Persalys 18 release

3/04/2025

The new version of Persalys 18 is ready. For this version, two main new features have been added to the data models: quantile estimation and the ability to create a 1-d field data model.
Quantile estimation in the data model involves determining quantile values for given probability levels. The user can choose to calculate a quantile in the low or high tail or bilaterally. Two methods are implemented:

Monte Carlo: this is the empirical quantile calculated from the sample, and the confidence interval is also provided.
Excess method: the generalised Pareto distribution is fitted to the tail of the distribution, enabling quantiles to be estimated for low probabilities. The aim is to select only the data located in the tail of the distribution and to adjust the distribution solely on this selection, called the excess sample. In order to define the threshold for creating this sample, the mean excess plot (or mean residual life plot) is provided to the user.

The data field model is used to load a file containing the field's 1D trajectories. The mesh can either be defined in Persalys, or also loaded from a file. The analysis that can then be performed is the same as for the physical field model, namely a central tendency analysis performing a Karhunen-Loève decomposition of the sample.