Scalable Distributed Memory Implementation of the Quasi-Adiabatic Propagator Path Integral

Area of Science:

• Quantum dynamics
• Computational physics
• Chemical physics

Background:

• Simulating quantum dynamics in non-Markovian environments is computationally challenging.
• Structured environments with sharp resonances cause long-time system-bath correlations.
• Existing methods struggle with memory bottlenecks for these complex systems.

Purpose of the Study:

• To present a scalable, distributed memory implementation of the MACGIC-QUAPI method.
• To address memory limitations using a premerging algorithm and hash-based look-up (hMACGIC-QUAPI).
• To enable accurate simulations of dissipative quantum dynamics in structured non-Markovian environments.

Main Methods:

• Developed a distributed memory implementation using MPI for path spreading.
• Implemented efficient path management with a hash map for constant access time.
• Utilized mask-assisted coarse graining of influence coefficients (MACGIC)-quasi-adiabatic propagator path integral (QUAPI).

Main Results:

• The hMACGIC-QUAPI method demonstrates scalability and preserves numerical accuracy.
• Simulations reveal resonance splitting and sideband emergence due to strong system-environment interactions.
• The method accurately captures non-Markovian system-bath correlations, outperforming perturbative approaches.

Conclusions:

• The hMACGIC-QUAPI method offers a versatile and efficient solution for simulating complex quantum dynamics.
• It overcomes memory bottlenecks, enabling large-scale studies of systems with structured non-Markovian environments.
• The open-source implementation facilitates broader research in quantum dynamics and condensed matter physics.