Capstone Projects

Image	A Reduced Order Model on Nonlinear Manifolds using Interpretable Variational Autoencoders Year: 2021–2022 Student Trainees: Peiyi Chen Faculty Advisors: Johann Guilleminot Other Authors: Tianchen Hu An interpretable variational autoencoder-based nonlinear manifold reduced-order model improves simulation efficiency for complex, advection-dominated problems. The approach achieves comparable accuracy to traditional linear models using a much smaller latent space while providing insight into the learned nonlinear manifolds.
Image	Combining Methods of Machine Learning and Elastodynamics to Reduce the Cost of Viscoelastic Nanocomposite Predictions Year: 2021–2022 Student Trainees: Yingfan Wang, Nicholas Finan Faculty Advisors: L. Catherine Brinson, Cynthia D. Rudin Other Authors: NA In order to predict the complete behavior of complex microstructural materials, one accurate canonical solution is 3D finite element modeling (FEM). This method offers physically accurate predictions for microstructures of high complexity but incurs a large computational cost in doing so.
Image	Operator learning for computational homogenization on 2D material Year: 2021–2022 Student Trainees: Yixin Tan, Hao Zhang Faculty Advisors: Johann Guilleminot, Jianfeng Lu Other Authors: NA Computational homogenization is a well-employed scheme to calculate the ”averaged” property of the multi-scale materials that have both large computational domain while small characteristic scale.
Image	Single-phase hierarchical phononic metamaterial templates with effectively scale-separated band gaps Year: 2021–2022 Student Trainees: Zhi Chen Faculty Advisors: L. Catherine Brinson, Cynthia D. Rudin Other Authors: Mary Bastawrous Hierarchical phononic materials have sparked interest as they can manipulate the dispersive characteristics of vibrational waves at multiple frequency ranges. Typically, the hierarchical length scales are relatively close, which makes multiscale methods unsuitable. Thus, the design process can only be carried out on a fine grid, which significantly increases the computational cost.
Image	Deep inverse design on hydrophobic patches on DNA origami for mesoscale assembly of superlattices. Year: 2022–2023 Student Trainees: Po-An Lin, Ben Ren Faculty Advisors: Gaurav Arya, Jordan Milton Malof Other Authors: Leslie Collins We plan to work on inverse-designing the hydrophobic patches on DNA origami for mesoscale assembly of superlattices. This problem is important as understanding the complex assembly rules of patchy DNA origami will create avenues for functioning mesoscale materials with practical applications, such as therapeutics, drug delivery, or metamaterials.
Image	Operator Learning for Predicting Fracture Paths in Heterogeneous Materials Year: 2022–2023 Student Trainees: Ariana Quek, JinYi Yong Faculty Advisors: Johann Guilleminot Other Authors: NA We present a data-driven approximation framework of a phase field model for brittle fracture. The proposed approach aims to alleviate the computational cost associated with full-scale simulations in studying brittle fracture in heterogeneous materials; where the fracture toughness of the material is spatially varying.
Image	Predicting the mechanical response profile of porous materials via their microstructures Year: 2022–2023 Student Trainees: Winston Lindqwister, Jacob Peloquin Faculty Advisors: Ken Gall Other Authors: Manolis Veveakis Porous media represent a wide range of materials that have yet to be fully understood and present a challenge in being harnessed to their full capacity. From rocks to bones, the ubiquity of porous materials across engineering disciplines cannot be overstated.
Image	Applied Machine Learning as a Driver for Biomedical Polymers Year: 2022–2023 Student Trainees: Sam McDonald, Quinn Lanners Faculty Advisors: Cynthia D. Rudin, L. Catherine Brinson Other Authors: Emily Augustine, Matthew Becker Polymers are ubiquitous to almost every aspect of modern society and their use in medical products is similarly pervasive. Despite this, the diversity in commercial polymers used in medicine is stunningly low.
Image	Machine Learning Guided Optimization of Hydrogel Composition for Enhanced Metal Year: 2023–2024 Student Trainees: Darryl Taylor, Vasishta Somayaji, Daniel Michael Faculty Advisors: Jie Liu, Wilkins Aquino Other Authors: Andrew Jones This interdisciplinary research project combines expertise in material science and machine learning to address the environmental challenges associated with metal extraction, specifically focusing on strategic metals like nickel, cobalt, and lithium. The team aims to develop nanocellulose/cyclodextrin-based hydrogels that efficiently and selectively capture these metals from aqueous [...]
