Pith. sign in

REVIEW 1 major objections 5 minor 26 references

Loss plus survival conditioning lets partially distinguishable photons bunch more than identical ones in a three-photon circuit.

Reviewed by Pith at T0; open to challenge. T0 means a machine referee read the full paper against a public rubric. the ladder, T0–T4 →

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗

T0 review · deepseek-v4-flash

2026-08-01 02:48 UTC pith:DAK7YASE

load-bearing objection A clean, honest paper that lowers the photon-number threshold for anomalous bunching from seven to three in the survival-conditioned setting; the main caveat is a tiny numerical margin in the key row-normalized claim that needs exact re-checking. the 1 major comments →

arxiv 2607.25306 v1 pith:DAK7YASE submitted 2026-07-28 quant-ph

Loss-induced anomalous generalized bunching in multiphoton interference

Rikizo Ikuta This is my paper
classification quant-ph
keywords generalized bunchingmultiphoton interferencepartial distinguishabilityinternal losssurvival conditioningpermanentlinear optical circuits
verification ladder T0 review T1 audit T2 compute T3 formal T4 reserved

The pith

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

This paper shows that internal loss, when combined with conditioning on photon survival, can activate anomalous generalized bunching in passive linear optics. The central result is that for three photons in a three-mode lossy interferometer, the conditional probability that all photons exit in a target region is maximized at partial distinguishability, not at full indistinguishability—a behavior forbidden for the corresponding unconditioned probability. For two photons, the conditional probability is always monotonic, though a multimode target region can reverse the trend. The anomaly survives row normalization, so it is not merely an artifact of detector-efficiency imbalance.

Core claim

The core discovery is that survival conditioning changes the minimal hierarchy of generalized bunching: for the transfer matrix Tint = R13(π/8)R23(π/4)L(3)F(3) with L(3)=diag(1/√2,1/2,1), the conditional bunching probability B_K^{Tint}(x) for K={1,2} attains a global maximum at x≈0.655, with value ≈0.241, which is larger than the value ≈0.233 at x=1. The paper proves a necessary and sufficient condition for such an anomaly—Δ2>0 and q(1)<0—showing that pairwise-exchange contributions drive the initial rise while cyclic three-photon contributions reverse the trend. The anomaly persists after row normalization and over finite parameter regions.

What carries the argument

The central object is the survival-conditioned bunching probability B_K^T(x)=P_K^T(x)/P_A^T(x), where P_R^T(x)=perm(H_R⊙S(x)). Here H_R is the Gram matrix of transfer columns into region R and S(x) is the one-parameter overlap matrix with x=1 for identical photons. This permanental form decomposes both numerator and denominator as a_R(1+β_R x^2+γ_R x^3), making the derivative controlled by the cubic q(x)=2Δ2+3Δ3x+Δ23x^3. The anomaly occurs exactly when Δ2>0 and q(1)<0, which is impossible for single-mode targets or for the unconditioned N=3 probability due to a determinantal constraint.

Load-bearing premise

The result assumes the permanent formula for partially distinguishable photons in lossy circuits remains valid when loss is internal; it also specializes to equal pairwise overlaps, so it is not proven for arbitrary Gram matrices.

What would settle it

A three-photon experiment on the circuit Tint = R13(π/8)R23(π/4)diag(1/√2,1/2,1)F(3) should show B_K(x) peaking near x≈0.655 above the x=1 value; seeing monotonic behavior or a peak only at x=1 would falsify the claim. Additionally, row-normalizing using single-photon transmission measurements should preserve the interior peak.

Watch this falsifier — get emailed when new claim-graph text bears on it.

If this is right

  • For two-photon inputs, the conditional bunching probability is always monotonic; single-mode targets increase with indistinguishability, while multimode targets can reverse the direction.
  • For three photons with a single-mode target, the conventional trend is proven for any number of accessible modes, and the minimal anomalous setting is N=M=3 with a two-mode target region.
  • The anomaly persists after row normalization, so it cannot be reduced to unequal output loss or detector efficiencies; it reflects the internal nonunitary interference structure.
  • The effect is robust over finite parameter regions, providing a benchmark for testing loss and partial-distinguishability models in photonic hardware.

