New method robustly discovers governing equations from complex system data

Researchers have developed a method called PDE-SINDy to discover governing partial differential equations (PDEs) from spatiotemporal data. Their study demonstrates that the accuracy of this discovery process is highly dependent on the amount of data available, with more data leading to the suppression of spurious terms and a more robust identification of the correct equation. However, increasing the size of the function library used for discovery can reduce efficiency. For the Glauber spin-flip Ising model, the method revealed a hierarchy of equations, and a stringent selection threshold successfully recovered a Model-A-like dynamical equation that accurately captures phase separation and domain growth dynamics. AI

IMPACT This research advances methods for scientific discovery by enabling the robust derivation of complex equations from data, potentially accelerating research across various scientific domains.

RANK_REASON Academic paper detailing a new method for scientific discovery. [lever_c_demoted from research: ic=1 ai=1.0]

Read on arXiv cs.LG →

• arXiv
• cs.LG
• Glauber spin-flip Ising model
• Model A
• Partha Sarathi Mondal
• partial differential equations
• PDE-SINDy

• paper
• other

AI-generated summary · Google Gemini · from 1 sources. How we write summaries →