Project Grant 2411204

This $300,000 Project Grant awarded by the National Science Foundation's Computer and Information Science and Engineering (CISE) program (CFDA 47.070) to the Massachusetts Institute of Technology (MIT) will support the development of advanced machine learning (ML) algorithms and data compression techniques to address the 200 TB/s data rate challenge for the LHCb Upgrade II experiment at CERN's Large Hadron Collider.

Specifically, the project will extend the functionality of the HLS4ML software package to enable the deployment of ML models on the Field Programmable Gate Arrays (FPGAs) used in the LHCb experiment's electronics. It will also develop low-latency, ML-based lossy compression algorithms to significantly reduce the data rate before it reaches the computing infrastructure. Additionally, the project will build on prior work to implement reconstruction algorithms that can run efficiently on both FPGA and GPU architectures, providing flexibility for the LHCb collaboration to choose the most cost-effective solution. These advancements are critical to ensuring the LHCb Upgrade II experiment can effectively manage its unprecedented data rates and maintain its ability to explore new physics beyond the Standard Model.