Where Pith is reading between the lines

These are editorial extensions of the paper, not claims the author makes directly.

  • A natural test would use unequal pairwise overlaps rather than the single-parameter family S(x); if the interior maximum persists for generic Gram matrices, the anomaly is not an artifact of symmetric distinguishability.
  • The ratio mechanism—opposite signs of exchange and cyclic three-photon terms—suggests analogous conditional statistics could reveal other interference features in lossy boson sampling beyond the N=3 minimal case.
  • Extending the analysis to mixed internal states or non-Fock inputs, as the paper suggests, might show whether the anomaly is a generic feature of survival-conditioned multiphoton statistics.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit.

Referee Report

1 major / 5 minor

Summary. The paper studies generalized bunching of partially distinguishable photons in lossy linear optical circuits, conditioning on the event that all photons survive. It introduces the conditional bunching probability B_K^(T)(S)=P_K^(T)(S)/P_A^(T)(S). For N=2 it proves that B_K is monotone in the one-parameter indistinguishability x for any circuit, with single-mode targets increasing and multimode targets possibly decreasing. For N=3, in a minimal three-mode circuit with a two-mode target and internal loss, it exhibits an explicit circuit for which B_K has a strict interior maximum at partial distinguishability, despite the unconditioned P_K being maximized at full indistinguishability by the permanent-on-top theorem. The anomaly is reported to persist after row normalization, and Appendix D gives a necessary and sufficient condition (Δ2>0 and q(1)<0) for such an anomaly.

Significance. If correct, the paper establishes that internal loss plus survival postselection lowers the minimal photon number for anomalous generalized bunching from N=7 in the lossless unconditioned setting to N=3 in the lossy conditional setting. The N=2 monotonicity proof is clean and fully algebraic; the N=3 example is explicit with hand-specified parameters and is supported by finite parameter regions in Appendix C; the necessary and sufficient condition in Appendix D is rigorously proved. The paper also clarifies the distinction between raw lossy HOM signatures and survival-conditioned statistics. These are concrete, reproducible theoretical results that should be of interest to the quantum-optics and multiphoton-interference community.

major comments (1)
  1. [§VI, values after Eq. (21)] The row-normalized anomaly B(T̄_int)(x_max)≈0.271 > B(T̄_int)(1)≈0.270 rests on a margin of 0.001 at three-decimal rounding. This persistence is central to the claim in §VII that the effect is due to internal nonunitary interference rather than output-row-weight imbalance. Please provide exact or higher-precision values (or a code/Colab notebook) for the main parameter set, and report the numerical resolution of the Appendix C scans so the reader can certify that the row-normalized anomaly is not a rounding artifact. Consider also choosing a parameter point with a larger margin for the row-normalized curve.
minor comments (5)
  1. [§IV, Eq. (16)] The permanental formula P_R^(T)(S)=perm(H_R⊙S) is imported for nonunitary transfer matrices. A brief derivation or an explicit statement that it follows by tracing out loss modes in the full unitary model, with precise pointers to Refs. [12,13], would improve accessibility.
  2. [Appendix C] Please specify the grid resolution and the numerical method used to compute Δ(T)_K. The color-bar labels in Fig. 5 are hard to read from the caption; a brief description of the plotted range would help.
  3. [Eq. (22)] The notation Δ23 is easy to confuse with a two-mode index. Consider renaming this coefficient (e.g., Δ_c) or clarifying that it is a combination of β and γ differences.
  4. [§V] When passing from monotonicity in y=x^2 to monotonicity in x, state explicitly that x is nonnegative, so the direction is preserved.
  5. [§VI] The observation that row normalization reduces T=D_ηU to U for M=N=3 is stated tersely; a one-line proof or a pointer to the definition of q_j would make the argument easier to follow.

