AI speeds up laser simulations for X-ray lasers
Published
Researchers at Stanford University, UCLA and SLAC National Accelerator Laboratory have built an AI model that reproduces a key laser calculation far faster than standard methods, according to a ScienceDaily report of the university release. The team says the model gives results in milliseconds. The ScienceDaily headline puts the speed-up at 250 times.
The process is called second-order nonlinear optics. Light waves exchange energy while passing through special crystals, which changes their colour and the shape of the laser pulse. Predicting that behaviour normally takes a lot of computing time.
The work matters for SLAC's upgraded LCLS-II X-ray laser. There, infrared light is converted to green and then to ultraviolet before X-rays are made. According to the release, faster and more accurate simulations could improve the precision and efficiency of such lasers.
Why it matters: the gain is in computing time. Teams could test laser settings on a computer before they change a real instrument.
Would you trust an AI simulation to tune a research instrument, or only to check a human's settings?
Sources: ScienceDaily