Manifold-mapping optimization applied to linear actuator design

Optimization procedures, in practice, are based on highly accurate models that typically have an excessive computational cost. By exploiting auxiliary models that are less accurate, but much cheaper to compute, space mapping (SM) has been reported to accelerate such procedures. However, the SM solution does not always coincide with the accurate model optimum. We introduce manifold mapping, an improved version of SM that finds this precise solution with the same computational efficiency. By two examples in linear actuator design, we show that our technique delivers a significant speedup compared to other optimization schemes.