Circularity Check

0 steps flagged

No significant circularity: the N=3 conditional-bunching anomaly is a self-contained explicit computation from the standard permanent formula, and the row-normalization persistence is supported by contrast circuits.

full rationale

The central claim is derived, not assumed. The conditional bunching probability B_K^(T)(S) is defined in Eq. (8) as P_K/P_A, and P_R is given by the standard permanent formula perm(H_R⊙S) in Eq. (16), imported from the external Refs. [12,13]. For N=3 the paper writes P_R(x)=a_R(1+β_R x^2+γ_R x^3), derives the derivative q(x) in Eq. (22), and proves in Appendix D that an interior maximum is equivalent to Δ2>0 and q(1)<0. This is algebra from the permanent representation; no target maximum is inserted in advance. The reported anomaly is an explicit example with stated circuit parameters θ13=π/8, θ23=π/4, η1=1/2, η2=1/4. These are existence witnesses, not fitted constants, and x is the independent distinguishability variable. The persistence after row normalization is nontrivial and is checked against control circuits: T0=L(3)F(3) is monotone, and Appendix B shows an output-attenuation-induced anomaly disappears under row normalization. Thus the claim that the effect reflects internal nonunitary interference rather than output imbalance is supported by a computed contrast, not by definition. The permanent-on-top obstruction of Ref. [6] is external and acts as a constraint to be bypassed by the new conditional quantity. The only self-citation, Ref. [22], concerns a future extension to nontrivial photon statistics and is not load-bearing. Numerical margin concerns are verification issues, not circularity.

Axiom & Free-Parameter Ledger

2 free parameters · 5 axioms · 0 invented entities

The paper introduces no new physical entities and no new postulates beyond standard linear-optical modeling. The free parameters are explicit circuit parameters chosen as existence witnesses; they are not fit to data. The main load-bearing background is the permanental formula for partially distinguishable photons.

free parameters (2)
  • Internal loss parameters η1, η2 = η1=1/2, η2=1/4 for the three-photon example; η1=1/3, η2=1/6 in Appendix B; η1=η2=1/3 for the two-photon example
    Mode-dependent loss rates chosen by hand to exhibit the anomaly. They are existence witnesses, not fit to external data.
  • Beamsplitter angles θ13, θ23 = θ13=π/8, θ23=π/4 for the three-photon example; θ13=θ23=π/8 for the two-photon reversal example
    Angles chosen within the finite anomaly regions demonstrated in Appendix C. They are not optimized against data.
axioms (5)
  • standard math P_R^(T)(S)=perm(H_R⊙S) for partially distinguishable photons
    The permanental representation of generalized bunching is imported from Refs. [12,13] and used in Eq. (16). It is the backbone of the entire derivation.
  • domain assumption Loss is modeled as coupling to unobserved vacuum modes with T†T≤I, and conditioning is only on final survival
    Eqs. (4)-(5) define the lossy linear-optical model. If loss were internal-state-dependent or if loss modes were monitored, the conditional probability would not take the stated permanent form.
  • domain assumption The one-parameter Gram family S(x)=x+(1-x)I covers the distinguishability behavior of interest
    The paper proves the anomaly for equal pairwise overlaps only (Eq. 14). It does not address arbitrary Gram matrices, leaving open whether the anomaly occurs more broadly.
  • standard math Unconditioned N=3 generalized bunching is monotone, by the permanent-on-top result of Ref. [6]
    Used to establish that the conditioned anomaly bypasses the unconditioned obstruction. This is a prior theorem, not derived in the paper.
  • standard math Positive-semidefiniteness and Hadamard inequality for the 3×3 matrices H_K and H_A
    Used in Appendices A and D to bound βA, γA and to derive the necessary-and-sufficient condition.

pith-pipeline@v1.3.0-alltime-deepseek · 12047 in / 13084 out tokens · 129937 ms · 2026-08-01T02:48:26.309518+00:00 · methodology

0 comments
read the original abstract

We show that internal loss and survival conditioning can activate anomalous generalized bunching in passive linear optical circuits. We introduce a conditional bunching probability that all photons occupy a target region of accessible output modes, given that all photons survive. For two-photon inputs, we prove that this probability is always monotonic for any circuit size and loss configuration, although a multimode target region can reverse the monotonic direction. For three-photon inputs in a minimal three-mode lossy interferometer, we find a nonmonotonic anomaly in which the conditional bunching probability is maximized for partially distinguishable photons. This behavior is forbidden for the corresponding unconditioned target-region probability, demonstrating that survival-conditioned loss changes the minimal hierarchy of generalized bunching.

