logo Persalys
fr en

Persalys 13.1 release


The new version of Persalys 13.1 is now available. Two new features have been added to this version :

  • Polynomial regression: this is the classic linear regression in a canonical basis of degree 1 or 2 with or without interactions. The results provide information on the residuals, the confidence intervals on the estimated coefficients and of course the regression formula. The validation techniques already present for the other metamodels also apply for this case.
  • Multi-objective optimization: the pagmo library is used to solve this type of problem. You can now set multiple outputs to be minimized or maximized. The result obtained is a Pareto front, which is provided if the number of objectives is 2. The table with all the values of the different fronts is provided with the classic parallel coordinates and scatter diagram tabs.
    We thank Naval Group for their contribution regarding the integration of this feature.

image pareto front

    Overall optimization has been improved:
  • Input parameters can be defined as continuous variables, integers or binaries.
  • Outputs can be added as constraints of the optimization problem.

image optim input choice

    Other changes include:
  • The ability to test an independent copula during correlation inference. This makes it possible to automatically test whether it is preferable not to define a dependency between the variables of a sample.
  • A sliding bar has been added in the results of the Morris method to be able to more simply modify the vertical bar for selecting influential variables.

image morris result