Challenges
Datasets
Workspaces
Papers
Discussions
Leaderboard
Log inSign up
Challenges
Datasets
Workspaces
Papers
Discussions
Leaderboard
Blog
Job Board
Q3AS

© 2026 Aqora Quantum S.A.S.

TermsPrivacyLegal Notice

Paired Quantum Trajectories: Common Random Numbers for Protocol Selection in Continuously Monitored Quantum Systems

Jean-Jacques Dubray

Published Aug 31, 2026• 0 citations
Publisher

Abstract

A quantum jump trajectory is a discrete-event process in the literal sense: coherent evolution between jumps, a waiting time drawn from the decay of the no-jump norm, a channel selection at the jump. We take this correspondence in both directions. First, we build a monitored quantum laboratory (a driven emitter, detectors with efficiency and dead time, a click-driven feedback controller) inside a verification-first discrete-event engine that machine-checks the event structure and the bookkeeping on every transition, validated against exact finite-horizon renewal expectations and against QuTiP. Second, we import a comparison discipline that is standard in discrete-event simulation and that we did not find in the quantum trajectory literature: common random numbers. Addressing the randomness by event ordinal keeps two protocol arms on identical draws until their event histories diverge, and a mechanical audit verifies the pairing rather than assuming it. Pairing pays in proportion to trajectory alignment: for close protocol variants, where the selection decision is hard, the variance of the estimated protocol gap falls by a factor of 40 or more (replicated across five seeds), and the paired design resolves a 0.7% rate difference at $n = 400$ trajectories that the unpaired analysis of the same data cannot. Against the Fisher information of the click record, that paired design recovers about a tenth of the achievable precision at the count endpoint and the unpaired design under two percent, with the remaining gap attributable to the endpoint rather than the pairing. Two physics results follow. Dead-time losses for antibunched light sit well below the textbook Poisson correction, an effect reaching 21% that we compute exactly for resonance fluorescence and reproduce within the measurement intervals. And the rate-optimal feedback pulse after a click is not a $\pi$ pulse, and not because of latency: the mean wait is exactly $a_0 + a_1\cos\theta + a_2\sin\theta$ for every drive, detuning and latency, which puts the zero-latency optimum at our drive at exactly $3\pi/4$ and the deficit of the $\pi$ pulse at exactly $\sqrt{2} - 1$, both confirmed by measurement at the shortest latency we can represent. Latency lowers the optimum further, and $\pi$ is recovered only as the drive weakens. The optimum does not maximize post-pulse excitation, because the waiting time depends on coherence with the drive, not population alone.

DOI
10.20944/preprints202608.2297.v2

Related papers

Unravelling quantum jumps by watching the fluorescence of a qubit

Philippe Campagne-Ibarcq, Pierre Six, S. Jézouin +6 more

When the main source of qubit relaxation comes from its coupling to a photonic channel, each relaxation event is associated with the release of a photon. The corresponding discrete quantum jumps of the qubit state can be observed by counting the number of photons emitted by fluorescence. This discreteness of the quantum jumps is in fact related to the nature of the light detector. What does the evolution of the qubit state become if fluorescence is measured using a heterodyne detector instead? In this talk, an experiment will be discussed, in which the records of heterodyne measurements of fluorescence is used to reconstruct the quantum trajectories of a qubit during relaxation. Using a large number of experiments, it is shown that these trajectories can be used to better predict the probability to find given measurement outcomes during the evolution. This heterodyne measurement of the fluorescence is a quantum demolition continuous measurement, which is very different from the more common dispersive measurement sensitive to qubit occupation. These trajectories are thus expected to exhibit some exotic properties, particularly when using past and future knowledge. Besides, using measurement based feedback based on the fluorescence signal alone, it is possible to stabilize any chosen qubit state. This work thus demonstrates that relaxation into an efficiently monitored channel is not a limit for quantum information protocols but can instead be a resource.

