arXiv:2609.37620
Optimal interferometric certification of multi-photon indistinguishability
Marco Robbio, Michał Oszmaniec, Nicolas J. Cerf, Ernesto Galvão, Leonardo Novo
Abstract
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.
AI Summary
The paper asks how to certify collective $N$-photon indistinguishability without resolving internal degrees of freedom or reconstructing the internal state. It considers inputs with one photon in each occupied spatial mode, allowing arbitrary internal correlations. The fidelity to the closest perfectly indistinguishable state is shown to be $F_{\mathrm{ind}}(\rho)=\operatorname{Tr}(\rho\Pi_{\mathrm{sym}})$. Its defect $1-F_{\mathrm{ind}}$ equals the trace distance of the reduced external state from the ideal input, bounding deviations of all internal-insensitive measurement statistics. Two linear-optical protocols use photon-number-resolving detection. A fixed $N$-mode Fourier interferometer measures the cyclic-invariant weight $P_c$ through a modular condition on output occupations. For arbitrary states, $F_{\mathrm{ind}}\leq P_c$; for separable states, $\max\{0,2P_c-1\}\leq F_{\mathrm{ind}}$. For product states with $P_c=1-\varepsilon$, a refined bound determines the fidelity up to a second-order gap, $F_{\mathrm{ind}}\geq1-\varepsilon-\frac{N-2}{2(N-1)}\varepsilon^2+\mathcal{O}(\varepsilon^3)$. A second protocol directly estimates the fidelity for arbitrary states by sampling permutations and implementing Fourier interferometers on their disjoint cycles. The constructions use internal–external permutation identities, overlap inequalities and representation-theoretic bounds. Bernstein concentration gives $\mathcal{O}(\varepsilon^{-1}\log\delta^{-1})$ samples for near-perfect certification at confidence $1-\delta$, independent of $N$; binary state discrimination establishes optimal scaling. The paper also gives product-state counterexamples to positive-partition distinguishability models after permutation twirling. The analysis assumes lossless optics and ideal number-resolving detectors. The randomized protocol requires shot-to-shot reconfiguration, while fixed-interferometer lower bounds require separability. Loss, detector imperfections, higher input occupations and continuous-variable inputs remain untreated.
Additionally, these models were used for polishing the narrative and spellchecking throughout the paper.
Error found: The external-mode/internal-state intertwining lemma incorrectly maps every mode permutation to the inverse internal permutation, although correcting this leaves the protocol averages and real-valued scores intact. The appendix's advertised sufficient conditions for comparing the cyclic and HOM lower bounds are not sufficient, as demonstrated by a four-photon product-state counterexample. The Bargmann-invariant table reports values inconsistent with its ceiling headings, including 2 instead of 3 for N=3 and 8 instead of 9 for N=4.
Written by gpt-6.1-sol from the paper's full text. Automated: it can be wrong.
- Comments
- 25 pages, 2 figures