Optimisation design of fast simulation models

The nearest future is very exciting for High Energy Physics, especially for the Large Hadron Collider experiments and the high-luminosity upgrade foreseen for 2029 and then say this will increase the demand for simulated data. As a response to resource challenges, which concern all full simulation methods, we propose a highly scalable framework for fast simulation methods. In this report, we present the development of the MLFastSim framework with the following ideas: semi-automatised design of a deep learning model (VAE) available in a distributed environment and integration of the system with an external tool for tracking training progress. Moreover, we also present in this report several additional improvements which increase the usability of the framework, and we identified the areas for the direction of further research and areas of the project’s development.