arXivDaily arXiv每日学术速递 周一至周五更新
arXiv周末暂无论文更新,休息一下吧,周末愉快~~
arXiv 2210.16623cond-mat.softcond-mat.dis-nn
收藏

通过物理启发的机器学习预测玻璃形成液体中的动态非均质性

Predicting dynamic heterogeneity in glass-forming liquids by physics-inspired machine learning

Gerhard Jung, Giulio Biroli, Ludovic Berthier

AI总结:

提出物理启发的机器学习框架 GlassMLP,利用结构信息预测深度过冷液体的长时间动力学与动态非均质性,并在二维、三维模型中展现高精度、低数据需求和跨尺寸迁移能力。

摘要:

我们介绍了 GlassMLP,这是一个采用物理启发的结构输入来预测深度过冷液体中长时间动力学的机器学习框架。我们将该深度神经网络应用于二维和三维原子模型。其性能优于现有最先进方法,同时在训练数据和拟合参数方面更为简约。GlassMLP 可定量预测四点动态关联以及动态非均质性的几何形态。跨系统尺寸的可迁移性使我们能够高效探究空间动态关联随温度的演化,揭示了重排区域几何形态随温度发生的深刻变化。

英文摘要:

We introduce GlassMLP, a machine learning framework using physics-inspired structural input to predict the long-time dynamics in deeply supercooled liquids. We apply this deep neural network to atomistic models in 2D and 3D. Its performance is better than the state of the art while being more parsimonious in terms of training data and fitting parameters. GlassMLP quantitatively predicts four-point dynamic correlations and the geometry of dynamic heterogeneity. Transferability across system sizes allows us to efficiently probe the temperature evolution of spatial dynamic correlations, revealing a profound change with temperature in the geometry of rearranging regions.

机构、总结、中文摘要均由AI分析而来,可能存在错误,请以论文正文为准。

↑