Neural Renormalization Group Flow advances percolation theory with ML

Researchers have developed a novel machine learning approach called Neural Renormalization Group Flow to tackle complex problems in condensed matter physics, specifically focusing on two-dimensional site percolation. This method utilizes a supervised, scale-shared neural architecture that recursively applies learned coarse-graining rules across different scales. The model demonstrates an impressive ability to extrapolate from small lattice training data to significantly larger systems, accurately predicting crossing probabilities and reconstructing the largest cluster. Crucially, the learned latent representation exhibits critical fluctuations and scale-dependent flows that align with renormalization-group principles, enabling high-fidelity predictions and accurate finite-size scaling near critical points. AI

IMPACT This novel ML approach could enable data-driven solutions for complex physical systems where traditional methods struggle.

RANK_REASON The cluster contains a research paper detailing a new machine learning method for a physics problem. [lever_c_demoted from research: ic=1 ai=1.0]

Read on arXiv cs.LG →

• alphaXiv
• arXiv
• CatalyzeX Code Finder for Papers
• CORE Recommender
• Cyril Furtlehner
• DagsHub
• Gotit.pub
• Hugging Face
• IArxiv Recommender
• Influence Flower
• Neural Renormalization Group Flow for Percolation
• ScienceCast

• paper
• other

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