A new computational framework, called cQUEDA, simulates ultrafast transient-absorption pump-probe spectra of molecular polaritons, with a demonstration on strongly coupled pyrazine. The calculated output includes lower- and upper-polariton states as well as separate contributions to the cavity-QED signal.
The method uses quasi-classical surface-hopping trajectory simulations as its input and turns them into cavity-QED doorway-window signals. For pyrazine, it calculates ground-state bleach (GSB), stimulated emission (SE) and excited-state absorption (ESA), allowing the three parts of the transient-absorption response to be examined separately.
The simulated spectra contain both a lower polariton, or LP, and an upper polariton, or UP. LP has higher intensity than UP, while the separation between the two SE peaks assigned as Rabi splitting is 0.949 eV.
A calculation built on existing trajectories
The study compares four computational settings for pyrazine stimulated-emission signals: no cavity coupling, a long-lived cavity, moderate cavity coupling and a short-lived cavity. The cavity cases show both UP and LP features, extending the same trajectory-based calculation across several cavity regimes.
For the main illustrative resonance calculation, the cavity and pump frequencies were both 5.2 eV. The pulse duration was 5 fs, probe frequencies ran from 3.0 to 7.0 eV in 0.02 eV steps, the cavity loss rate was 0.018 fs−1 and the dipole coupling strength was 0.56 eV.
The implementation is in WaveMixings.jl, and the calculations used Julia 1.13.0-rc3 on an Intel i5-12500H platform with 16 GB of RAM. The authors report that the pyrazine calculations took minutes on modern laptops, while the supporting information directs readers to the repository for the cQUEDA implementation.
Signals that unfold over time
The model also produces a time-dependent picture of the simulated response. In the cavity-QED calculation, the SE signal has a non-radiative decay approximately 6 fs faster than the bare-molecular counterpart, and the simulation lacks wavepacket bifurcation in the B2u nπ state.
Other features appear as the simulated spectra are followed over longer delays. The calculation shows an apparent overlap between GSB and ESA intensities and a decay from ESA to GSB. Signatures connecting UP and LP appear beyond 145 fs, while GSB repopulation occurs within 22 fs.
Despite those separate contributions, the total spectrum is mainly dominated by GSB. Comparing the total and GSB signals with mean squared absolute error gives an approximately 5% difference, consistent with GSB providing most of the calculated total signal.
The boundary of the shortcut
The framework rests on a stated model assumption: wavepacket dynamics during the waiting time are simulated without cavity influence. The paper justifies that treatment when the cavity-mode lifetime is short relative to the molecular-dynamics timescale.
The authors also report a convergence test for the quasi-classical trajectories, saying that the trajectory set yielded spectrally converged results. However, the supplied text does not state the trajectory count.
The document is an arXiv version 1 preprint dated 25 Aug 2026. Its central result is a computational framework for generating and inspecting cavity-QED spectra from simulated trajectory data, demonstrated through pyrazine calculations.
Paper data and sources
Original title: An Ab initio Framework for Simulating Ultrafast Nonlinear Cavity Quantum Electrodynamics Spectra
Authors: Luis Vasquez, Kewei Sun, Haibo Ma, Maxim F. Gelin
Journal/Repository: arXiv
Status: Preprint, not yet peer-reviewed
First online: 2026-08-25
DOI: Not available
Original paper · Full text