Using the OSG to test theories of nature at the LHC

Two cars collide on the motorway. It happens so fast that we cannot see details. From debris that is scattered around the scene we try to reconstruct interesting facts: Did the driver wear a seatbelt? Did the airbag work? What happened to the purse in the glove compartment?

Now imagine doing this for collisions that happen forty million times per second. We are talking about the Large Hadron Collider in Geneva. It smashes protons at nearly the speed of light. They behave much like a car and disintegrate. Physicists study the debris to reveal which forces were at work, and find the fundamental theory of nature.

–          Stefan Höche

~ Greg Moore

Making sense of the debris of forty million crashes per second is daunting. That is why Stefan Höche at SLAC National Accelerator Laboratory and his collaborators worldwide are developing more precise simulations for researchers at the Large Hadron Collider (LHC) and using the Open Science Grid (OSG) to run them. More precise simulations mean better analysis of LHC data.

Stefan Hoche at SLAC

Stefan Höche
Photo courtesy of Matt Beardsley/SLAC National Accelerator Laboratory


Höche and his colleagues have developed a simulation program called Sherpa that is capable of computing signals and backgrounds at the LHC at unprecedented accuracy. Sherpa’s simulations are based on quantum chromodynamics, the theory describing the strong nuclear force (the strongest basic force in nature, which holds together the subatomic particles of the nucleus).

A fundamental theory of nature will describe both what is observed in the universe and what is still unobserved – like dark matter and dark energy. In order to find the unobserved, physicists at the LHC record collision debris with detectors that work like giant digital cameras.


Figure 1: Representation of a proton-proton collision. Image courtesy of Frank Krauss/Durham University


To extract information from the ‘pictures,’ the researchers need to compare recordings with what their theory predicts. This is done only for the most interesting collisions, which happen about a thousand times per second.  Physicists use computer simulations that generate the same quanta of light that would occur in the detector during a collision. Then they compute how these quanta interact with the detector. They average the result over many pictures and compare the simulated average with the measured average. This comparison ultimately confirms or rejects their theory.

“Now, we need to make the theory prediction as precise as possible,” says Höche. “Getting precise theory predictions at the LHC is important, because we have lots of measurements and for most of them the experimental uncertainties are very small.” Höche says that his calculations using Sherpa gain them about a precision factor of two, as shown in Figure 2 below.


Figure 2: Transverse Momentum of a Reconstructed Top Quark. Image courtesy of Stefan Höche


“This may allow the experiments to discover a potential new theory of nature twice as fast and bring us closer, for example, to solving the mystery that surrounds dark matter,” says Höche. “These LHC calculations are running mostly on the OSG. Each prediction needs 250,000 CPU hours, and you have dozens of predictions to make. The OSG helps us to produce these results in as timely a manner as possible.”

Höche says the OSG has been an obvious choice for their work, but he thinks that researchers from many fields should use it. “Don’t be afraid of using it,” he says. “Some might be leery because they are so used to using their local resources. They also might need help in getting started. But the OSG is a great resource because you get a quicker turnaround. And when you don’t need the cycles, they are not going to waste. A site can simply contribute their small resource and, when you have a peak demand, use the outside resources that are available. The complexity of resource provisioning is entirely hidden from the user.”

In 2013, Höche, a member of SLAC’s Particle Physics Theory Group, was among 61 scientists – including three from SLAC – to receive Early Career Research Program grants from the U.S. Department of Energy (DOE). SLAC National Accelerator Laboratory (originally named Stanford Linear Accelerator Center) is one of 10 DOE Office of Science laboratories, and is operated by Stanford University on behalf of the DOE.

For more about SLAC, see their website

For more about Sherpa, see their website