PAM-OPT 2020.5

ESI PAM-OPT 2020.5 has the following new main options:

Improvement of the Adaptive RSM algorithm:

• Extension of the design parameter bounds. The design parameter bounds move when a bound prevents PAM-OPT from improving the solution.
• Relaxation of the constraint function bounds. Reserved for optimization experts.

Improvement of curve treatments:

• New options, mainly for material fitting studies.

Handling of called applications:

• Improvement of the distribution of the processes on several machines. Thanks to a better handling of the variability of the elapsed time used by the called applications.

The extension of the design parameter bounds is an innovative and efficient optimization algorithm. As a matter of fact, Adaptive RSM (or any RSM based method) may not be effective on very wide bounds, that is because of the following main reasons:

• Bounds could be so wide that almost all calculations on these large bounds will generate non-exploitable or misleading results.
• At the beginning of an optimization, the optimizer could be trapped in a wrong remote local minimum.

On the one hand this new algorithm allows to avoid the latter problems, and on the other hand, it allows to handle unknown bounds.

This algorithm keeps the robustness and the reduced CPU of Adaptive RSM. It can be used in all types of optimizations with few or many parameters, continuous or discrete values.

Please refer to PAM-OPT Release Notes for more details about this release.