Simulation in the 21st century

This CMS simulated event was created using Geant4 simulation software. Image: CMS Collaboration

Simulation is not magic, but it can certainly produce the feeling. Although it can’t miraculously replace particle physics experiments, revealing new physics phenomena at the touch of a key, it can help scientists to design detectors for best physics at the minimum cost in time and money.

Geant4 is a detector simulation software toolkit originally created at CERN and currently developed by about 100 physicists and computer scientists from all around the world to model the passage of particles through matter and electromagnetic fields. For example, physicists use simulation to optimize detectors and software algorithms with the goal to measure, with utmost efficiency, marks that previously unobserved particles predicted by new theories would leave in their experimental devices.

Particle physics detectors are typically large and complex. Think of them as a set of hundreds of different shapes and materials. Particles coming from accelerator beams or high-energy collisions traverse the detectors, lose energy and transform themselves into showers of more particles as they interact with the detector material. The marks they leave behind are read by detector electronics and reconstructed by software into the original incident particles with their associated energies and trajectories.

We wouldn’t even dream of starting detector construction, much less asking for the funding to do it, without simulating the detector geometry and magnetic fields, as well as the physics of the interactions of particles with detector material, in exquisite detail. One of the goals of simulation is to demonstrate that the proposed detector would do the job.

Geant4 includes tools to represent the detector geometry by assembling elements of different shapes, sizes and material, as well as the mathematical expressions to propagate particles and calculate the details of the electromagnetic and nuclear interactions of particles with matter.

Geant4 is the current incarnation of Geant (Geometry and Tracking, or “giant” in French). It has become extremely popular for physics, medical and space science applications and is the tool of choice for high-energy physics, including CERN’s LHC experiments and Fermilab’s neutrino and muon programs.

The Fermilab Scientific Computing Simulation Department (SCS) has grown a team of Geant4 experts that participate actively in its core development and maintenance, offering detector simulation support to experiments and projects within Fermilab’s scientific program. The focus of our team is on improving physics and testing tools, as well as time and memory performance. The SCS team also spearheads an exciting R&D program to re-engineer the toolkit to run on modern computer architectures.

New-generation machines containing chips called coprocessors, or graphics processing units such as those used in game consoles or smart phones, may be used to speed execution times significantly. Software engineers do this by exploiting the benefits of the novel circuit design of the chips, as well as by using parallel programming. For example, a program execution mode called “multi-threading” would allow us to simulate particles from showers of different physics collisions simultaneously by submitting these threads to the hundreds or thousands of processor cores contained within these novel computer systems.

As the high-energy community builds, commissions and runs the experiments of the first half of the 21st century, a world of exciting and promising possibilities is opening in the field of simulation and detector modeling. Our Fermilab SCS team is at the forefront of this effort.

V. Daniel Elvira, Scientific Computing Simulation Department head