Figures

Figures reproduced from arXiv: 2607.25306 by Rikizo Ikuta.

Figure 1
Figure 1. Figure 1: FIG. 1. (a) Standard generalized bunching setting. (b) Gen [PITH_FULL_IMAGE:figures/full_fig_p001_1.png] view at source ↗
Figure 2
Figure 2. Figure 2: FIG. 2. The circuit in Eq. (20) with [PITH_FULL_IMAGE:figures/full_fig_p004_2.png] view at source ↗
Figure 3
Figure 3. Figure 3: FIG. 3. (a) Conditional bunching probability for the full cir [PITH_FULL_IMAGE:figures/full_fig_p005_3.png] view at source ↗
Figure 5
Figure 5. Figure 5: FIG. 5. Parameter regions for the generalized bunching [PITH_FULL_IMAGE:figures/full_fig_p007_5.png] view at source ↗

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Reference graph

Works this paper leans on

26 extracted references · 2 linked inside Pith

  1. [1]

    Thus, q′(x) < 0 is satisfied for all 0 ≤ x ≤ 1

    q′(1) < 0 follows from γA > β A − 1, βA ≥ 0, and 0 < 3x∗ − x3 ∗ < 2. Thus, q′(x) < 0 is satisfied for all 0 ≤ x ≤ 1. This means that whenever an interior root exists, q(x) is strictly decreasing on 0 ≤ x ≤ 1, and the root is unique. Since q(0) = 2∆ 2 > 0, the condition q(1) < 0, to- gether with the continuity of q(x), guarantees the exis- tence of an x∗ ∈...

  2. [2]

    Using this relation, we obtain q(x) = ∆ 2(2 + 3x + (1 − dA)x3) − dA(3x + βAx3), (D1) where dA = 1 − βA + γA

    It follows that det(HK)/aK = 1 − βK + γK = 0. Using this relation, we obtain q(x) = ∆ 2(2 + 3x + (1 − dA)x3) − dA(3x + βAx3), (D1) where dA = 1 − βA + γA. Since HA is positive semidef- inite, det(HA) ≥ 0, while Hadamard’s inequality gives det(HA) ≤ aA. Therefore, 0 ≤ dA = det( HA)/aA ≤ 1. Since βA ≥ 0, if ∆2 ≤ 0, Eq. (D1) gives q(x) ≤ 0 for 0 ≤ x ≤ 1. Thu...

  3. [3]

    C. K. Hong, Z. Y. Ou, and L. Mandel, Measurement of subpicosecond time intervals between two photons by interference, Phys. Rev. Lett. 59, 2044 (1987)

    2044

  4. [4]

    R. P. Feynman, R. B. Leighton, and M. Sands, The Feyn- man Lectures on Physics, Vol. III, Chapter 4: Identi- cal Particles , https://www.feynmanlectures.caltech. edu/III_04.html (1965), section 4–3

    1965

  5. [5]

    V. S. Shchesnovich, Universality of generalized bunching and efficient assessment of boson sampling , Phys. Rev. Lett. 116, 123601 (2016)

    2016

  6. [6]

    Seron, L

    B. Seron, L. Novo, and N. J. Cerf, Boson bunching is not maximized by indistinguishable particles , Nature Photon- ics 17, 702 (2023)

    2023

  7. [7]

    R. B. Bapat and V. S. Sunder, On majorization and Schur products, Linear Algebra and its Applications 72, 107 (1985)

    1985

  8. [8]

    Bapat and V

    R. Bapat and V. Sunder, An extremal property of the permanent and the determinant , Linear Algebra and its Applications 76, 153 (1986)

    1986

  9. [9]

    Pioge, B

    L. Pioge, B. Seron, L. Novo, and N. J. Cerf, Anoma- lous bunching of nearly indistinguishable bosons , arXiv preprint arXiv:2308.12226 (2023)

    Pith/arXiv arXiv 2023

  10. [10]

