Neptune method estimates complex PDE parameters from sparse data

Researchers have introduced Neptune, a novel method designed to infer parameter fields in complex multi-physics partial differential equations (PDEs) using scarce measurements. This approach utilizes independent coordinate neural networks to represent parameter fields, outperforming existing methods like physics-informed neural networks (PINNs) and neural operators. Neptune demonstrates robust parameter estimation with as few as 45 measurements, significantly reducing errors and showing superior physical extrapolation capabilities. AI

IMPACT Facilitates more data-efficient parameter inference for complex physical and biomedical models.

RANK_REASON Academic paper detailing a new method for parameter estimation in PDEs. [lever_c_demoted from research: ic=1 ai=0.7]

Read on arXiv cs.LG →

• alphaXiv
• arXiv
• CatalyzeX
• DagsHub
• Gotit.pub
• Hugging Face
• Neptune
• physics-informed neural networks
• ScienceCast
• Xuyang Liu

• paper
• other

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