Illustration of the CMS detector at the Large Hadron Collider, detecting the results of a proton-proton collision. Reproduced with permission from CERN.
CMS Experiment Using Bridges Supercomputer to Simulate Collisions for Upgraded LHC Particle Accelerator
Oct. 10, 2018
The Pittsburgh Supercomputing Center (PSC) is now supplying computation for the world’s most powerful particle collider. Fermi National Accelerator Laboratory (Fermilab) scientists working on the Compact Muon Solenoid (CMS) experiment, in collaboration with the Open Science Grid, have begun analyzing data from the Large Hadron Collider (LHC) using PSC’s Bridges supercomputer. The work is driven by a vast upgrade to the LHC’s computational capabilities that scientists expect to need by 2026. In that year, the latest enhancement of the collider—the “High-Luminosity LHC,” or HL-LHC—comes online. The HL-LHC will increase the intensity of the collider, located at the CERN laboratory on the border between France and Switzerland, to 10 times that of its original design.
“What’s going to happen then is the luminosity of the LHC will increase by such a factor that the collision event data collected—and the event complexity—will expand our computing demands by an order of magnitude,” said Dirk Hufnagel of Fermilab’s Scientific Computing Division. The new collider is expected to generate on the order of an exabyte of data. That’s about 250 million 4-gigabyte movies.
PSC is a joint effort of Carnegie Mellon University’s Mellon College of Science and the University of Pittsburgh.
“We’re looking at how we can include HPC into our computing model,” Hufnagel added. CMS is working now with Bridges because they expect that successfully incorporating supercomputing (HPC, or high performance computing) into their existing procedures and data flow will require some experimentation in order to be ready to go by 2026. But it’s not just experimentation: “We’re gaining experience at this point, but we do get useful work out of it as well.” At the moment, the group is using Bridges to help refine its simulations of the signals generated by the possible particles produced by the upgraded collider. The idea is to eventually help scientists more quickly identify these particles in the real HL-LHC data.
“Flexibility to support Fermilab’s CMS workflow sets Bridges apart from traditional supercomputers,” said Paola Buitrago, Bridges co-principal investigator and PSC Director of Artificial Intelligence and Big Data. “This is an important milestone that we expect to help advance the frontiers of physics.”
Data from the CMS and ATLAS detectors helped generate headlines in 2012—and a 2013 Nobel Prize—by identifying the long-sought Higgs boson, which gives mass to elementary particles. Scientists don’t know what the more powerful LHC will discover. At the very least, it will provide an opportunity to measure the properties of the Higgs boson in detail and thus test the standard model of particle physics to great precision. But this model is believed to be incomplete, and the HL-LHC will provide an opportunity for the discovery of new particles and new interactions.
“No matter what, we are looking forward to using Bridges to explore these new horizons in particle physics,” Hufnagel said.