gaussian-process surrogates trained on 2500 two-dimensional finite-element solutions. The package contains the numerical design-of-experiments table, the trained surrogates with a leave-one-out accuracy report, reference Pareto fronts, a finite-element validation of the surrogate ON the optimised front, the full comparison protocol, and the MATLAB code that reproduces every published number. It requires no commercial finite-element licence to run.