[论文解读] Automated High-Throughput Screening of Polymers Using a Computational Workflow

Lois Smith, Samuel Ericson+4 more|arXiv (Cornell University)|Mar 5, 2026

Machine Learning in Materials Science被引用 0

一句话总结

本文提出一种自动化工作流程，通过将自动退火与自适应控制相结合，实现聚合物的高通量计算筛选，为基于机器学习的聚合物性质预测铺平了道路。

ABSTRACT

High-throughput computational screening of polymers offers a powerful way to address the imbalance between the vast number of polymers synthesised for diverse applications and the relatively small subset that can be studied using atomistic simulations. This work presents an automatic workflow designed to enable the rapid and efficient screening of an extensive polymer library. The workflow integrates an automated annealing protocol with adaptive control, allowing for reproducible simulations with minimal human intervention and minimisation of the computational cost. The availability of a homogenous large set of simulations enables the adoption of machine learning approaches for a variety of tasks. We exemplify this possibility by proposing rapid machine-learning-based method to predict the (computed) polymer density and (experimental) glass transition temperature.

研究动机与目标

• Motivate high-throughput computational screening to bridge the gap between vast polymer synthesis and limited atomistic study.
• Develop an automatic workflow that minimizes human intervention and computational cost while maximizing reproducibility.
• Create a homogeneous large dataset of simulations to enable machine learning on polymer properties.

提出的方法

• Integrates an automated annealing protocol with adaptive control.
• Achieves reproducible simulations with minimal user intervention.
• Generates a large, uniform set of polymer simulations suitable for machine learning applications.
• Demonstrates rapid ML-based prediction of computed polymer density and experimental glass transition temperature.

实验结果

研究问题

• RQ1 Can an automated, adaptive annealing workflow provide reproducible high-throughput polymer simulations at reduced cost?
• RQ2 Is a homogeneous simulation dataset sufficient to enable reliable machine learning predictors for polymer properties (e.g., density, Tg)?
• RQ3 What is the feasibility of using ML to predict computed densities and experimental Tg from the generated polymer data?

主要发现

• An automatic workflow enables rapid and efficient screening of an extensive polymer library.
• Adaptive control and automated annealing minimize human intervention and computational cost.
• The homogeneous simulation set supports proposed machine-learning approaches for property prediction.
• The paper exemplifies fast ML-based methods to predict (computed) polymer density and (experimental) Tg.