About Me


About me

Over the years, I have worked on a number of different things from experimentally studying fatigue in biomaterials to computationally modelling fatigue problems to finally working on GPU systems for material modelling. I enjoy tackling these challenging problems, figuring out how to make programs go screaming fast, and working out numerical methods. Outside of work, I take pleasure hanging out with my dog and partner, hiking, and shooting landscape photography.

Current Work

I am a computational engineer at Lawrence Livermore National Laboratory (LLNL) where my work is focused on the intersection between the bleeding-edge of HPC systems and solid mechanics. Part of this work has involved leading the development of LLNL’s new open-source crystal plasticity/solid mechanics finite element code, ExaConstit, as part of the ExaAM project. ExaAM is an Exascale Computing Project (ECP) project focused on modelling the additive manufacturing from the melt pool all the way up to the part scale. My work there has been leading the efforts to connect microstructure to the part scale response. As part of this work, the ExaConstit code was born. This new code has been designed from the ground-up around MFEM to run on the next-gen exascale computing systems such as Frontier and El Capitan. So, I have implemented a number of GPU friendly algorithms into ExaConstit that have led to speed-ups of 15x over the CPU runs. These improvements have been vital in providing users the ability to run simulations at unprecedented levels of detail without spending weeks waiting on the simulations. Outside this code, I also am a primary developer for a number of other libraries such as ExaCMech, a constitutive modelling library, and SNLS, a nonlinear solver library for small systems. For each of these other open-source libraries, I have ported the libraries over to Nvidia and AMD GPU systems and examined optimization schemes for both. Additionally, I have looked at adding new solvers, numerical schemes, and new models for each library. Overall, I have been lucky to work on open-source as part of my day job. These sorts of project are especially in the computational solid mechanics field as traditionally research codes are never released to the public.

PhD Thesis Work

Work related to my thesis centered around characterizing intragrain/intra-crystal deformation mechanisms and how these mechanisms evolve and form networks through a polycrystalline material during cyclic loading. I’ve accomplished a majority of this work through the use and expansion of various different single crystal elasto-viscoplastic models. Overall, my thesis work has led to me having a strong background in a number of fields ranging from: continuum mechanics; numerical methods (nonlinear FEMs, krylov subspace methods, ODE solvers); far-field high energy diffraction; high performance computing through the use of MPI and OpenMP; and data analysis/visualization of large data sets. It has also exposed me to various dimensionality reduction methods, statistical inference methods, graph theory, and image analysis such as connected component labeling.

During my time at Cornell, I spent a great deal of my time collaborating with Professor Matthew Miller and Dr. Mark Obstalecki for a combined set of experiment and modelling studies. They were responsible for running high energy x-ray diffraction experiments on copper samples using both far-field and near-field techniques. Through our combined modelling and experimental approach, we were able to show the strengths and weaknesses of various different micro-mechanical hardening laws. This interaction really imparted onto me the need for more combined modelling and experimental efforts. Without these efforts, it can be incredibly easy to get lost in the weeds on the modelling side and forget about how real materials actually behave. New experimental methods also allow us to verify and obtain new single crystal elastic constants along with varying plasticity parameters. On the modelling side, we can provide the experimentalist a wealth of detailed information, once we know our models limitations. Current experimental methods on average are either only able to return grain average information while in-situ or ex-situ intragrain data. Simulations therefore provide a valuable complement to experiments in that they are able to return highly resolved intragrain data for a wide number of parameters.

Disclaimer

Please note that any thoughts or opinions discussed throughout this website do not represent my employer Lawrence Livermore National Laboratory (LLNL).