Monte Carlo simulation is vital for the design, construction and operation of high-energy physics (HEP) experiments. The computing resources needed to generate the required amount of simulated data are increasing with the energy and the luminosity of particle accelerators and are starting to exceed the available computing budget. Faster simulation is therefore essential to maintain the accuracy of physics analyses. Speed-up of the simulation may be achieved by reducing the time spent in detailed simulation, by applying a parametrisation of the physics processes, or by directly reproducing the signal created in the detectors. With Machine Learning (ML)-based technologies, new possibilities have emerged where trained neural networks take the place of the most compute-intensive stages of HEP event simulation. They are of special interest for the simulation of calorimeters, where most of the simulation time is spent for HEP experiments. There are several on-going activities carried out in different groups.

In particular, the following activities are foreseen:

• Review and evaluation of existing generative models, focusing on their application to the simulation of particle showers.
• Testing, training, and validation of generalizable and fast adaptive generative models based on meta learning approaches for particle shower simulation.
• Testing and development of optimized data pipelines: evaluation of data loading and training strategies for big datasets (such as distributed training) for efficient hardware usage (for eg. to prevent GPU from data starvation).
• Review and evaluation of inference libraries (LWTNN, ONNX, Tensorflow lite, MNN,.. ).
• Integration of ML-aided simulation models within full simulation.
• Testing and evaluation of existing and new AutoML algorithms (e.g. evolutionary algorithms).
• Model profiling and memory footprint optimization strategies (graph optimizations, quantization, knowledge distillation,...).

Contact and Collaboration

ML4Sim is public forum where machine learning for simulation discussions can take place with a focus on technical issues and novel algorithms. Please sign up to the mailing list to keep in touch.

For more information and collaboration opportunities for Faster Simulation, please contact Anna Zaborowska and Dalila Salamani.