How an Astrophysicist Uses Codex to Help Simulate Black Holes

The gravity around black holes is so powerful that not even light can escape, and astrophysicists use computer simulations to study them. However, existing algorithms and computing power have had limits in fully reproducing the extreme physical phenomena around black holes.

In particular, modeling Plasma near black holes has been a major challenge. Particles in high-temperature, low-density plasma tend to spiral along magnetic field lines rather than collide with each other, and existing methods consumed excessive supercomputer computational resources trying to calculate all of this minute motion.

Chi-kwan Chan, a researcher at the University of Arizona, introduced Codex to solve this problem. Instead of tracking the particles' minute spiral motion one by one, he used AI to find a new algorithm that could mathematically bypass this process.

Codex helps generate candidate algorithms from among countless mathematical possibilities and supports the process of validating them. Chan directly reviews and physically verifies the numerical analysis techniques proposed by the AI, and through this he is achieving precise black hole simulations that were previously impossible.