Persalys 19.1 release
05/12/25
Persalys version 19.1 has just been released. Improvments have been made to data sets, design of experiments and coupling models.
Data models have been renamed to data sets. It is now possible to run data analysis and marginal inference on data containing missing values. The marginal analyses are done only on the available data for each variable. The multivariate analyses are performed on the sub-sample containing only valid values.Sensitivity analysis is now available for data sets using the SRC indices and Sobol' indices computed using the RankSobolSensitivityAlgorithm.
A button has been added in the marginal inference wizard to make it easier to select laws to test. The underlying design of experiment of any simulation analysis can now be exported as a Persalys DOE. Evaluations can also be added to a DOE from a CSV file. This allows to reuse the same DOE for different analyses or to perform evaluations outside Persalys and then import the results. Monte Carlo is now the default choice in the DOE wizard. Coupling models can now be run on a remote server via SSH. The user has to provide the address of a server he can connect to without a password. The command will be executed on the remote server and input/output files will be transferred automatically.
