Ab Initio Polariton Transport Dynamics with the Classical Path Approximation

From the paper

Why atomistic polariton transport is expensive

On-the-fly excited-state gradients for every molecule become prohibitive in a large cavity ensemble. The paper combines quantum exciton-photon propagation with precomputed ground-state Born-Oppenheimer trajectories, arguing that collective delocalization suppresses each molecule’s excited-state force correction.

From the paper

The classical path approximation

For a local harmonic bath coordinate, the full mean-field and CPA forces have the structure

If the wavepacket is delocalized over an effective number of molecules $N_{\mathrm{eff}}$, then $|c_n|^2\sim1/N_{\mathrm{eff}}$ and the omitted local correction becomes small. The inverse participation ratio provides a practical diagnostic:

From the paper

From model validation to BODIPY transport

1. Propagate exciton and photon amplitudes with Lindblad mean-field dynamics.
2. Validate CPA against full mean-field forces in a generalized HTC model using velocity, MSD, populations, and spatial wavepackets.
3. Generate ground-state molecular trajectories and electronic properties for immobilized BODIPY molecules.
4. Feed time-dependent excitation energies and dipoles into the quantum propagator.
5. Compute transmission, polariton/dark density, and transient MSD, then compare their trends with experiment.

1. Ground-state pathsObtain affordable nuclear trajectories and molecular observables.
2. Quantum propagationEvolve the coupled exciton-photon wavepacket with loss.
3. Adaptive checkRequest local excited-state gradients only where population becomes large.

Simplified educational example

How delocalization controls the omitted force

import numpy as np

x = np.arange(400)
for width in (1.5, 8.0, 30.0, 100.0):
    p = np.exp(-0.5*((x-200)/width)**2)
    p /= p.sum()
    N_eff = 1.0/np.sum(p**2)
    omitted = p.max()  # in units of |C|
    print(width, N_eff, omitted)

From the paper

Reported validation and application

• CPA and full-force model simulations agree closely for group velocity, transient MSD, populations, and spatial wavepackets across the tested losses and bath couplings.
• The agreement is explained by collective delocalization rather than treated as an empirical accident.
• The atomistic BODIPY calculation reproduces the cavity anti-crossing and qualitative transient transport behavior seen experimentally.
• The paper proposes an adaptive extension that evaluates excited-state gradients only for molecules whose local excitation probability exceeds a threshold.

Scope and breakdown conditions

When CPA may fail

• Localized initial states, strong static disorder, dark-state localization, or bond-localized photochemistry can make $|c_n|^2$ large.
• The study is restricted to a single excitation; higher manifolds increase force corrections with excitation number.
• Derivative couplings and nonradiative relaxation terms are neglected for the ultrafast transport window.
• The atomistic electronic structure is semiempirical, molecular centers are fixed, and only five atomistic realizations are propagated.
• Nuclear motion remains classical and the comparison to experiment is mainly qualitative or semiquantitative.