Chemistry

Home » Konstantinos Vogiatzis
Tag: Konstantinos Vogiatzis
Vogiatzis Featured Image

Vogiatzis Group Publishes in Journal of Physical Chemistry Letters

by

Grier Jones, fifth year chemistry PhD student, and Associate Professor Konstantinos Vogiatzis recently published a new data-driven quantum chemistry method, based on the reduced-density matrix (RDM) formulation of quantum mechanics, in the Journal of Physical Chemistry Letters. This publication was developed in collaboration with University of Tennessee, Knoxville alumnus Professor A. Eugene DePrince (’05) and his research group at Florida State University. DePrince’s group specializes in the development of novel RDM methods for the treatment of strongly correlated electrons.

Strong electron correlation lies at the heart of molecular quantum mechanics and, in particular, at the heart of electronic structure theory. Configuration interaction (CI) theory provides an exact description of strong correlation, but it suffers from exponential scaling with respect to the number of correlated electrons and orbitals. As an alternative, variational two-electron RDM (v2RDM) methods have been introduced since the energy of a many-electron system can be formulated exactly using the two-electron RDMs (2RDMs). One interesting property is that the 2RDM can be formulated without explicit knowledge of the wave function. In practice, finding a wave function that maps explicitly to the 2RDM can be very tricky, and the resulting deviation between CI- and RDM-based methods can be very large.

To resolve this issue, a collaboration between the Vogiatzis and DePrince groups lead to the development of the data-driven v2RDM (DDv2RDM) method to learn CI-quality energies using data generated using the v2RDM-complete active space self-consistent field (CASSCF) method. Using proof-of-principle calculations, they found that the model learns the correction the v2RDM energy near-chemical accuracy (1 kcal/mol). They also introduced the use of SHapley Additive exPlanation (SHAP) values, a feature importance method based on cooperative game theory, to analyze the how their physics-based features affect model performance. The SHAP analysis confirmed that the features that impact the model performance the most (and least) correspond well to insights based on physical principles.

Read the full article here.

Vogiatzis Group Publishes in npj Computational Materials

by

Associate Professor of Chemistry Konstantinos Vogiatzis, in collaboration with Professor of Mathematics Vasileios Maroulas and Eastman Chemical Company, has published a new machine learning model for predicting the properties of new polymeric materials.

Polymers are everywhere. From cookware to medical devices, polymers have become important to modern life due in part to a growing list of potential uses, and desirable properties like high durability and resistance to corrosion.

Creating new polymers can be an expensive, time-consuming process. Because of this, researchers attempt to predict the future properties of polymers using a variety of tools. Computational prediction methods allow researchers to screen polymer combinations for the desired properties before beginning experimentation. However, finding ways to represent polymers as machine-readable inputs can be difficult, creating a challenge for developing accurate prediction models.

Vogiatzis’ team is attempting to tackle these challenges by creating a deep learning method to predict polymer properties called PolymerGNN. PolymerGNN relies on state-of-the-art graph neural networks (GNN) and machine learning to predict the properties of new polymers using a database of complex polyesters.

“Polyesters offer a diverse material space formed by considering many different types of multifunctional acids and glycols, which are the building blocks of these materials,” said Vogiatzis. This, coupled with other complex properties of polyesters, creates a large materials design space Vogiatzis and his team were able to leverage in the development of PolymerGNN.

Vogiatzis worked with Vasileios Maroulas and students Owen Queen, Dr. Gavin McCarver and Sai Thatigotla to develop the general framework and GNN-based machine learning model for PolymerGNN. Collaborators from Eastman Chemical Company synthesized a set of more than 240 polymers and helped compile a database of properties which was used to train PolymerGNN.

Once trained, PolymerGNN accurately predicted both glass transition temperature and intrinsic viscosity. Glass transition temperature is the temperature at which a polymer shifts between a hard state and a softened state. Intrinsic viscosity is a measurement of a polymer’s molecular weight, which can indicate the polymer’s melting point, crystallinity, and tensile strength. These properties are fundamental to the ultimate physical traits of a given polymer and are critical to the development of adhesives, plastics, and more.

Vogiatzis’ team recently published this work in npj Computational Materials, an open access journal from Nature Research. They have also released PolymerGNN as an open-source codebase. Vogiatzis and Maroulas have collaborated on previous machine learning projects published by the American Chemical Society and Nature Communications. Read the most recent publication here.