Using Machine Learning to Simulate the Physical World

Weather forecasting, fluid dynamics and drug discovery tools run heavy physics simulations. YC wants ML models that predict the same results in minutes on smaller computers.

Many essential software tools simulate the world from known principles of physics and chemistry: weather prediction, computational fluid dynamics for designing rockets and airplanes, and drug discovery tools that predict how molecules interact. Many run full physics simulations that solve complex equations and need heavy computing.

YC notes that AI models are general function approximators that can predict these outcomes without being told the physics. According to YC, the result is predictions in minutes or seconds on much smaller computers, instead of days or weeks on supercomputers.

The ask: companies replacing existing simulations with ML-based ones, and companies using ML-based simulation to open markets that can't be served today.

Read the full request on YC →