CMT/LTC Seminars

Symbolic regression for discovering physics equations

by Arjun Dey (PSI)

→ Europe/Zurich
WBBC/111

WBBC/111

https://teams.microsoft.com/l/meetup-join/19%3ameeting_YzVmOTgzMjctMTAwNC00YTY3LWIxOGQtNmNjODYxNjY5M2Rh%40thread.v2/0?context=%7b%22Tid%22%3a%2250f89ee2-f910-47c5-9913-a6ea08928f11%22%2c%22Oid%22%3a%2204a638bf-9dac-4ed8-ad21-43ac02aa2cc2%22%7d Meeting ID: 360 918 567 009, Passcode: qA6Cv93s
Description

In 1601, Kepler obtained the best planetary data and, after 4 years and 40 failed attempts, showed that Mars follows an elliptical orbit. This result is an early example of symbolic regression, and it revealed a simple law behind raw measurements. In this article, the authors present modern approaches to the same challenge, combining physics methods (dimensional analysis, smoothness, symmetry, separability, compositionality) with modern machine learning to discover equations from data. As a concrete test, the authors re-discover 100 equations from the Feynman Lectures on Physics.

Ref:

Udrescu, S. M., Tegmark, M. (2020). AI Feynman: A physics-inspired method for symbolic regression. Science advances, 6(16), eaay2631. https://doi.org/10.1126/sciadv.aay2631

Organised by

Laboratory for Theoretical and Computational Physics

Host: Dr. Markus Müller