2D GRMHD Simulation & Parallel Tensor Engine

Role: Graduate Project Lead

This project integrates numerical astrophysics with distributed scientific computing, focusing on the physical modeling of relativistic jets in Active Galactic Nuclei (AGNs). As the graduate lead for this course project, I originated the research concept, authored the formal proposal, and am currently managing a team of two undergraduate researchers to build out the simulation and analysis pipeline.

Our objective is to investigate the computational and physical structure of a two-dimensional relativistic jet. The project is divided into two primary development tracks:

1. Simulation & Physical Modeling

We will utilize Athena++ to simulate a 2D slice of a relativistic magnetized jet. This involves configuring a 2D Cartesian
grid, establishing boundary conditions, and defining an ideal relativistic gas equation of state.

2. Custom Distributed Tensor Engine

We are developing a custom distributed-memory tensor post-processing pipeline built in C and MPI. This engine is
designed to handle parallel HDF5 input and compute complex invariant quantities—including the stress-energy and
magnetic tensors—while systematically measuring HPC efficiency metrics like strong/weak scaling and I/O throughput.

Current Status (As of March 2026)

We are currently in the initial architecture phase, focusing on environment setup, defining the Athena++ simulation parameters, and structuring the parallel processing workflow for the tensor engine.

Note — This research was developed and executed as my graduate project for the High Performance Computing class
(CS-6373) at the University of Tulsa.