    Pioge, L

    L. Pioge, L. Novo, and N. J. Cerf, A unified frame- work for anomalous boson bunching , arXiv preprint arXiv:2607.19499 (2026)

    Pith/arXiv arXiv 2026

  11. [11]

    Drury, A counterexample to a question of Bapat and Sunder, The Electronic Journal of Linear Algebra 31, 69 (2016)

    S. Drury, A counterexample to a question of Bapat and Sunder, The Electronic Journal of Linear Algebra 31, 69 (2016)

    2016

  12. [12]

    Vest, M.-C

    B. Vest, M.-C. Dheur, É. Devaux, A. Baron, E. Rousseau, J.-P. Hugonin, J.-J. Greffet, G. Messin, and F. Marquier, Anti-coalescence of bosons on a lossy beam splitter , Sci- ence 356, 1373 (2017)

    2017

  13. [13]

    L. Hong, Y. Zhang, Y. Chen, and L. Chen, Loss- Assisted Anomalous Hong-Ou-Mandel Interference Based on Nonunitary Multilayer Graphene , Phys. Rev. Lett. 133, 023601 (2024)

    2024

  14. [14]

    V. S. Shchesnovich, Partial indistinguishability theory for multiphoton experiments in multiport devices , Phys. Rev. A 91, 013844 (2015)

    2015

  15. [15]

    M. C. Tichy, Sampling of partially distinguishable bosons and the relation to the multidimensional permanent , Physical Review A 91, 022316 (2015). 9

    2015

  16. [16]

    Chabaud, D

    U. Chabaud, D. Markham, and A. Sohbi, Quantum ma- chine learning with adaptive linear optics , Quantum 5, 496 (2021)

    2021

  17. [17]

    Spagnolo, D

    N. Spagnolo, D. J. Brod, E. F. Galvão, and F. Sciarrino, Non-linear boson sampling , npj Quantum Information 9, 3 (2023)

    2023

  18. [18]

    Francalanci, G

    T. Francalanci, G. Rodari, E. Caruccio, F. Hoch, G. Car- vacho, T. Giordani, N. Spagnolo, R. Albiero, N. D. Gi- ano, G. Corrielli, et al., Generalized photon-bunching bounds in linear and non-linear quantum optical dynam- ics, Optica 13, 1137 (2026)

    2026

  19. [19]

    Chin and J

    S. Chin and J. Huh, Partial distinguishability as a coher- ence resource in boson sampling , Quantum Information Processing 19, 37 (2020)

    2020

  20. [20]

    A. E. Jones, A. J. Menssen, H. M. Chrzanowski, T. A. W. Wolterink, V. S. Shchesnovich, and I. A. Walmsley, Mul- tiparticle interference of pairwise distinguishable photons , Phys. Rev. Lett. 125, 123603 (2020)

    2020

  21. [21]

    Aaronson and D

    S. Aaronson and D. J. Brod, Bosonsampling with lost photons, Phys. Rev. A 93, 012335 (2016)

    2016

  22. [22]

    García-Patrón, J

    R. García-Patrón, J. J. Renema, and V. Shchesnovich, Simulating boson sampling in lossy architectures , Quan- tum 3, 169 (2019)

    2019

  23. [23]

    A. E. Moylett, R. García-Patrón, J. J. Renema, and P. S. Turner, Classically simulating near-term partially- distinguishable and lossy boson sampling , Quantum Sci- ence and Technology 5, 015001 (2020)

    2020

  24. [24]

    Ikuta, Statistical-noise-enhanced multiphoton interfer- ence, Phys

    R. Ikuta, Statistical-noise-enhanced multiphoton interfer- ence, Phys. Rev. A 113, 063731 (2026)

    2026

  25. [25]

    Khalid, D

    A. Khalid, D. Spivak, B. C. Sanders, and H. de Guise, Permutational symmetries for coincidence rates in mul- timode multiphotonic interferometry , Phys. Rev. A 97, 063802 (2018)

    2018

  26. [26]

    A. E. Jones, S. Kumar, S. D’Aurelio, M. Bayerbach, A. J. Menssen, and S. Barz, Distinguishability and mixed- ness in quantum interference , Phys. Rev. A 108, 053701 (2023)

    2023