6 Result(s)

Machine Learning-based Materials Design

Designing new environmentally friendly and cheap materials for practical applications is one of the main challenges of our century. This process is however very slow because synthesizing and testing new materials take time and have considerable cost. Computational methods provide an alternative method to screen materials faster and circumvent the costly and slow experimental trial-and-error approach. In this research area, machine learning-based methods have emerged as flexible tools recently to predict the properties of hitherto unknown materials based on previously known information. The Szilvasi group is working on developing databases and machine learning-based workflows to design new materials in the area of catalysis, energy storage, ...

Required Availability
The End of Time
Course Credit?
Yes - CHE 498
Paid Position?
No

Computational Catalysis

Catalysis is used to produce most chemicals worldwide. Thus, optimization of catalysts is relevant for both economic and environmental reasons. The ever-increasing computational power has led to the rise of computational research in catalysis that has been one of the main developments of the previous decades in the field. Computations have helped understanding chemical bonding, assign spectroscopic features, and explore reaction mechanisms among others. Regarding this latter, identifying rate-determining steps and analyzing critical chemical interactions have become standard tools to understand catalytic reactions and design more active, selective, and/or stable catalysts. As such, the Szilvasi group is interested in using computational met...

Required Availability
The End of Time
Course Credit?
Yes - CHE 498
Paid Position?
No

computational catalysis

The control of chemical transformation via catalysis is both an exceptional intellectual challenge and critically important to the Nation. Catalysis is central to energy production and utilization, to chemical manufacturing, to the minimization of environmental impact, and it has been arguably the single most important agent for sustainable development in the developing world. The revolutions in nanotechnology and high performance computing provide unprecedented new opportunities to elucidate the fundamental principles governing the control of chemical transformation by catalysts. Indeed, the coupling of theory, modeling and simulation with experiment will provide the most profound insights into catalyst behavior and thus enable the design ...

Required Availability
The End of Time
Course Credit?
Yes - CH396:398
Paid Position?
No

Analysis of physics data using deep learning methods

Students can work on the following projects: 1. Novel approaches to simulation of nuclear track detector data for the MoEDAL experiment at the Large Hadron Collider using Generative Adversarial Networks 2. Position reconstruction of gamma and beta decays in the EXO-200 neutrinoless double-beta decay experiment using Convolutional Neural Networks Fluency in programming languages (Python and/or C, Linux OS) are required. Familiarity with deep learning technique and relevant software (PyTorch, Keras) are preferred. Access to GPU computing system will be provided...

Required Availability
The End of Time
Course Credit?
Yes - PH495
Paid Position?
No

Computational peptide chemistry

Advanced computational electronic structure methods will be used to calculate the geometries, vibrational frequencies, energetics, and excited state properties of important compounds of biological interest. Both correlated molecular orbital theory and density functional theory will be used. The focus of the work is on charging of peptides for explaining mass spectrometry results for both cationic and anionic peptides. The cationic work will focus on transition metal ion charging. Both types of studies are relevant to the study of the Human proteome....

Required Availability
The End of Time
Course Credit?
Yes - CH396:398
Paid Position?
No

Computational heavy element chemsitry

We are interested in developing a fundamental and predictive understanding of actinide chemistry in aqueous solution under conditions relevant to nuclear-waste storage and reprocessing of spent fuel to address aggregate and colloid formation. Intractable, small aggregates in nuclear-waste streams can impair clean-up, forcing a low-level waste stream to be treated as high-level waste, thereby increasing treatment costs. Metal oligomers, aggregates, clusters, nanophases and colloids are ubiquitous in aqueous chemistry. Thought to form via the condensation reactions of hydrolyzed metal ions, intrinsic dissolved aggregates or colloids are generally described as poly-dispersed hydroxides or hydrous oxides with varying stoichiometry and no well-d...

Required Availability
The End of Time
Course Credit?
Yes - CH396:398
Paid Position?
No