Permutation-symmetric quantum trajectories

Elliot W. Lloyd, Aleksandra A. Ziolkowska, Jonathan Keeling · 2026

Read Permutation-symmetric quantum trajectories by Elliot W. Lloyd, Aleksandra A. Ziolkowska, Jonathan Keeling (2026) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2605-11103.txt.

Introduction

Elliot W. Lloyd, Aleksandra A. Ziolkowska, and Jonathan Keeling show how to unravel quantum trajectories while respecting weak permutation symmetry for models of many emitters coupled to a shared system, such as a cavity mode. Such models underlie lasing, superradiance, and Dicke-model phase transitions, but exact dynamics grow expensive fast.

For two-level emitters their symmetry-respecting unraveling cuts cost from fifth power to second power in emitter count, and refinements reach linear scaling, greatly widening the simulable size range. The method extends to d-level systems with cost scaling as N to the power d(d-1)/2, enabling large runs for three-level cases. Cavity-QED researchers get exact dynamics for far larger emitter groups.

Catalog record

CREATORElliot W. Lloyd, Aleksandra A. Ziolkowska, Jonathan Keeling
YEAR2026
COLLECTIONPAPERS
FORMATTXT
FILE SIZE50680
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2605.11103

Browse the full library · About the Tower of Babel

Full text

The opening of the work, from the library's hosted text:

Permutation-symmetric quantum trajectories Elliot W. Lloyd, Aleksandra A. Ziolkowska, Jonathan Keeling Abstract We show how one may perform a stochastic unraveling which respects weak permutation symmetry for models of N N emitters coupled to a common system (e.g. a cavity mode). For problems involving 2 2 -level emitters, such an unravelling reduces the computational cost from 𝒪 ⁡ ( N 5 ) \mathcal{O}(N^{5}) to 𝒪 ⁡ ( N 2 ) \mathcal{O}(N^{2}) , and with additional refinements, allows reduction to 𝒪 ⁡ ( N ) \mathcal{O}(N) . This significantly increases the range of system sizes for which one can model exact quantum dynamics of such systems. We further show how the method can also be applied to d d -level systems, with computational effort scaling as 𝒪 ⁡ ( N d ⁡ ( d − 1 ) / 2 ) \mathcal{O}(N^{d(d-1)/2}) , and we show it allows large- N N simulations for d = 3 d=3 . Many paradigmatic models of cavity QED involve N emitters coupled to a common cavity mode. Such models arise in contexts including simple models of lasing Haken 1970 ; Scully and Zubairy 1997 , superradiant dynamics Bonifacio and Preparata 1970 , superradiant phase transitions of the Dicke model Hepp and Lieb 1973 ; Wang and Hioe 1973 , and single-mode models of photon condensation Abouelela et al. 2025 . Recently there has been increasing interest in versions of such models where the emitters are more complex, with multiple electronic states, or dressed by vibrational modes Ćwik et al. 2014 ; Galego et al. 2015 ; Herrera and Spano 2016 . In the most general case, the cost of simulating such models scales exponentially with N , however, when the emitters are identical this cost reduces significantly. If all terms in the model (the Hamiltonian and all dissipators) involve only sums of single emitter terms then one has strong permutation symmetry and may rewrite the model in terms of collective operators. For two-level emitters, these are total spin operators, and the the simulation cost reduces to \mathcal{O}(N^{2}N_{c}^{2}) (where N_{c} is the truncation of cavity occupation). If there are also individual dissipation terms for the two-level emitters, the model still has weak Buča and Prosen 2012 ; Albert and Jiang 2014 permutation symmetry. This prevents using collective spins, but one can use permutation symmetric density matrix approaches with a cost \mathcal{O}(N^{3}N_{c}^{2}) . Such approaches have been applied to many of the models listed above Chase and Geremia 2008 ; Xu et al. 2013 ; Richter et al. 2015 ; Kirton and Keeling 2018 ; Shammah et al. 2018 ; Freter et al. 2025 . While simulating the density matrix evolution can provide most properties of interest for a Markovian open quantum system—i.e. operator expectations and correlations including multi-time correlations—there has recently been much interest in the properties that can only be found from the dynamics of individual quantum trajectories rather than their ensemble average Cao et al. 2019 ; Li et al. 2018 ; Skinner et al. 2019 ; Choi et al. 2020 ; Gullans and Huse 2020 . Quantum trajectories Gisin and Percival 1992 ; Dalibard et al. 1992 ; Dum et al. 1992 ; Carmichael 1993 ; Plenio and Knight 1998 ; Daley 2014 can be thought of as simulating individual experimental runs, and thus the state of the system conditioned on a measurement record. Different forms of measurement (e.g. photon counting vs homodyne measurement) lead to different forms of quantum trajectory, such as the quantum jump formalism vs the quantum state diffusion equations Plenio and Knight 1998 ; Daley 2014 . Different measurement apparatus correspond to different forms of “stochastic unravelling”, and thus different forms of entanglement within the trajectories, even though the ensemble-averaged density matrix may be the same Vovk and Pichler 2022 . Quantum trajectories also provide a route to cheaper numerical simulations, as one simulates a wavefunction with a size given by the Hilbert space dimension, d_{H} , rather than the density matrix size d_{H}^{2} . For problems with weak permutation symmetry—i.e. involving individual dissipation terms—direct applications of stochastic unravelling causes a problem. Individual dissipation of the emitters means that measurements corresponding to that dissipation make the emitter states inequivalent: a naive unravelling would require trajectories that break the permutation symmetry. Rather than reducing the computational time, such unravelling increases it back to exponential in N . In this paper we show that when formulated in terms of collective spin representations, one may perform a stochastic unravelling of the master equation that holds within the space of permutation symmetric density matrices. Considering emitters as two-level systems (2LS), this approach leads to a computational cost that grows only linearly with system size, \mathcal{O}(NN_{c}) , the same scaling as would occur in the absence of individual dissipation. We further show that the factor N_{c} (which typically needs to be of order N_{c}\propto N ) can be eliminated. This means a reduction from \mathcal{O}(N^{5}) for density matrix simulations to \mathcal{O}(N) for the approaches we describe here. This enables massively larger simulations than were previously possible with individual dissipation. We further show how the method can be extended beyond 2LS, to d -level systems, allowing large quantum simulations of more complex models. We consider models consisting of i=1\ldots N 2LS coupled to a common system (e.g. a bosonic mode). We write the general case as follows: \displaystyle\hat{H} \displaystyle=\sum_{i=1}^{N}(\hat{h}_{i}+\sum_{\alpha}\hat{x}_{i,\alpha}\hat{X}_{c,\alpha})+\hat{H}_{c}, (1) \displaystyle\partial_{t}\rho \displaystyle=-i[\hat{H},\rho]+\sum_{i,\alpha}\mathcal{D}[\hat{\ell}_{i,\alpha}]+\sum_{\alpha}\mathcal{D}[\hat{C}_{\alpha}] (2) where \mathcal{D}[\hat{X}]=\hat{X}\rho\hat{X}^{\dagger}-(1/2)\{\hat{X}^{\dagger}\hat{X},\rho\} . Here \hat{h}_{i} is the individual Hamiltonian for the i th 2LS, \hat{x}_{i,\alpha} is an operator of type \alpha for that 2LS which couples to an operator of type \alpha of the common bosonic mode \hat{X}_{c,\alpha} . \hat{H}_{c} is the Hamiltonian for the common mode. The master equation also involves individual dissipation terms \hat{\ell}_{i,\alpha} for the emitters and common bosonic mode dissipation terms \hat{C}_{\alpha} . It is important to note that to ensure permutation symmetry, we consider only homogeneous systems, i.e. not subject to disorder. As a concrete example (discussed further below), the Dicke model has \hat{h}_{i}=\omega_{0}\sigma^{z}_{i},\hat{H}_{c}=\omega_{c}\hat{a}^{\dagger}\hat{a} and a single type of coupling \hat{x}_{i,1}=\sigma^{x}_{i},\hat{X}_{c,1}=(g/\sqrt{N})(\hat{a}+\hat{a}^{\dagger}) . Adding dephasing and loss for the 2LS requires two individual dissipation terms, \hat{\ell}_{i,1}=\sqrt{\Gamma_{\phi}}\sigma^{z}_{i},\hat{\ell}_{i,2}=\sqrt{\Gamma_{\downarrow}}\sigma^{-}_{i} , while cavity photon loss requires \hat{C}_{1}=\sqrt{\kappa}\hat{a} . We start from the approach to weak permutation symmetry based on collective spin states labeled by total spin J and z-magnetization M discussed in Refs. Chase and Geremia 2008 ; Barberena 2025 . As described there, one can write a permutation symmetric density matrix for the cavity-emitter system in the form: \rho=\sum_{J,M,M^{\prime}}\sum_{n_{c},n_{c}^{\prime}}\rho^{n_{c},n_{c}^{\prime}}_{J,M,M^{\prime}}|n_{c}\rangle\langle n_{c}^{\prime}|\otimes\overline{|J,M\rangle\langle J,M^{\prime}|}. (3) Here |n_{c}\rangle denotes a Fock state for the cavity; for clarity, we will suppress the cavity state in the following discussion and restore it when summarizing our main result. The notation for the spin represents three important points. (1) The labels J,M do not in general uniquely specify a unique state of the N -spin system; one needs an additional label, which we denote T , to specify how the individual spins are combined to create the total spin (see discussion on d -level case in SM for details). (2) Permutation symmetry means that the density matrix weights \rho_{J,M,M^{\prime}} do not depend on this label T . (3) Both the label J and the extra label T must be the

Read the complete text