(UC Berkeley) Devised and implemented a discrete event simulation of a domain-based fog architecture model and demonstrated different queuing methods (FCFS, EDF, CR) to improve QoS for fog networks, formulating and modeling a discrete-event simulation based on various constraints on job size, deadline, and location.Ī Method of Modeling Asteroid Intercept Trajectory (UC Berkeley) Rigorous implementation of finite volume schemes (Godunov, Roe) in evaluating the Lighthill-Whitham-Richards traffic model, a flow-based traffic congestion model based on vehicle density that accounts for disturbance shocksĮfficiency of a Domain-based Fog Computing Architecture Optimal Traffic Signal Timing with a Continuum Flow Congestion Model (UC Berkeley) Rigorous formulation and implementation of quadratic finite elements and custom meshing algorithm to simulate frequency response and resonance phenomena of a waveguide using the 2D Helmholtz Equation with Sommerfeld radiation conditions Time-Harmonic Wave Propagation Simulation in Waveguides (UC Berkeley) Finite-difference method numerical modeling of hydrodynamically and thermally developed convective flow of different fluids in various geometries and analysis of their behaviors, comparison of scheme efficacies with dimensionless quantities Natural Convection Boundary-Layer Flow Analysis (UC Berkeley) Implemented and formulated efficient variations of the Uzawa and GMRES algorithms applied to the matrix computations arising from the numerical solution to the instabilities observed from the 2d Euler Equations for inviscid flows (Georgia Tech) Developing a one-stop course planning platform to aggregate a course catalog, a requirements planner, visualized grading distributions, enrollment histories, a schedule optimizer, a ratings system, a course recommender, and an exam/syllabi database.Įfficient Computational Methods for Kelvin-Helmholtz Instabilities (Georgia Tech) Used confounding factors within historical congestion data to develop predictive autoregressive integrated moving average (ARIMA) analyses, modeling the cyclic nature of transit demand Performance profiling and optimization across different pragmas and strategiesĬorrelative Analysis and Visualization of Transportation Ecosystems
ICMD STANFOR PORTABLE
(NASA Supercomputing) Developed CPU and GPU portable iterative method solvers with new OpenMP implementations across different specialized compilers. Portable OpenMP CPU/GPU Parallelization of Iterative Methods (Georgia Tech) Regression analyses on individual and intersectional demographic factors affecting faculty salaries to determine best predictors across five public university systems Intersectional Demographic Regression Analysis of University Employee Salaries Testing on a conceptual operation of a future Mars mission (Georgia Tech) Developed generalized discrete event simulation model of space mission timelines, providing logistical and risk analyses. Space Mission Logistics Discrete Event Simulation Framework (Stanford) Developing a quadratic finite element solver with multigrid and adaptive mesh refinement accelerations using Regent, a Nvidia Legate based research task-based parallel language for proof of concept purposes
Task-based Parallel Multigrid and Adaptive Mesh Refinement FEM Solver (Stanford) Leveraged Jax to optimize Discontinuous Galerkin numerical method code, to understand methods of parallelizing numerical methods across CPUs/GPUs/TPUs using matrix computations to enable compiler-driven parallelism NeurIPS Workshop on Machine Learning for Geophysical and Geochemical Signals.Accelerated High-Order Discontinuous Galerkin Methods Machine Learning in Solid Earth Geoscience Conference, Santa Fe, NM (2019) Women in Data Science Stanford Earth Workshop, Stanford, CA (2019) National Academies Committee on Seismology and Geodynamics Fall Meeting, Washington D.C. IRIS Distributed Acoustic Sensing Virtual Workshop and Tutorial, virtual (2020)Ĭongressional briefing on Machine Learning in Seismology, hosted by the Seismological Society of America, Washington D.C. Seismological Society of America Annual Meeting: Machine Learning Plenary Panel, virtual (2021) Symposium on Artificial Intelligence and Earthquake Engineering, EERI San Diego Chapter, hybrid (2022)
0 kommentar(er)
