[论文] TetrisCNN for interpretable detection of phases of matter from experim...

论文概要

研究领域: ML
作者: Kacper Cybiński, Björn van Zwol, James Enouen, Guillaume Bornet, Thierry Lahaye, Antoine Browaeys, Antoine Georges, Anna Dawid
发布时间: 2026-09-17
arXiv: 2609.20693

中文摘要

物相检测的核心在于找到正确的序参量——对未知相变而言这项任务 notoriously 困难，传统上依赖物理直觉与有根据的猜测。神经网络近期提供了一条替代路径：无需任何先验物理知识即可在已知模型中定位相变。但这些方法仍是黑箱：只识别相而不阐明其性质；且在面对构成自动化方法终极考场的真实噪声实验数据时往往力不从心。本文弥合两个视角：提出 TetrisCNN——一个由不同形状滤波器（ reminiscent 俄罗斯方块）的并行分支构成的卷积架构，它直接以自旋关联子学习稀疏、可解释的潜表示。应用于以多基矢测量的二维 Ising 与 XY 量子模拟器实验快照，该网络不仅能检测相变与 crossover，还能将其潜表示与决策边界表达为由实验可测自旋关联子构成的符号公式。该框架为可解释神经网络与量子模拟器结合、以发现和理解新物相开辟了道路。

原文摘要

Detecting phases of matter in general relies on identifying the correct order parameter - a task that remains notoriously difficult for unknown transitions and traditionally is guided by physical intuition and educated guess. Neural networks have recently offered an alternative route by locating phase transitions in known models without any a priori physical knowledge. Yet these approaches remain black boxes and only identify phases without elucidating their properties. Moreover, they often struggle when confronted with realistic, noisy experimental data, which constitute the ultimate testbed for automated methods in physics. Here, we bridge these perspectives by introducing TetrisCNN, a convolutional architecture with parallel branches of differently shaped filters, reminiscent of Tetris ...

自动采集于 2026-09-20

#论文 #arXiv #ML #小凯