Connecting Modeling, Simulations, and Machine Learning with Experiments for Soft Materials Design - Structure-Property Relationships

Explore a virtual seminar presented by Prof. Arthi Jayaraman from the University of Delaware, focusing on connecting modeling, simulations, and machine learning with experiments for design-structure-property relationships in soft materials. Delve into the Jayaraman lab's study of soft materials, including polymers and colloids, using molecular modeling, simulations, theory, and machine learning tools. Learn about the Computational Reverse Engineering Analysis of Scattering Experiments (CREASE) method, its application in structural characterization of soft materials using Small Angle X-ray Scattering (SAXS), and how machine learning enhances this process. Examine case studies on vesicles, fibrillar structures in amphiphilic polymers, methylcellulose's unique phase behavior in aqueous solutions, and synthesized spherical particles. Discover the 'PairVAE' technique for pairing structural characterization data from complementary techniques, and gain insights into predicting color for CREASE's reconstructed structures.

Intro
Connecting molecular modeling, simulations, & machine le
Jayaraman lab studies soft materials polymers, colloids, pa
Our tools: Molecular modeling, simulations, theory, & machin
Focus of today's talk
Structural Characterization of Soft Materials using Small Angl
Computational Reverse Engineering Analysis of Scattering Ex
CREASE Step 1: Genetic algorithm (GA)
How machine learning has helped CREASE
CREASE for analyzing vesicles' structure
CREASE vs. SASVIEW fit with core-multi-shell mode vesicles with dispersity in all relevant dimensions
CREASE: Step 2: Molecular reconstruction within GA informed
CREASE applied to fibrillar structures in amphiphilic polym
Methylcellulose and its unique phase behavior in aqueous s
Dimensions from SAXS data analyzed by CREASE vs. analytical
CREASE applied to SAXS on synthesized spherical particle
Predict color for CREASE's reconstructed structure
'PairVAE' for Pairing Structural Characterization Data Complementary Techniques