Researchers have been able to accelerate fusion energy simulations by up to 50 times while preserving critical physical detail.

A team at the Helmholtz-Zentrum Dresden-Rossendorf (HZDR), a major German research facility, have developed a new method to speed up complex simulations of how matter behaves under extreme conditions, reducing the computational demands on supercomputers.

For instance, fusion energy is being researched around the world as a potential source of almost limitless, clean energy. However, before it is anywhere near deployable, researchers need to fully understand what temperatures and pressures are present in such extreme states of matter.

To understand what happens, researchers use X-ray scattering. Essentially, a high-intensity X-ray beam penetrates the sample and the resulting scattering signal is studied to learn about its properties. However, the resulting data is often not enough on its own to determine key characteristics such as density and temperature.

So, in order to interpret their experiments, researchers have to calculate numerous combinations of temperature and density, which requires a lot of computing time on supercomputers. Analysing the data also produces heavy numerical noise (small, often imperceptible errors that occur during simulations), making the process incredibly slow and resource-intensive.

HZDR’s new method enables scientists to precisely interpret measured data from such X-ray scattering experiments and run related simulations far more quickly. 

Instead of researchers constantly refining the simulations, the basic idea is to systematically identify which elements of the signal calculated are physically relevant – and which are just numerical noise. So instead of simply smoothing out the data, which often obscures important details, it preserves the signal’s physical structure.

“Building on this, we combine a reliable convergence test with a filtering procedure that removes artificial ringing without distorting the physical information,” said Dr Zhandos Moldabekov, a researcher at HZDR who came up with the idea for the method.

During testing, their method was able to run simulations 50 times faster, according to Moldabekov. This means that instead of only being able to do a few simulations on expensive supercomputers, it will be possible to run more simulations, as well as analyse experimental data faster and more accurately. 

“If we want to have a fusion power plant, we have to understand what really happens in such extreme states of matter. Now, our new method makes it possible to comprehensively and precisely analyse the datasets from such experiments,” said Dr Tobias Dornheim, a physicist and head of the high energy density department in HZDR’s Institute of Radiation Physics.

“It should be possible to develop our method into a standard tool for interpreting modern X-ray experiments. In the future, it could play a central role in exploring extreme states of matter,” added Moldabekov.