研究人员开发了卷积循环神经网络代理模型来模拟晶体生长动力学。这些模型基于 Allen-Cahn 动力学数据进行训练，并能考虑可变过饱和度水平。该研究比较了两种架构：一种从帧的迷你序列中隐式推断过饱和度，另一种将过饱和度作为显式输入。结果表明，显式参数条件能够产生最准确的预测，尽管隐式方法在更大的训练数据集下也能获得可比的结果。 AI
影响 引入了模拟复杂物理过程的新型神经网络架构，有望加速材料科学研究。
排序理由 这是一篇详细介绍使用神经网络模拟晶体生长动力学的新方法的学术论文。
在 Hugging Face Daily Papers 阅读 →
AI 生成摘要 · Google Gemini · 来自 2 个来源。 我们如何撰写摘要 →
完整方法见我们的编辑标准。
Simulations of crystal growth are performed by using Convolutional Recurrent Neural Network surrogate models, trained on a dataset of time sequences computed by numerical integration of Allen-Cahn dynamics including faceting via kinetic anisotropy. Two network architectures are d…
Simulations of crystal growth are performed by using Convolutional Recurrent Neural Network surrogate models, trained on a dataset of time sequences computed by numerical integration of Allen-Cahn dynamics including faceting via kinetic anisotropy. Two network architectures are d…