Skip to Main Content

Title ImagePublic Abstract

 
Collapse

DE-SC0025203: Quantum Tensor Network Algorithms for Simulating Physics of Hot Dense Plasmas

Award Status: Active
  • Institution: The Johns Hopkins University, Baltimore, MD
  • UEI: FTMTDMBR29C7
  • DUNS: 001910777
  • Most Recent Award Date: 08/20/2024
  • Number of Support Periods: 1
  • PM: Akli, Kramer
  • Current Budget Period: 08/01/2024 - 07/31/2025
  • Current Project Period: 08/01/2024 - 07/31/2027
  • PI: Titum, Paraj
  • Supplement Budget Period: N/A
 

Public Abstract

Quantum Tensor Network Algorithms for Simulating Physics of Hot Dense Plasmas
Project Summary/Abstract

 

Paraj Titum, Johns Hopkins University Applied Physics Laboratory (Principal Investigator)
Frank R. Graziani, Lawrence Livermore National Laboratory (Co-Principal Investigator)
Gregory Quiroz, Johns Hopkins University Applied Physics Laboratory (Co-Investigator)
John Van Dyke, Johns Hopkins University Applied Physics Laboratory (Co-Investigator)
Michael L. Wall, Johns Hopkins University Applied Physics Laboratory (Co-Investigator)
Ilon Joseph, Lawrence Livermore National Laboratory (Co-Investigator)

 

Project Objective
We propose the use of quantum computing to investigate the classical dynamics of hot dense
plasmas using a novel representation of plasma properties in terms of tensor networks which will
subsequently be simulated with quantum circuits. Our multi-disciplinary team will focus on hydrodynamic
instabilities and turbulence utilizing recent advances in classical and quantum algorithms
using tensor networks to investigate plasma dynamics.


Project Description

The project will be split into two tracks focusing on classical and quantum algorithm development
based on tensor networks. The classical thrust will focus on developing novel tensor network representations
of the physical properties of plasma and its dynamics. Not only will the tensor network
representations allow subsequent simulation via quantum tensor networks, these simulations will be
compared directly with the state-of-the-art classical simulation algorithms based on finite element
discretization of mix/turbulence equations approaches currently used for simulating Inertial Confinement
Fusion (ICF) and Magnetic Fusion Energy experiments. The quantum algorithms thrust
will focus on necessary advances to simulate plasma dynamics on the noisy quantum hardware of
the near-future: designing low-depth quantum circuits from the tensor network representations,
developing variational approaches to optimizing tensor networks directly on a quantum computer,
and incorporating noise-robustness to improve the performance of the quantum algorithms on noisy
quantum hardware.

Our focus will be on studying the Large Eddy Simulation (LES) model which is extensively
used in ICF design codes for simulating instabilities in plasma dynamics. This model is based on
the Navier-Stokes equations but with a turbulent viscosity, which depends on the eddy size and
kinetic energy per unit mass. Classical tensor network methods will be explored to compress the
representation of relevant initial data as well as its dynamics. The classical tensor network representations
will be mapped to equivalent quantum tensor networks through a quantum compilation
algorithm. Generalizing these quantum tensor networks to a variational ansatz, we will explore
variational quantum algorithms, and noise-robust encoding to improve the performance on noisy
quantum hardware. Since the LES model is non-linear, we will investigate different approaches
to mapping the dynamics to quantum algorithms such as using multiple copies of the same state,
or linearizing the constituent equations. Finally, by increasing the number of qubits, and thus
the bond-dimension of the encoded tensor networks, we will explore the path towards a quantum
advantage for plasma dynamics simulations on quantum computers of the near future.


Impact
This project will exploit rapid progress in quantum information science to develop algorithms
that will utilize quantum computers of the near future to accelerate simulations that advance
our fundamental understanding of hot dense plasmas for fusion energy sciences. By developing
high-fidelity quantum algorithms for simulating instabilities in plasma dynamics, we will be able
to leverage quantum computers of the near future in concert with classical supercomputing to
explore through simulation some of the most challenging problems for the design of future fusion
experiments.

The proposed work here has potential impact for a variety of DOE interests. The algorithms
developed in this proposal can be repurposed for solving other classes of non-linear partial differential
equations describing a wide variety of phenomenon in nature and are of relevance to DOE
program areas beyond Fusion Energy Sciences. For example, these algorithms could be adapted
to solve dynamical systems in computational fluid dynamics which have applications ranging from
renewable energy generation to large scale climate modeling.



Scroll to top