26 July 2026 to 1 August 2026
University of Maryland, College Park
US/Eastern timezone

H$^2$MC: A Hybrid Hamiltonian Monte Carlo framework for interacting quantum wires

28 Jul 2026, 17:10
20m
Benjamin Banneker B (Adele H. Stamp Student Union)

Benjamin Banneker B

Adele H. Stamp Student Union

3972 Campus Dr, College Park, MD 20742
Contributed talk Algorithms and artificial intelligence Algorithms and artificial intelligence

Speaker

Finn Temmen (Forschungszentrum Jülich, IAS-4)

Description

Stochastic methods are indispensable tools for studying strongly correlated fermionic systems due to their far more favorable volume scaling than direct approaches such as exact diagonalization (ED) or tensor network methods. However, this improved scaling comes at the cost of new challenges, most notably the sign problem and long autocorrelation times, which severely restrict the accessible parameter space.

In this talk, we consider a 2D system of interacting quantum wires and show how partially retaining ED in Hamiltonian Monte Carlo (HMC) simulations mitigates these limitations while preserving scaling advantages over full ED. Specifically, we demonstrate that this Hybrid Hamiltonian Monte Carlo framework, dubbed H$^2$MC, significantly alleviates the sign problem and shortens autocorrelation times compared to pure HMC formulations based on real and imaginary Hubbard-Stratonovich transformations. We further show how incorporating pseudofermion analogues into this framework yields additional computational speedups while simultaneously reducing memory requirements.

Together, these results illustrate how combining seemingly disparate numerical methods can overcome limitations that are otherwise intrinsic to each approach individually.

Authors

Finn Temmen (Forschungszentrum Jülich, IAS-4) Martina Gisti (University of Bonn)

Co-authors

Prof. David Luitz (University of Bonn) Johann Ostmeyer (University Bonn) Prof. Thomas Luu (Forshungszentrum Jülich)

Presentation materials