Parallel DOE, Optimization, Monte-Carlo

parallel DOE

optimization

Monte-Carlo

• Same principle as for Batch Runs: several simulation started with different parameters (data files)

• Monte Carlo / DOE runs are totally independent

• Optimization runs are independent during gradient computation (N consecutive runs, with N = number of degrees of freedom)

• Genetic algorithm runs are independent within a given generation (population size = number of individuals)

Remark: It actually looks like Data Parallelism.