Skip to documentation content
On this page
  1. Where to find it
  2. How to use it
  3. Settings and initial values
  4. Result
  5. Troubleshooting

CMA-ES

Explore a design with CMA-ES, a population-based search that does not require objective derivatives.

Explore a design with CMA-ES, a population-based search that does not require objective derivatives.

Where to find it

System Solver → Optimization

How to use it

Define an objective, select the design controls and their bounds, and add constraints before running the optimization study. Choose CMA-ES, provide sensible bounds and scales, and budget enough model evaluations for several generations.

Settings and initial values

SettingInitial value
Maximum concurrent trial solves8
Maximum number of outer iterations100
Optimality tolerance0.001
Constraint tolerance0.001
Step tolerance0.001
Population sizenull
Parent countnull
Random seed1729
CMA-ES covarianceDiagonal
CMA-ES covariance precisionAutomatic
Full covariance maximum controls4096
Full covariance maximum bytesnull
Total memory limit bytesnull
Resident memory estimate bytesnull
Candidate memory estimate bytesnull
Resident memory estimate scopenull
Candidate memory estimate scopenull
Retain all candidate payloadsfalse
Full covariance FP64 threshold bytes8589934592
Covariance rank-update block bytes67108864
Initial step size0.25
Minimum step size1e-07
Maximum step size0.8
Covariance learning ratenull
Constraint penalty1000.0
Penalty increase factor2.0
Maximum constraint penalty1000000000000.0
Parallel trial solves1
Cpu thread budgetnull
Native threads per candidate1
Maximum number of model evaluations1000
Maximum wall-clock timenull
Elapsed time offset seconds0.0
Objective stalled-iteration limit12

Result

Compare the final objective, constraint values, design and termination reason together.

Troubleshooting

Limits on iterations or evaluations can end a search before an optimum is reached. Inspect feasibility and termination reason, then verify the final design with a solve. Each candidate can require a complete solve. Compare multiple runs when randomness changes the outcome.