•Mar 2, 2015•0 citations

A New Horizon in Open Quantum Systems: The Phase-Dependent Jump Rate

Saadat Samadi Dinani

We propose that the quantum-jump (decay) rate of an open quantum system can depend on the relative phase δ = θ − φ between the system coherence and a reference oscillation, taking the form Γ(δ) = Γ₀ exp(α cos δ) ≈ Γ₀[1 + α cos δ]. The resulting distribution of jump phases is a von Mises distribution whose first Fourier coefficient is M₁ = I₁(α)/I₀(α) ≈ α/2. Standard Lindblad theory employs a constant rate γ₀; this work replaces it with a phase-dependent rate inside the Lindblad dissipator. The operational definitions are explicit: θ = arg ρ_eg is the phase of the system coherence in a fixed reference frame, and φ is the phase of a classical reference oscillation (a drive tone or a phase-lockable bath mode). The relative phase δ is therefore defined only when such a reference exists — pure Fock-state (photon or phonon number) jumps carry no phase information and cannot test the hypothesis. A two-dimensional metric on the torus (θ, φ), g = [[1, −cos δ], [−cos δ, 1]], is offered solely as geometric motivation for the cos δ form. Its Hessian is singular at δ = nπ, but this singularity does not by itself imply an enhanced decay rate, and it cannot select the sign of α. The step from "Legendre failure" to "rate modulation" is an explicit postulate, not a derivation. This work does not claim to solve the measurement problem. A reproducible injection–recovery test with null controls (4000 trajectories per replicate, seed 2026) recovers injected values of α to within statistical error. For the linear law, the estimator α̂ = 2⟨cos δ⟩ is exactly unbiased with standard error √(2/N); for the exponential (von Mises) law, a maximum-likelihood estimator is obtained by solving M₁ = I₁(α)/I₀(α). Under the null (α = 0), the maximum |M₁| observed across 1500 independent replicates is 0.046, with 95th and 99th percentiles of 0.027 and 0.035. Power analysis indicates that a 3σ detection requires N ≈ 2(3/α)² jumps and a 5σ detection requires N ≈ 2(5/α)² jumps. These simulations validate only the statistical estimator: because α was injected by construction, they are not evidence for the physical hypothesis. The manuscript states the platform requirements for a real test (a driven qubit or oscillator with an explicit classical reference phase, measurable system phase from continuous I/Q records, and trajectory-resolved jump detection), the pre-registration requirements, the null controls, and the rival explanations that must be excluded (phase-dependent detection efficiency, ordinary drive-phase dependence, population-dependent rates, and intrinsic heterodyne detector phase response). No experimental result is claimed. The link between the singular metric and the rate modulation is postulated rather than derived; the phases θ and φ must be tied to a concrete Hamiltonian and reference in each physical system; and the sign of α remains a free parameter of the hypothesis. Keywords: open quantum systems; quantum jumps; phase-dependent decay; von Mises distribution; Lindblad equation; quantum trajectories; injection–recovery; null controls; estimator validation

•Jul 30, 2026•0 citations

To Catch and Reverse a Quantum Jump Mid-Flight (In Russian)

Zlatko Minev, Shantanu Mundhada, Shyam Shankar

Quantum physics was invented to account for two fundamental features of measurement results their discreetness and randomness. Emblematic of these features is the Bohr idea of quantum jumps between two discrete energy levels of an atom. Experimentally, quantum jumps were first observed in an atomic ion driven by a weak deterministic force while under strong continuous energy measurement. The times at which the discontinuous jump transitions occur are reputed to be fundamentally unpredictable. Can there be, despite the indeterminism of quantum physics, a possibility to know if a quantum jump is about to occur or not? Here, we answer this question affirmatively by experimentally demonstrating that the jump from the ground to an excited state of a superconducting artificial three-level atom can be tracked as it follows a predictable flight, by monitoring the population of an auxiliary energy level coupled to the ground state. The experimental results demonstrate that the jump evolution when completed is continuous, coherent, and deterministic. Furthermore, exploiting these features and using real-time monitoring and feedback, we catch and reverse a quantum jump mid-flight, thus deterministically preventing its completion. Our results, which agree with theoretical predictions essentially without adjustable parameters, support the modern quantum trajectory theory and provide new ground for the exploration of real-time intervention techniques in the control of quantum systems, such as early detection of error syndromes.

•Jul 1, 2019•0 citations

Parameter estimation for quantum jump unraveling

Marco Radaelli, Joseph A. Smiga, Gabriel Teixeira Landi +1 more

We consider the estimation of parameters encoded in the measurement record of a continuously monitored quantum system in the jump unraveling, corresponding to a single-shot scenario, where information is continuously gathered. Here, it is generally difficult to assess the precision of the estimation procedure via the Fisher Information due to intricate temporal correlations and memory effects. In this paper we provide a full set of solutions to this problem. First, for multi-channel renewal processes we relate the Fisher Information to an underlying Markov chain and derive a easily computable expression for it. For non-renewal processes, we introduce a new algorithm that combines two methods: the monitoring operator method for metrology and the Gillespie algorithm which allows for efficient sampling of a stochastic form of the Fisher Information along individual quantum trajectories. We show that this stochastic Fisher Information satisfies useful properties related to estimation on a single run. Finally, we consider the case where some information is lost in data compression/post-selection and provide tools for computing the Fisher Information in this case. All scenarios are illustrated with instructive examples from quantum optics and condensed matter.

arXiv:2402.06556•Quantum•Feb 2, 2026•19 citations

Patterns in the jump-channel statistics of open quantum systems

Gabriel Teixeira Landi

A continuously measured quantum system with multiple jump channels gives rise to a stochastic process described by random jump times and random emitted symbols, representing each jump channel. While much is known about the waiting time distributions, very little is known about the statistics of the emitted symbols. In this letter we fill in this gap. First, we provide a full characterization of the resulting stochastic process, including efficient ways of simulating it, as well as determining the underlying memory structure. Second, we show how to unveil patterns in the stochastic evolution: Some systems support closed patterns, wherein the evolution runs over a finite set of states, or at least recurring states. But even if neither is possible, we show that one may still cluster the states approximately, based on their ability to predict future outcomes. We illustrate these ideas by studying transport through a boundary-driven one-dimensional XY spin chain.

arXiv:2305.07957•May 13, 2023•11 citations

Non-Markovian Quantum Trajectories

Michael W. Jack

The technique of quantum trajectories (stochastic Schrödinger equations or Monte Carlo wave functions) for open systems is generalized to the non-Markovian regime. I consider a microscopic model of an open system consisting of a boson field coupled linearly (with an excitation preserving coupling) to a localized system. The model allows for a field with an arbitrary dispersion relation and an arbitrary mode-dependent coupling to the system. The trajectories are formulated as continuous measurements of the output field from the system. For a general dispersive field these measurements must be distributed in space for this formulation to be possible. The result of this formulation is a non-Markovian equation for the system conditioned on the measurements. A method of numerically simulating this equation has been determined and implemented in some test cases. Numerical simulation is possible if one can introduce a finite memory time for the evolution of the reduced system. As an illustration, the method is applied to the spectral detection of the emission from a driven two-level atom and also to an atom radiating into an electromagnetic field where the free space modes of the electromagnetic field are altered by the presence of a cavity. In both cases the non-Markovian behaviour arises from the uncertainty in the time of emission of a photon that is later detected (or reabsorbed), although, in the second case, the non-Markovian behaviour is intrinsic to the system environment coupling whereas, in the spectral detection case, it is a consequence of the choice of measurement process. The generalization of the techniques of quantum trajectories to the non-Markovian regime promises to make a range of open system problems where the Born-Markov approximation is invalid tractable to numerical simulation.

•Jan 1, 1999•3 citations
A quantum jump trajectory is a discrete-event process in the literal sense: coherent evolution between jumps, a waiting time drawn from the decay of the no-jump norm, a channel selection at the jump. We take this correspondence in both directions. First, we build a monitored quantum laboratory (a driven emitter, detectors with efficiency and dead time, a click-driven feedback controller) inside a verification-first discrete-event engine that machine-checks the event structure and the bookkeeping on every transition, validated against exact finite-horizon renewal expectations and against QuTiP. Second, we import a comparison discipline that is standard in discrete-event simulation and that we did not find in the quantum trajectory literature: common random numbers. Addressing the randomness by event ordinal keeps two protocol arms on identical draws until their event histories diverge, and a mechanical audit verifies the pairing rather than assuming it. Pairing pays in proportion to trajectory alignment: for close protocol variants, where the selection decision is hard, the variance of the estimated protocol gap falls by a factor of 40 or more (replicated across five seeds), and the paired design resolves a 0.7% rate difference at n=400n = 400n=400 trajectories that the unpaired analysis of the same data cannot. Against the Fisher information of the click record, that paired design recovers about a tenth of the achievable precision at the count endpoint and the unpaired design under two percent, with the remaining gap attributable to the endpoint rather than the pairing. Two physics results follow. Dead-time losses for antibunched light sit well below the textbook Poisson correction, an effect reaching 21% that we compute exactly for resonance fluorescence and reproduce within the measurement intervals. And the rate-optimal feedback pulse after a click is not a π\piπ pulse, and not because of latency: the mean wait is exactly a0+a1cos⁡θ+a2sin⁡θa_0 + a_1\cos\theta + a_2\sin\thetaa0​+a1​cosθ+a2​sinθ for every drive, detuning and latency, which puts the zero-latency optimum at our drive at exactly 3π/43\pi/43π/4 and the deficit of the π\piπ pulse at exactly 2−1\sqrt{2} - 12​−1, both confirmed by measurement at the shortest latency we can represent. Latency lowers the optimum further, and π\piπ is recovered only as the drive weakens. The optimum does not maximize post-pulse excitation, because the waiting time depends on coherence with the drive, not population alone.
When the main source of qubit relaxation comes from its coupling to a photonic channel, each relaxation event is associated with the release of a photon. The corresponding discrete quantum jumps of the qubit state can be observed by counting the number of photons emitted by fluorescence. This discreteness of the quantum jumps is in fact related to the nature of the light detector. What does the evolution of the qubit state become if fluorescence is measured using a heterodyne detector instead? In this talk, an experiment will be discussed, in which the records of heterodyne measurements of fluorescence is used to reconstruct the quantum trajectories of a qubit during relaxation. Using a large number of experiments, it is shown that these trajectories can be used to better predict the probability to find given measurement outcomes during the evolution. This heterodyne measurement of the fluorescence is a quantum demolition continuous measurement, which is very different from the more common dispersive measurement sensitive to qubit occupation. These trajectories are thus expected to exhibit some exotic properties, particularly when using past and future knowledge. Besides, using measurement based feedback based on the fluorescence signal alone, it is possible to stabilize any chosen qubit state. This work thus demonstrates that relaxation into an efficiently monitored channel is not a limit for quantum information protocols but can instead be a resource.
We propose that the quantum-jump (decay) rate of an open quantum system can depend on the relative phase δ = θ − φ between the system coherence and a reference oscillation, taking the form Γ(δ) = Γ₀ exp(α cos δ) ≈ Γ₀[1 + α cos δ]. The resulting distribution of jump phases is a von Mises distribution whose first Fourier coefficient is M₁ = I₁(α)/I₀(α) ≈ α/2. Standard Lindblad theory employs a constant rate γ₀; this work replaces it with a phase-dependent rate inside the Lindblad dissipator. The operational definitions are explicit: θ = arg ρ_eg is the phase of the system coherence in a fixed reference frame, and φ is the phase of a classical reference oscillation (a drive tone or a phase-lockable bath mode). The relative phase δ is therefore defined only when such a reference exists — pure Fock-state (photon or phonon number) jumps carry no phase information and cannot test the hypothesis. A two-dimensional metric on the torus (θ, φ), g = [[1, −cos δ], [−cos δ, 1]], is offered solely as geometric motivation for the cos δ form. Its Hessian is singular at δ = nπ, but this singularity does not by itself imply an enhanced decay rate, and it cannot select the sign of α. The step from "Legendre failure" to "rate modulation" is an explicit postulate, not a derivation. This work does not claim to solve the measurement problem. A reproducible injection–recovery test with null controls (4000 trajectories per replicate, seed 2026) recovers injected values of α to within statistical error. For the linear law, the estimator α̂ = 2⟨cos δ⟩ is exactly unbiased with standard error √(2/N); for the exponential (von Mises) law, a maximum-likelihood estimator is obtained by solving M₁ = I₁(α)/I₀(α). Under the null (α = 0), the maximum |M₁| observed across 1500 independent replicates is 0.046, with 95th and 99th percentiles of 0.027 and 0.035. Power analysis indicates that a 3σ detection requires N ≈ 2(3/α)² jumps and a 5σ detection requires N ≈ 2(5/α)² jumps. These simulations validate only the statistical estimator: because α was injected by construction, they are not evidence for the physical hypothesis. The manuscript states the platform requirements for a real test (a driven qubit or oscillator with an explicit classical reference phase, measurable system phase from continuous I/Q records, and trajectory-resolved jump detection), the pre-registration requirements, the null controls, and the rival explanations that must be excluded (phase-dependent detection efficiency, ordinary drive-phase dependence, population-dependent rates, and intrinsic heterodyne detector phase response). No experimental result is claimed. The link between the singular metric and the rate modulation is postulated rather than derived; the phases θ and φ must be tied to a concrete Hamiltonian and reference in each physical system; and the sign of α remains a free parameter of the hypothesis. Keywords: open quantum systems; quantum jumps; phase-dependent decay; von Mises distribution; Lindblad equation; quantum trajectories; injection–recovery; null controls; estimator validation
Quantum physics was invented to account for two fundamental features of measurement results their discreetness and randomness. Emblematic of these features is the Bohr idea of quantum jumps between two discrete energy levels of an atom. Experimentally, quantum jumps were first observed in an atomic ion driven by a weak deterministic force while under strong continuous energy measurement. The times at which the discontinuous jump transitions occur are reputed to be fundamentally unpredictable. Can there be, despite the indeterminism of quantum physics, a possibility to know if a quantum jump is about to occur or not? Here, we answer this question affirmatively by experimentally demonstrating that the jump from the ground to an excited state of a superconducting artificial three-level atom can be tracked as it follows a predictable flight, by monitoring the population of an auxiliary energy level coupled to the ground state. The experimental results demonstrate that the jump evolution when completed is continuous, coherent, and deterministic. Furthermore, exploiting these features and using real-time monitoring and feedback, we catch and reverse a quantum jump mid-flight, thus deterministically preventing its completion. Our results, which agree with theoretical predictions essentially without adjustable parameters, support the modern quantum trajectory theory and provide new ground for the exploration of real-time intervention techniques in the control of quantum systems, such as early detection of error syndromes.
We consider the estimation of parameters encoded in the measurement record of a continuously monitored quantum system in the jump unraveling, corresponding to a single-shot scenario, where information is continuously gathered. Here, it is generally difficult to assess the precision of the estimation procedure via the Fisher Information due to intricate temporal correlations and memory effects. In this paper we provide a full set of solutions to this problem. First, for multi-channel renewal processes we relate the Fisher Information to an underlying Markov chain and derive a easily computable expression for it. For non-renewal processes, we introduce a new algorithm that combines two methods: the monitoring operator method for metrology and the Gillespie algorithm which allows for efficient sampling of a stochastic form of the Fisher Information along individual quantum trajectories. We show that this stochastic Fisher Information satisfies useful properties related to estimation on a single run. Finally, we consider the case where some information is lost in data compression/post-selection and provide tools for computing the Fisher Information in this case. All scenarios are illustrated with instructive examples from quantum optics and condensed matter.
A continuously measured quantum system with multiple jump channels gives rise to a stochastic process described by random jump times and random emitted symbols, representing each jump channel. While much is known about the waiting time distributions, very little is known about the statistics of the emitted symbols. In this letter we fill in this gap. First, we provide a full characterization of the resulting stochastic process, including efficient ways of simulating it, as well as determining the underlying memory structure. Second, we show how to unveil patterns in the stochastic evolution: Some systems support closed patterns, wherein the evolution runs over a finite set of states, or at least recurring states. But even if neither is possible, we show that one may still cluster the states approximately, based on their ability to predict future outcomes. We illustrate these ideas by studying transport through a boundary-driven one-dimensional XY spin chain.
The technique of quantum trajectories (stochastic Schrödinger equations or Monte Carlo wave functions) for open systems is generalized to the non-Markovian regime. I consider a microscopic model of an open system consisting of a boson field coupled linearly (with an excitation preserving coupling) to a localized system. The model allows for a field with an arbitrary dispersion relation and an arbitrary mode-dependent coupling to the system. The trajectories are formulated as continuous measurements of the output field from the system. For a general dispersive field these measurements must be distributed in space for this formulation to be possible. The result of this formulation is a non-Markovian equation for the system conditioned on the measurements. A method of numerically simulating this equation has been determined and implemented in some test cases. Numerical simulation is possible if one can introduce a finite memory time for the evolution of the reduced system. As an illustration, the method is applied to the spectral detection of the emission from a driven two-level atom and also to an atom radiating into an electromagnetic field where the free space modes of the electromagnetic field are altered by the presence of a cavity. In both cases the non-Markovian behaviour arises from the uncertainty in the time of emission of a photon that is later detected (or reabsorbed), although, in the second case, the non-Markovian behaviour is intrinsic to the system environment coupling whereas, in the spectral detection case, it is a consequence of the choice of measurement process. The generalization of the techniques of quantum trajectories to the non-Markovian regime promises to make a range of open system problems where the Born-Markov approximation is invalid tractable to numerical simulation.