About


My research revolves around High Performance Computing and its applications in Radiation Material Science. Computing offers many advantages in Radiation Material Science in that we can see the fundamental physics taking place in a material at short time scales. For example we can see the damage produced by an high energy ion. Currently I work heavily with Density Functional Theory (DFT) [VASP, Quantum Espresso] and Molecular Dynamics (MD) [LAMMPS] to create MD potentials through Machine Learning.

Education

University of Tennessee, Knoxville, TN
Chansellors Fellow and PhD Candidate in Material Science and Engineering with minor in Computation Sciences, Fall 2013 - Current
Clemson University, Clemson SC
B.S. in Applied Mathematics with minor and Computer Science and Material Science, May 2013

Work

Graduate Teaching Assistant, University of Tennessee Knoxville
Graduate Thermodynamics 513 (Spring 2016 - Fall 2017)
Graduate Research Assistant, University of Tennessee Knoxville
Advisor Dr. William Weber

Publications

Journal Publications

Department Publications

  • Ostrouchov, C., Haranczyk, M., Computational Tools for Geometry Based Analysis of Crystalline Porous Materials, NSF - Lawrence Berkeley National Lab - Research Report , Jul. 2012
  • Ostrouchov, C., Wactel, P., Richardson, K., Examination of High Temperature Infrared Optical Materials in the GeAsS Ternary Glass System, NSF - International Research Experience for Undergraduates , July 2011

Poster and Presentations

[Presentation] The golden python packaging template
knoxpy, Knoxville, Tennessee, June 2018
[Invited Presentation] Machine learning oportunities in material
science, Joint Institute for Advanced Materials, Knoxville Tennessee, May 2018

[Presentation] celery: Tool for managing distributed task queues : knoxpy, Knoxville, Tennessee, August 2017

[Presentation] DFTFIT: Potential Generation for Molecular Dynamics Calculations
Materials Research Society, Boston, Massachusetts November 2016
[Presentation] DFTFIT: Potential Generation for Molecular Dynamics Calculations
American Ceramics Society HTCMC, Toronto, Canada, July 2016

[Poster] DFTFIT: Potential Generation for Molecular Dynamics Calculations : American Society for Metals (AMS) Student Poster Night, Knoxville, TN, November 2015

[Presentation] Creation of MD Potentials from Ab Initio Calculations
Oak Ridge National Laboratory, Monthly Energy Dissipation to Defect Evolution (EDDE) meeting, November 2015
[Poster] Tools for Geometry Based Analysis of Crystalline Porous Materials
Lawrence Berkeley National Laboratory (LBNL), Berkeley, CA, August 2012
[Poster] A Combined Numerical and Experimental Approach to Measuring Gap Conductance for Precision Glass Molding
Clemson University Creative Inquiry Poster Session, Clemson, SC, May 2011
[Presentation] A Combined Numerical and Experimental Approach to Measuring Gap Conductance for Precision Glass Molding
MS\&T Materials Science and Technology Journal of Undergraduate Research Symposium, Columbus, OH, October 2011

Awards and Leadership

Graduate Student Award for Excellence in Teaching
University of Tennessee, Spring 2017
Chancellors Fellowship
University of Tennessee 2015 - 2017
President, Materials Research Society
University of Tennessee 2016 - 2017
Secretary, Materials Research Society
University of Tennessee 2014 - 2015
Vice-President, Keramos Engineering Honors Society
Clemson University 2011 - 2012
Vice-President, Materials Advantage Society
Clemson University 2010 - 2011

Research

Molecular Dynamics Studies of LiNbO3

LiNbO3 is a material that our group is very interested in that has seen few MD studies due to it’s complex triclinic crystal structure. We are modeling the effects of high energy ions passing through LiNbO3 by applying the thermal shock model. The above image is a rendering of the electron density of LiNbO3 from a ground state Quantum Espresso run.

Monte Carlo Simulation of Ion Cascades

Density Map of Ion Collisions

Density Map of Ion Collisions

SRIM is a closed source code originally written in the 1970’s to calculate the range and stopping power of ions in matter. The code has gone through periodic updates with the most recent coming out in June 2013. Many groups in Nuclear Engineering and Material Science at the University of Tennessee use this software frequently and are frustrated with its restrictions, speed, and ability to only run on windows. Our group’s aim is to provide an open source, more stable, platform independent alternative to SRIM. See Github for updates.

Machine Learning and Molecular Dynamic Potentials

DFTFIT

The creation of Molecular Dynamics potentials typically are made through the use of fitting experimental data and Density Functional Theory data. Currently there are limited tools available online for creation of MD potentials. My goal for this research is to tie in well established codes (VASP, Quantum Espresso, and LAMMPS) for usage in constructing molecular dynamics potentials. More information to come! Please see Github for real-time updates.

Software

Zeo++

MFI Zeolite

MFI Zeolite

During the summer of 2012, I was given the opportunity of an internship at Lawrence Berkeley National Lab (LBNL) working under Dr. Maciej Haranczyk. I worked on Zeo++ throughout that summer and added features to characterize the distribution of the voids within a given Zeolite. This work materialized into a paper published in early 2013. Zeo++ is written in C++ with interfaces through python and VMD. Zeo++ allows for the calculation of geometrical parameters describing pores within the material and easy visualization of these parameters.

PYQE

LiNbO3

LiNbO3

While taking MSE 613 (Introduction to Density Functional Theory) by Dr. Haixuan Xu we made heavy use of Quantum Espresso. I was often frustrated by syntax errors in my input scripts along with how many steps could not be easily automated. BASH scripting was used at first but I quickly learned that by creating a Python library I could speed up the process. The software is open-source and available on Github PyQE.

DFTFIT

DFTFIT is a software that uses DFT simulation data for fitting molecular dynamic potentials. DFTFIT uses the least square method to fit the stresses, total energy, and forces of a given set of configurations. The user has the flexibility of changing the weights for fitting and uses relative error to judge the fitness of parameters. While this code is in its early stages, it has already proven that it can come up with good potentials for MgO. After a potential haso been created, DFTFIT has tools for easily calculating measurable experimental properties such as the lattice constant and bulk modulus. The software is open-source and available on Github DFTFIT.