Image	Enhancing Solar Cell Performance through Machine Learning: A 2D Perovskite Perspective Year: 2023–2024 Student Trainees: Migon Choi, Aiden Sable, Nicky Ramos Faculty Advisors: Olivier Delaire Other Authors: Hai Li, David Mitzi Two-dimensional (2D) perovskites are emerging as promising candidates to address the stability issues of three-dimensional perovskites. However, the application of machine learning (ML) to predict the performance of 2D perovskite solar cells is still in its infancy. Furthermore, organic cation spacers in 2D perovskites are often omitted in predictions of solar cell performance.
Image	Inverse Design of Quasicrystals and Complex Nanoparticle Architectures formed at Interface of Two Mediums through ML Approaches. Year: 2023–2024 Student Trainees: Safak Callioglu, Jack McErlean Faculty Advisors: Gaurav Arya Other Authors: Hau-Tieng Wu Nanoparticle assemblies have attracted considerable interest due to their various industrial applications, which motivate the discovery of untraditional complex nanoparticle architectures.
Image	Retrieving Structured Data from Polymer Nanocomposite Literature using LLMs Year: 2023–2024 Student Trainees: Defne Çirci, Ghazal Khalighinejad Faculty Advisors: L. Catherine Brinson Other Authors: Bhuwan Dhingra Automated materials synthesis requires historical data, but extracting detailed data and metadata from publications is challenging. In this project, we develop strategies for using large language models (LLMs) for rapid, autonomous data extraction from materials science articles in a format curatable by a materials database.
Image	Towards transition path sampling, an off-policy learning approach for energy manifolds Year: 2024–2025 Student Trainees: Alexander Bonacci, JinYi Yong Faculty Advisors: Johann Guilleminot Other Authors: John Dolbow Understanding transition paths in molecular systems is key to uncovering rare but essential events that govern transitions between metastable states, such as conformational changes and chemical reactions. Traditional molecular dynamics (MD) simulations often struggle to capture these phenomena due to the immense timescales and dimensionality involved.
Image	A general, domain-agnostic approach for the inverse design of complex microstructures Year: 2024–2025 Student Trainees: Yiyang Sun, Anlan Chen, Yasha Saxena Faculty Advisors: Cynthia D. Rudin, L. Catherine Brinson Other Authors: Tatiana Segura In this project, we aim to develop a general framework to calculate system behaviors from microstructure images. To establish the generalizability of the approach, we employ convolutional neural networks (CNNs) to process microstructure images from two domains - polymer nanocomposites (PNCs) and granular biomaterials.
Image	Inverse Design of Acoustic Metamaterials using Diffusion Models Year: 2024–2025 Student Trainees: Daniel Duke, Han Zhang Faculty Advisors: Gaurav Arya, L. Catherine Brinson Other Authors: NA The inverse design of acoustic metamaterials presents significant challenges due to the complex, non-intuitive mapping between desired acoustic properties and corresponding geometries. This project explores the application of diffusion models to solve this problem, leveraging their proven capability in handling one-to-many mappings.
Image	Extracting Anomalous Data from Materials Science Images with Machine Learning Year: 2024–2025 Student Trainees: Tyler Wilson, Ja’Nya Breeden, Io Saito Faculty Advisors: L. Catherine Brinson, Olivier Delaire Other Authors: Jian Pei In many scientific fields it is imperative to analyze data in numerical or image formats. However, there are many challenges that accompany this process. Image data of experimental signals from materials science often contains features which hinder both qualitative and quantitative analysis in the form of anomalies.
Image	Machine learning based optical metasurfaces design for signal processing Year: 2025–2026 Student Trainees: Yuruo Zheng Faculty Advisors: Other Authors: Jiaren Tan, Natalia Litchinitser Optical metasurfaces enable compact and ultrafast analog optical computing functions such as spatial and temporal differentiation, which are critical for edge detection, motion sensing, and real-time signal processing.
Image	A Data-efficient Multi-property Design of Electroactive Polymer Formulations via Bayesian Optimization Year: 2025–2026 Student Trainees: Zijie Xu Faculty Advisors: Xiaoyue Ni, Simon Mak Other Authors: Yilei Bu, Feilin Feng, Jingtong Ma, Matthew Becker Data-efficient optimization of multifunctional polymer systems remains challenging because of high-dimensional formulation spaces and competing property requirements. Here, we present a multi-objective Bayesian optimization (BO) framework for accelerated design of electroactive polymer (EAP) formulations, demonstrated on PEDOT:PSS/PVA composite films.
Image	Effects of Crystal Symmetry Encoding on Large Language Models for Crystal Structure Prediction Year: 2025–2026 Student Trainees: Gabriel Graf Faculty Advisors: Volker Blum Other Authors: Oliver You, Andy Kapoor Crystal structure prediction (CSP) is a foundational component of high-throughput materials discovery and design. Despite the success of classical CSP methods, their applicability to large, complex systems of physical interest has been limited by the computational cost of methodically exploring an exponentially large configuration space.