FastParam is a project that aims to enhance the detection of gravitational waves by optimising the analysis of pulsar signals through advanced Machine Learning (ML) techniques. The objective is to reduce the analysis time of current scientific models. The project focuses on analysing data from the European Pulsar Timing Array (EPTA).

The FastParam project has been selected and funded under the ICSC Programme, financed as part of the PNRR, funded by the European Union – NextGenerationEU.

The models used for analysing pulsar signals are complex due to the numerous parameters that must be considered to account for various astrophysical phenomena. This complexity leads to a vast parameter space, resulting in computational challenges and extended processing times.

Researchers are therefore forced to make simplifying assumptions about certain parameters, which can compromise the accuracy of the results.

Furthermore, Bayesian analyses present challenges such as the computational expense of calculating the posterior distribution and the impact of prior distribution choices.

To accelerate the running of parametric models for detecting gravitational waves in pulsar signals, FastParam explores two main strategies:

1. Intelligent sampling of the parameter space, inspired by ML techniques such as hyper-parameter optimisation. Methods like grid search, random search, and genetic algorithms will be explored.
2. Dimensionality reduction techniques, like PCA, LDA, LASSO, and Autoencoders will be considered.

These techniques are then combined together into an integrated pipeline for further analysis acceleration.