Polymer Sequence Design via Active Learning

TL;DR

This paper explores how active learning combined with physics-based and data science methods can efficiently optimize polymer sequences, addressing the challenge of navigating large chemical design spaces for material development.

Contribution

It introduces an integrated approach for polymer sequence design using active learning, assessing surrogate models and strategies for improved convergence and efficiency.

Findings

Active learning accelerates polymer sequence optimization.

Surrogate models significantly influence convergence.

Optimal initial conditions improve design efficiency.

Abstract

Analysis of molecular scale interactions and chemical structure offers an enormous opportunity to tune material properties for targeted applications. However, designing materials from molecular scale is a grand challenge owing to the practical limitations in exploring astronomically large design spaces using traditional experimental or computational methods. Advancements in data sciences and artificial intelligence have produced a host of tools and techniques that can facilitate the efficient exploration of large search spaces. In this work, a blended approach integrating physics-based methods and data science techniques is implemented in order to effectively screen the materials configuration space for accelerating materials design. Here, we survey and assess the efficacy of data-driven methods within the framework of active learning for a challenging design problem viz, sequence optimization of copolymers. We report the impact of surrogate models, kernels, and initial conditions on the convergence of an active learning method for the sequence design problem. This work outlines optimal strategies and parameters for inverse design of polymer sequence via active learning.