Thursday, August 15, 2019

Optimus - new metaheuristic optimization plug-in for Grasshopper


Optimus is a new metaheuristic optimization plug-in for Grasshopper. The first version implements the Self-Adaptive Differential Evolution with Ensemble of Mutation Strategies (jEDE). While the self-adaptive approach updates crossover and mutation rates, the ensemble of mutation strategies uses three operators to find the optimal result. The main purpose is to support the Performative Computational Architecture framework during the design decisions in the conceptual phase.

The plug-in is experimented using standard test problems, some of the test problems proposed in IEEE CEC 2005, and one design optimization problem (all these can be found below). Results show that Optimus (jEDE) outperformed some of the single-objective optimization algorithms available in Food4Rhino by finding much better solutions for 20 out of 21 problems. The target audience of the tool is frequent users of parametric design modeling, e.g., architects, engineers, and designers.

The background of the algorithm and the results can be found here.

No comments: