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

arXiv:2610.25471

Marginal bunching for \textit{in situ} certification of multiparticle indistinguishability

Valery Shchesnovich

quant-phmath-phphysics.atom-phSubmitted Oct 18, 2026• Announced Oct 20, 2026• 0 citations• Revised Nov 2, 2026 (v2)
v1 · Oct 18, 2026v2 · Nov 2, 2026
arXivHTMLPDF

Abstract

We propose an \textit{in situ} method for certifying the multiparticle indistinguishability of identical bosons or fermions in generic linear interferometers. For particles occupying distinct input ports, the method uses two-particle marginal bunching or anti-bunching probabilities to bound full multiparticle indistinguishability through Aldous' spectral gap theorem. For Haar-random unitary interferometers, the number of experimental runs required to estimate this bound to a prescribed precision, at fixed confidence, is polynomial in the particle number, the interferometer size, and the inverse precision, with high probability over the choice of interferometer. Losses are incorporated without conditioning on particle survival. The method is particularly suited to certifying photon indistinguishability in large-scale interference experiments, including Boson Sampling experiments targeting quantum computational advantage.

AI Summary

The paper asks how to certify full multiparticle indistinguishability directly in the linear interferometer used for an experiment, without estimating exponentially rare output events. It defines marginal bunching as the probability that a uniformly selected subset of input particles is detected within chosen output ports. Two-particle marginals are estimated from number-resolved records using $N(N-1)/[n(n-1)]$, including loss events without postselection. For generic output subsets, two-boson marginal bunching is maximized only by fully indistinguishable bosons; for fermions in distinct inputs, the corresponding minimum certifies indistinguishability. An analogous maximality result holds for three-boson marginals. For particles in distinct input ports, weighted pair-interference strengths define a graph Laplacian $L_\omega$. Aldous’ spectral gap theorem yields the tight bound $\mathcal D\ge1-\binom n2\Delta P/\lambda_2(L_\omega)$, where $\mathcal D$ is the label-state weight in the fully symmetric subspace and $\Delta P$ is the measured defect from ideal bunching or anti-bunching. The required gap is computable from an $n\times n$ matrix despite arbitrary, potentially correlated internal states. Hoeffding concentration gives sampling cost scaling as $O(n^4/[\lambda_2(L_\omega)^2\epsilon^2])$ at fixed confidence. For Haar-random interferometers, inverse-polynomial sensitivity holds with high probability. Uniform survival probability $\eta$ increases this sufficient cost by $\eta^{-4}$. The analysis assumes calibrated interferometer parameters and independent, identically prepared trials. Polynomial sampling guarantees concern Haar-random interferometers, not arbitrary fixed devices. The certificate bounds $\mathcal D$ rather than estimating it exactly; three-cycle information is not yet incorporated into quantitative certification. The exponential comparison with full bunching requires fixed nonvanishing Haar-averaged contrast.

No sign of AI-generated textPossible error

Error found: Equation (Runs) understates the sufficient number of runs for its claimed high-confidence certification by a factor of four: the Appendix C argument requires R>=n^2(n-1)^2 ln(2/alpha)/(2g(L_omega)^2 epsilon^2), whereas the published count guarantees only statistical half-width epsilon.

Written by gpt-6.1-sol from the paper's full text. Automated: it can be wrong.

Comments
21 pages, 1 figure (in color); in this revision: new Section VI introduces a weighted series of subsets and a lower bound based on Aldous' spectral gap theorem, thus removing a connectivity bottleneck; updated fig1 is placed in Section VI

Related papers

Measuring multiparticle indistinguishability with the generalized bunching probability

Shawn Geller, Emanuel H. Knill

The indistinguishability of many bosons undergoing passive linear transformations followed by number basis measurements is fully characterized by the visible state of the bosons. However, measuring all the parameters in the visible state is experimentally demanding. In this work, we seek to perform partial characterization of the visible state by measuring properties of it that are available after randomization. First we study the case where the occupied visible modes are randomly permuted, and second we study the case where Haar random linear optical unitaries are applied. In each case, we find that the generalized bunching probability -- which is the probability that all the input bosons arrive in a given subset of the output modes -- obeys monotonicity with respect to some partial order of distinguishability of the input bosons. As an intermediate result, we show that Lieb's permanental-dominance conjecture for immanants is equivalent to the following statement: for states that are invariant under permutations of the occupied visible modes, the generalized bunching probability is maximized when the bosons are perfectly indistinguishable. We also prove that a consequence of the monotonicity of the generalized bunching probability after Haar averaging is that this average is maximized when the bosons are perfectly indistinguishable. Finally, we discuss applications of our results to thermometry of cold-atom systems.

arXiv:2509.04550•Physical Review A•Feb 3, 2026•5 citations

Optimal interferometric certification of multi-photon indistinguishability

Marco Robbio, Michał Oszmaniec, Nicolas J. Cerf +2 more

Multiphoton indistinguishability is a key resource for photonic quantum technologies, yet its characterization typically relies on resource-intensive methods. In this work, we develop two efficient and experimentally friendly protocols to estimate or bound the fidelity $F_{\mathrm{ind}}$ of an $N$-photon state to the closest perfectly indistinguishable state. The first protocol applies to sources preparing separable states, uses a single Fourier interferometer together with photon-number-resolving detection, and yields tight two-sided bounds on $F_{\mathrm{ind}}$. The second protocol combines randomized implementations of linear-optical interferometers with photon counting, enabling direct estimation of $F_{\mathrm{ind}}$ for arbitrary $N$-photon states. Both protocols can certify $F_{\mathrm{ind}} = 1 - \mathcal{O}(ε)$ using provably optimal $\mathcal{O}(1/ε)$ samples, in contrast to previous approaches which required prior assumptions on the model of partial distinguishability. Our methods, based on a multiphoton generalization of the Hong-Ou-Mandel test, bring the rigorous and operationally meaningful certification of multiphoton indistinguishability within reach of current photonic technologies.

quant-pharXiv:2609.37620•Sep 30, 2026•2 citations

Optimal photonic indistinguishability tests in multimode networks

Niko Viggianiello, Fulvio Flamini, Marco Bentivegna +6 more

Particle indistinguishability is at the heart of quantum statistics that regulates fundamental phenomena such as the electronic band structure of solids, Bose-Einstein condensation and superconductivity. Moreover, it is necessary in practical applications such as linear optical quantum computation and simulation, in particular for Boson Sampling devices. It is thus crucial to develop tools to certify genuine multiphoton interference between multiple sources. Our approach employs the total variation distance to find those transformations that minimize the error probability in discriminating the behaviors of distinguishable and indistinguishable photons. In particular, we show that so-called Sylvester interferometers are near-optimal for this task. By using Bayesian tests and inference, we numerically show that Sylvester transformations largely outperform most Haar-random unitaries in terms of sample size required. Furthermore, we experimentally demonstrate the efficacy of the transformation using an efficient 3D integrated circuits in the single- and multiple-source cases. We then discuss the extension of this approach to a larger number of photons and modes. These results open the way to the application of Sylvester interferometers for optimal assessment of multiphoton interference experiments.

arXiv:1710.03578•中国科学通报:英文版•Oct 24, 2018•22 citations

Universality of Generalized Bunching and Efficient Assessment of Boson Sampling

V. S. Shchesnovich

It is found that identical bosons (fermions) show a generalized bunching (antibunching) property in linear networks: the absolute maximum (minimum) of the probability that all N input particles are detected in a subset of K output modes of any nontrivial linear M-mode network is attained only by completely indistinguishable bosons (fermions). For fermions K is arbitrary; for bosons it is either (i) arbitrary for only classically correlated bosons or (ii) satisfies K≥N (or K=1) for arbitrary input states of N particles. The generalized bunching allows us to certify in a polynomial in N number of runs that a physical device realizing boson sampling with an arbitrary network operates in the regime of full quantum coherence compatible only with completely indistinguishable bosons. The protocol needs only polynomial classical computations for the standard boson sampling, whereas an analytic formula is available for the scattershot version.

arXiv:1509.01561•Physical Review Letters•Mar 22, 2016•63 citations

Universality of boson bunching and fermion antibunching in linear networks

V. S. Shchesnovich

It is found that identical bosons (fermions) show generalized bunching (antibunching) property in linear networks: The absolute maximum (minimum) of probability that all $N$ input particles are detected in a subset of $\mathcal{K}$ output modes of any nontrivial linear $M$-mode network is attained \textit{only} by completely indistinguishable bosons (fermions). For fermions $\mathcal{K}$ is arbitrary, for bosons it is either ($i$) arbitrary for only classically correlated bosons or ($ii$) satisfies $\mathcal{K}\ge N$ (or $\mathcal{K}=1$) for arbitrary input states of $N$ particles. The generalized bunching allows to certify in a \textit{polynomial} in $N$ number of runs that a physical device realizing Boson Sampling with \textit{an arbitrary} network operates in the regime of full quantum coherence compatible \textit{only} with completely indistinguishable bosons. The protocol needs \textit{only polynomial} classical computations for the standard Boson Sampling, whereas an \textit{analytic formula} is available for the scattershot version.

•Sep 4, 2015•0 citations

Complementarity between bosonic and fermionic many-body interferences with partially distinguishable particles

Marco Robbio, Michael G. Jabbour, Nicolas J. Cerf

It is well known that bosons and fermions exhibit opposite behaviors when experiencing interference, in the sense that bosons have a tendency to bunch whereas fermions have a tendency to antibunch. Recently, this complementarity was mathematically characterized in [arXiv:2312.17709] by means of an identity relating the transition probabilities of both types of particles in a linear interferometer. Here, we show that such a complementarity still holds even when particles become partially distinguishable, for example, when they have slightly different polarizations or time delays. Namely, we establish a relation that combines bosonic and fermionic multiparticle interferences in an arbitrary linear interferometer, in the presence of partial distinguishability. Incidentally, this also provides a new mathematical identity relating the permanent and determinant of tensors of order 3. Importantly, this complementarity has direct operational consequences in quantum metrology. Indeed, we show that the correlation matrices for bosonic and fermionic particle number distributions at the output of the interferometer obey a simple sum rule: their sum equals twice the correlation matrix for classical particles. This, in turn, constraints the achievable quantum Fisher information in phase-estimation protocols, highlighting a trade-off whereby greater indistinguishability enhances bosonic sensitivity whereas reduced indistinguishability can benefit fermionic schemes.

quant-pharXiv:2604.23316•Apr 28, 2026•3 citations
We propose an \textit{in situ} method for certifying the multiparticle indistinguishability of identical bosons or fermions in generic linear interferometers. For particles occupying distinct input ports, the method uses two-particle marginal bunching or anti-bunching probabilities to bound full multiparticle indistinguishability through Aldous' spectral gap theorem. For Haar-random unitary interferometers, the number of experimental runs required to estimate this bound to a prescribed precision, at fixed confidence, is polynomial in the particle number, the interferometer size, and the inverse precision, with high probability over the choice of interferometer. Losses are incorporated without conditioning on particle survival. The method is particularly suited to certifying photon indistinguishability in large-scale interference experiments, including Boson Sampling experiments targeting quantum computational advantage.
The paper asks how to certify full multiparticle indistinguishability directly in the linear interferometer used for an experiment, without estimating exponentially rare output events. It defines marginal bunching as the probability that a uniformly selected subset of input particles is detected within chosen output ports. Two-particle marginals are estimated from number-resolved records using N(N−1)/[n(n−1)]N(N-1)/[n(n-1)]N(N−1)/[n(n−1)], including loss events without postselection.
For generic output subsets, two-boson marginal bunching is maximized only by fully indistinguishable bosons; for fermions in distinct inputs, the corresponding minimum certifies indistinguishability. An analogous maximality result holds for three-boson marginals. For particles in distinct input ports, weighted pair-interference strengths define a graph Laplacian LωL_\omegaLω​. Aldous’ spectral gap theorem yields the tight bound D≥1−(n2)ΔP/λ2(Lω)\mathcal D\ge1-\binom n2\Delta P/\lambda_2(L_\omega)D≥1−(2n​)ΔP/λ2​(Lω​), where D\mathcal DD is the label-state weight in the fully symmetric subspace and ΔP\Delta PΔP is the measured defect from ideal bunching or anti-bunching. The required gap is computable from an n×nn\times nn×n matrix despite arbitrary, potentially correlated internal states.
Hoeffding concentration gives sampling cost scaling as O(n4/[λ2(Lω)2ϵ2])O(n^4/[\lambda_2(L_\omega)^2\epsilon^2])O(n4/[λ2​(Lω​)2ϵ2]) at fixed confidence. For Haar-random interferometers, inverse-polynomial sensitivity holds with high probability. Uniform survival probability η\etaη increases this sufficient cost by η−4\eta^{-4}η−4.
The analysis assumes calibrated interferometer parameters and independent, identically prepared trials. Polynomial sampling guarantees concern Haar-random interferometers, not arbitrary fixed devices. The certificate bounds D\mathcal DD rather than estimating it exactly; three-cycle information is not yet incorporated into quantitative certification. The exponential comparison with full bunching requires fixed nonvanishing Haar-averaged contrast.
The indistinguishability of many bosons undergoing passive linear transformations followed by number basis measurements is fully characterized by the visible state of the bosons. However, measuring all the parameters in the visible state is experimentally demanding. In this work, we seek to perform partial characterization of the visible state by measuring properties of it that are available after randomization. First we study the case where the occupied visible modes are randomly permuted, and second we study the case where Haar random linear optical unitaries are applied. In each case, we find that the generalized bunching probability -- which is the probability that all the input bosons arrive in a given subset of the output modes -- obeys monotonicity with respect to some partial order of distinguishability of the input bosons. As an intermediate result, we show that Lieb's permanental-dominance conjecture for immanants is equivalent to the following statement: for states that are invariant under permutations of the occupied visible modes, the generalized bunching probability is maximized when the bosons are perfectly indistinguishable. We also prove that a consequence of the monotonicity of the generalized bunching probability after Haar averaging is that this average is maximized when the bosons are perfectly indistinguishable. Finally, we discuss applications of our results to thermometry of cold-atom systems.
Multiphoton indistinguishability is a key resource for photonic quantum technologies, yet its characterization typically relies on resource-intensive methods. In this work, we develop two efficient and experimentally friendly protocols to estimate or bound the fidelity FindF_{\mathrm{ind}}Find​ of an NNN-photon state to the closest perfectly indistinguishable state. The first protocol applies to sources preparing separable states, uses a single Fourier interferometer together with photon-number-resolving detection, and yields tight two-sided bounds on FindF_{\mathrm{ind}}Find​. The second protocol combines randomized implementations of linear-optical interferometers with photon counting, enabling direct estimation of FindF_{\mathrm{ind}}Find​ for arbitrary NNN-photon states. Both protocols can certify Find=1−O(ε)F_{\mathrm{ind}} = 1 - \mathcal{O}(ε)Find​=1−O(ε) using provably optimal O(1/ε)\mathcal{O}(1/ε)O(1/ε) samples, in contrast to previous approaches which required prior assumptions on the model of partial distinguishability. Our methods, based on a multiphoton generalization of the Hong-Ou-Mandel test, bring the rigorous and operationally meaningful certification of multiphoton indistinguishability within reach of current photonic technologies.
Particle indistinguishability is at the heart of quantum statistics that regulates fundamental phenomena such as the electronic band structure of solids, Bose-Einstein condensation and superconductivity. Moreover, it is necessary in practical applications such as linear optical quantum computation and simulation, in particular for Boson Sampling devices. It is thus crucial to develop tools to certify genuine multiphoton interference between multiple sources. Our approach employs the total variation distance to find those transformations that minimize the error probability in discriminating the behaviors of distinguishable and indistinguishable photons. In particular, we show that so-called Sylvester interferometers are near-optimal for this task. By using Bayesian tests and inference, we numerically show that Sylvester transformations largely outperform most Haar-random unitaries in terms of sample size required. Furthermore, we experimentally demonstrate the efficacy of the transformation using an efficient 3D integrated circuits in the single- and multiple-source cases. We then discuss the extension of this approach to a larger number of photons and modes. These results open the way to the application of Sylvester interferometers for optimal assessment of multiphoton interference experiments.
It is found that identical bosons (fermions) show a generalized bunching (antibunching) property in linear networks: the absolute maximum (minimum) of the probability that all N input particles are detected in a subset of K output modes of any nontrivial linear M-mode network is attained only by completely indistinguishable bosons (fermions). For fermions K is arbitrary; for bosons it is either (i) arbitrary for only classically correlated bosons or (ii) satisfies K≥N (or K=1) for arbitrary input states of N particles. The generalized bunching allows us to certify in a polynomial in N number of runs that a physical device realizing boson sampling with an arbitrary network operates in the regime of full quantum coherence compatible only with completely indistinguishable bosons. The protocol needs only polynomial classical computations for the standard boson sampling, whereas an analytic formula is available for the scattershot version.
It is found that identical bosons (fermions) show generalized bunching (antibunching) property in linear networks: The absolute maximum (minimum) of probability that all NNN input particles are detected in a subset of K\mathcal{K}K output modes of any nontrivial linear MMM-mode network is attained \textit{only} by completely indistinguishable bosons (fermions). For fermions K\mathcal{K}K is arbitrary, for bosons it is either (iii) arbitrary for only classically correlated bosons or (iiiiii) satisfies K≥N\mathcal{K}\ge NK≥N (or K=1\mathcal{K}=1K=1) for arbitrary input states of NNN particles. The generalized bunching allows to certify in a \textit{polynomial} in NNN number of runs that a physical device realizing Boson Sampling with \textit{an arbitrary} network operates in the regime of full quantum coherence compatible \textit{only} with completely indistinguishable bosons. The protocol needs \textit{only polynomial} classical computations for the standard Boson Sampling, whereas an \textit{analytic formula} is available for the scattershot version.
It is well known that bosons and fermions exhibit opposite behaviors when experiencing interference, in the sense that bosons have a tendency to bunch whereas fermions have a tendency to antibunch. Recently, this complementarity was mathematically characterized in [arXiv:2312.17709] by means of an identity relating the transition probabilities of both types of particles in a linear interferometer. Here, we show that such a complementarity still holds even when particles become partially distinguishable, for example, when they have slightly different polarizations or time delays. Namely, we establish a relation that combines bosonic and fermionic multiparticle interferences in an arbitrary linear interferometer, in the presence of partial distinguishability. Incidentally, this also provides a new mathematical identity relating the permanent and determinant of tensors of order 3. Importantly, this complementarity has direct operational consequences in quantum metrology. Indeed, we show that the correlation matrices for bosonic and fermionic particle number distributions at the output of the interferometer obey a simple sum rule: their sum equals twice the correlation matrix for classical particles. This, in turn, constraints the achievable quantum Fisher information in phase-estimation protocols, highlighting a trade-off whereby greater indistinguishability enhances bosonic sensitivity whereas reduced indistinguishability can benefit fermionic schemes.