Professor Thomas Jones

This group's computational program at MSI centers on pioneering, high fidelity fluid dynamic simulations that address timely and fundamental astrophysical problems, specially including the physics of hot, magnetized cosmic plasmas often accounting for associated "subgrid micro-physics." In addition to non-adiabatic fluid behaviors and such things as self-gravity, much of this work includes acceleration and transport of high energy cosmic ray particles and their feedback on bulk fluids. The group's simulations use high performance magnetohydrodynamic (MHD) codes and supporting particle transport codes developed within the group in partnership with academic and industrial collaborators. The MHD codes include important, complementary physics (e.g., self-gravity, radiative energy balance, and relativistic cosmic ray (CR) transport and acceleration, plus in some contexts gravitational interactions of the fluids with "collisionless dark matter" particles that often dominate local gravity). As part of this effort the researchers have focused on development of a new generation, high order, high performance astrophysical MHD code named "WOMBAT." This code is unique in the astrophysical community in both performance and accuracy. It incorporates state-of-the-art parallelization and vectorization methods (utilizing both CPUs and GPUs), along with a unique domain decomposition strategy. It now includes the exceptionally high accuracy WENO 5th order MHD solver along with constrained transport of magnetic flux to assure maintenance to machine precision of a divergence-free magnetic field. The researchers have demonstrated in multiple tests that "WENO-WOMBAT" generally achieves approximately twice the effective resolution of the most popular community MHD codes for a given computational cost. They also have also demonstrated sustained single core CPU performance exceeding 20% theoretical on new generation CPU chips and near theoretical weak scaling to more than 200,000 cores on external Cray systems made available to to the group for testing. The exceptional properties of WENO-WOMBAT are documented in a 2019 issue of the Astrophysical Journal Supplements. The code also can make very effective use of current generation GPUs. Among other key strengths, this code allows researchers, for the first time ever, to properly simulate the cosmological MHD turbulent dynamo (an astrophysical "Holy Grail"). The dynamo is thought (but not yet explicitly verified) to be primarily responsible for amplification of magnetic fields during cosmic structure formation following the Big Bang. Significant magnetic fields, in fact, are observed to be ubiquitous in diverse settings, but their origins and controling dynamics remain problematic. The outstanding and unique capabilities of WENO-WOMBAT open up totally new territories to exploration and position this group at the head of the line to contribute to several vital astrophysical research areas. In addition to the mentioned cosmological questions these include the basic character of cosmic MHD turbulence, the interactions of hypersonic MHD jets with multiple dynamical cosmic media, and the creation of giant, magnetized, radio emitting structures in galaxy clusters. Additionally, the path forward with WENO-WOMBAT addresses outstanding, complementary issues in understanding the origins and interpretation of amazingly rich magnetized structures just now being revealed in and around galaxies and galaxy clusters by the new generation of radio telescopes that are looking deeper with better spatial and spectral information than could be imagined a few years ago. The structures are visible through emissions by relativistic, CR electrons embedded, accelerated and transported within the local, hot (10⁸ K) plasma. Many of the observed structures appear to be the result of interactions between the local plasma and plasma expelled by high energy plasma jets from the nuclei of embedded galaxies and from multiple supernovae in galaxies. These interactions are quite complex; WENO-WOMBAT is likely the only code currently able to address these issues properly.

The group is working on three specific projects for 2024:

• Large Scale, High Fidelity MHD Simulations of Galaxy Cluster Formation 
• MHD Simulations of Interactions of High Energy Galaxy Outflows with Realistic Galaxy Cluster Environments
• Machine Language Analysis and Structure Identification of 3D Simulation Data 

This research was featured on the MSI website in October 2014: Computational Astrophysics.