A key component of the DOE Scientific Discovery through Advanced Computing (SciDAC) program is the Scientific Computation Application Partnerships. Through these partnerships, scientists conduct complex scientific and engineering computations simulating real-world conditions in targeted application projects. Essential to these activities is the exploitation of leadership-class computing resources. The partnerships capitalize on expertise in applied mathematics and computer science, algorithms and methods, and scientific software tools at the SciDAC Institutes.
Researchers in the Mathematics and Computer Science Division at Argonne are involved in ten SciDAC partnerships cofunded by the DOE Advanced Scientific Computing Research (ASCR) program and the various DOE Office of Science programs listed.
Biological and Environmental Research
The CANGA project seeks to develop new high-performance coupling approaches and capabilities for coupled Earth system models on next-generation computing architectures. Argonne efforts will focus on verification and the coupling of miniapps.
Fusion Energy Sciences
This project seeks to develop and integrate high-performance simulation tools for predicting plasma facing component (PFC) operating lifetime and the impact of the evolving surface morphology and composition of tungsten-based PFCs on plasma contamination.
Tokamak Disruption Simulation – PI Barry Smith
Plasma disruptions are a major problem facing the development of reactor-scale devices. To address this problem, we will simulate the strongly coupled, nonlinear physical phenomena in such disruptions.
High Energy Physics
Community Project for Accelerator Science and Simulation 4 (ComPASS-4) – PI Stefan Wild
ComPASS-4 is a partnership of computational accelerator physicists, applied mathematicians, and computer scientists. The mission is to develop and deploy the state-of-the-art accelerator modeling tools necessary to advance new designs, concepts and technologies for particle accelerators.
HEP Data Analytics on HPC – PI Rob Ross
This project involves development and deployment of tools and algorithms that will enable high-performance computing facilities to meet the new data analysis demands in both energy and intensity frontiers and allow computationally expensive physics studies to be completed on time scales that are not currently feasible. Specifically, we will focus on three areas: high-dimensional parameter fitting, workflow automation, and introduction of HPC data resources into applications.
HPC Framework for Event Generation at Colliders – PI Paul Hovland
Monte Carlo event generation is the first step in simulating the data collected by experiments at the Large Hadron Collider. A key step in event generation is using intelligent Monte Carlo sampling algorithms to map out the phase space of the integrand for some interaction amplitude. This project will implement new Monte Carlo techniques for tackling this integration process more efficiently and for scaling such calculations on high-performance computers.
This project involves developing and applying new methods and algorithms in extreme-scale inference and machine learning. We will apply the new techniques to cosmological surveys across multiple wavebands. The results will improve the sensitivity and robustness of the final cosmological constraints in multiple ways.
Nuclear Computational Low Energy Initiative (NUCLEI) – PI Stefan Wild
The NUCLEI project seeks to advance large-scale nuclear physics computations in order to dramatically increase our understanding of nuclear structure and reactions and the properties of nucleonic matter.
Towards Exascale Astrophysics of Mergers and Supernovae (TEAMS) – PI Anshu Dubey
The TEAMS project seeks to develop improved simulations of supernovae explosions and neutron-star mergers, with the aim of predicting signatures of these events and advancing scientific understanding of the creation of the heaviest elements. Achieving these goals requires software capable of exploiting current and future architectures.
Simulation of Fission Gas in Uranium Oxide Nuclear Fuel – PI Barry Smith
In uranium oxide nuclear fuel, understanding the retnetion and release of fissuon gas atoms is important. In this project we use multiscale simulations to study the behavior of the diffusion mechanisms. The results are then compared with experimental data.