Synergia pushes the state of the art

Jim Amundson

Jim Amundson, deputy head of the Computational Physics Department and leader of the Computational Physics for Accelerators Group, wrote this column.

In an era when the field of particle physics is looking to decide what accelerator projects to pursue, accelerator modeling expertise is of tremendous importance. Fermilab’s Synergia simulation tool is helping accelerator experts here and at CERN optimize their machines and plan for the future.

A little over 10 years ago, a small accelerator modeling team at Fermilab received its first grant from the then newly established Scientific Discovery through Advanced Computing (SciDAC) program. The thrust of this grant was to combine state-of-the-art space charge calculations with similarly advanced software for single-particle beam dynamics—a capability that did not exist in this field at that point. This work requires the sort of advanced high-performance computing (HPC) platforms championed by the SciDAC program. We named our new program Synergia (Συνεργια)—the Greek word for synergy.

Our first application of Synergia in 2002 was to improve the modeling and, ultimately, performance of the Fermilab Booster, when the delivery of protons for the Tevatron collider experiments and MiniBooNE were the lab’s highest priorities. After our initial successes modeling the Booster, we have continued to use SciDAC to enhance Synergia through both funding and collaborating with other physicists and computer scientists. In the past decade Synergia has evolved into a general framework for the calculation of intensity-dependent effects in beam dynamics. And while running Synergia efficiently on 128 parallel processors used to seem like a major accomplishment, we now have demonstrated efficient running on 131,072 cores, keeping us at the leading edge of the rapidly changing field of HPC.

The Synergia project continues today under the third incarnation of the SciDAC program as part of the Community Project for Accelerator Science and Simulation (ComPASS) collaboration, which is led by SCD’s Panagiotis Spentzouris. This year, I am the principal investigator of a sought-after INCITE award for computing time on Argonne’s state-of-the-art high-performance Blue Gene supercomputers. The award is for 80,000,000 core hours.

At Fermilab, the shift in emphasis to the Intensity Frontier is a natural fit for Synergia, since studying intensity-limiting effects is our bread and butter. We are currently applying Synergia to simulations of the Main Injector for Project X and the Debuncher for Mu2e while continuing studies of the Booster.

Synergia’s reputation as an important accelerator modeling framework is growing worldwide. CERN recently announced that it will use Synergia for simulations of space charge as part of the Lattice Injector Upgrades portion of the High-Luminosity LHC upgrades.

We are looking forward to continuing the success of this program by expanding Synergia’s capability to efficiently use the best new computing hardware in order to enable more and better accelerator science applications.