A domestic research team has developed an algorithm that utilizes technology to determine if bullets hit targets in games to detect collisions of high-speed particles rushing into the interior walls of a nuclear fusion reactor. It is expected to improve the stability and design efficiency of the nuclear fusion reactor by quickly predicting collisions even in complex three-dimensional structures.
Professor Yoon Ui-sung of the Ulsan National Institute of Science and Technology (UNIST) noted on the 17th that his research team developed an algorithm capable of quickly identifying collision points of high-speed particles within a virtual nuclear fusion device. The research results were published in the April issue of the international journal Computer Physics Communications.
Nuclear fusion power, known as the artificial sun, injects high-speed neutral particles to heat the interior of the fusion reactor to temperatures similar to the sun. If some particles go out of control and collide with the device's interior walls, it could damage the reactor walls or halt the nuclear fusion reaction.
The research team applied a collision detection algorithm used in the gaming industry to the problem of detecting colliding particles. The existing method divided the space into predetermined sections and checked each one for particles, but this new algorithm excludes over 99.9% of the calculations through simple arithmetic, only computing when necessary.
Applying this algorithm to 'V-KSTAR' increased the collision detection speed by up to 15 times compared to before. V-KSTAR is a digital twin that replicates the Korean nuclear fusion experimental reactor KSTAR in a three-dimensional virtual space.
Professor Yoon Ui-sung stated, "The collision algorithm is one of the core technology elements for the three-dimensional expansion of V-KSTAR, as well as tracking the neutral particle beam," adding, "We are planning additional research using GPU supercomputers, which have faster processing speeds than CPU computers, for high-speed calculations."
References
Computer Physics Communications (2025), DOI: https://doi.org/10.1016/j.cpc.2024.109490