REVIEW 2 major objections 5 minor 42 references
Large speed-up of quantum emitter detection via quantum interference
T0 review · 2 major / 5 minor · reviewed 2026-08-16 · deepseek-v4-flash
Pith's one-line read Extended Hong-Ou-Mandel interference, analyzed with optimal Bayesian testing, detects quantum emitters orders of magnitude faster than direct photon counting, with the advantage growing under loss and background noise.
desk verdict Plausible and useful extension of HOM statistics to Bayesian emitter detection, but the speed-up numbers rest on a log-normal approximation validated only for direct measurement, and the main text's likelihood ratio is inverted. read the letter →
The pith
A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.
The reading
What carries the argument
The load-bearing object is the derived joint photon-count distribution for extended Hong-Ou-Mandel interference, Eq. (2): $p_{jk}(\xi)$, the probability of seeing $j$ and $k$ photons at the two interferometer outputs after tracing over loss and convolving Poissonian background noise. Its non-Poissonian terms—one phase-sensitive term proportional to $\cos\theta\sqrt{(1-\xi)/\xi}\,(j-k)$ and one phase-insensitive term proportional to $-\eta\epsilon\bar{n}_c (j-k)^2/\bar{n}^2$—create the HOM anti-bunching that distinguishes emitter-present from emitter-absent statistics. Around this distribution the paper builds a Bayesian decision rule: each measurement contributes a log-likelihood $\ln\lambda_{jk}$, the overall log-likelihood $\ln\Lambda$ is a sum of independent variables, and the central limit theorem turns $\Lambda$ into a log-normal random variable, giving the closed-form confidence expression, Eq. (5). That formula is what allows the paper to convert protocol parameters into the required measurement count $N_{2\sigma}$ without brute-force simulation.
What would settle it
Run the paper's own trajectory simulation for coherent and incoherent HOM at the parameters of Figs. 2 and 3 (for example $\eta=0.8$–$0.9$, $\bar{n}_e=\bar{n}_i=1$, with optimized $\bar{n}_c$), tallying the fraction of simulated experiments that reach the correct decision after exactly the predicted $N_{2\sigma}$ rounds; if that fraction deviates from 95.4%, the log-normal confidence formula is over- or under-optimistic for those protocols.
Extended reading notes
Core claim
On its own terms, the paper's central discovery is that the anti-bunching signature of extended Hong-Ou-Mandel interference encodes the presence of a quantum emitter far more efficiently per measurement than the intensity bump seen in direct detection. The author derives the full joint photon-number distribution for a coherent superposition state $\sqrt{1-\xi}|0\rangle+\sqrt{\xi}|1\rangle$ interfering with a coherent state, including mode overlap $\epsilon$, detection efficiency $\eta$, Poissonian background, and detector noise, and shows that the phase-sensitive interference term is linear in $\eta$ while the phase-insensitive anti-bunching term is quadratic. Feeding these distributions into an optimal Bayesian likelihood-ratio test, with the overall log-likelihood treated as log-normal via the central limit theorem, yields a closed-form confidence expression; inverting it at 95.4% confidence gives the number of measurements needed. The paper reports that coherent HOM beats direct measurement by several orders of magnitude across broad parameter ranges, that the speed-up improves as background noise and loss increase, that simple photon counters still give order-of-magnitude gains, and that even incoherent fluorescent emission is detected faster whenever noise is high enough and detectors do not saturate.
Load-bearing premise
The calculations assume that the accumulated evidence, measured on a logarithmic scale, follows a bell-curve distribution after enough measurements; the paper checks this assumption explicitly only for direct photon counting, not for the interference protocols whose speed-ups are quoted.
Editorial extensions
If this is right
- Coherent HOM detection reaches two-sigma (95.4%) confidence in several orders of magnitude fewer measurement rounds than direct photon counting across wide ranges of loss and background noise.
- The speed-up grows as detection efficiency falls and background noise rises, so the protocol is most advantageous exactly where direct detection struggles most.
- Simple saturating photon counters, including single-photon click/no-click detection, retain order-of-magnitude speed-ups for coherent HOM when the coherent field brightness is chosen appropriately.
- Incoherent, fluorescence-like emission can still be detected faster than direct measurement with saturation-free detectors and sufficient noise, though the margin is about one order of magnitude and disappears at low noise or with saturation.
- Because the statistics are derived with realistic noise and loss included, the protocol is claimed to be implementable with existing HOM-microscopy technology.
Reading between the lines
- If the log-normal approximation holds for the interference protocols as it does for direct counting, the same closed-form confidence machinery could be exported to other two-mode interference metrology tasks, such as phase estimation or photonic state tomography, where a likelihood ratio is the natural decision statistic.
- The linear-in-loss phase-sensitive term suggests coherent HOM detection could retain an advantage in ultra-low-efficiency settings such as deep-tissue fluorescence imaging, where direct intensity measurements are photon-starved; this extrapolates beyond the parameter ranges plotted.
- A camera-based HOM microscope imaging a known sparse emitter sample could provide a direct experimental test: the paper's predicted speed-up should appear as far fewer frames needed to reach the same per-pixel certainty as direct counting.
- Applying the same Bayesian ratio test to other non-classical fields, such as squeezed vacuum or two-photon emission, would test whether this loss-and-noise robustness generalizes beyond zero-one photon superpositions.
Signed reviews
Editorial analysis
A structured set of objections, weighed in public.
Referee Report
Summary. The paper proposes using extended Hong-Ou-Mandel (HOM) interference between a quantum emitter's fluorescence and a coherent field, together with Bayesian hypothesis testing on full photon-counting statistics, to detect the presence of a quantum emitter much faster than direct photon counting. The authors derive the two-detector photon-number distribution including loss, background noise, detector noise, and imperfect mode overlap, and use it to define per-shot likelihood ratios. Assuming the log-normal approximation for the distribution of the cumulative log-likelihood ratio, they derive a closed-form expression for the confidence as a function of the number of measurements, invert it to obtain the number of measurements needed for 95.4% confidence, and compute speed-up factors relative to direct measurement. They report that coherent HOM provides order-of-magnitude or larger speed-ups over a wide parameter range, that the advantage improves with increasing loss and background noise, and that incoherent HOM also helps in certain regimes.
Significance. If the quantitative claims hold, this is a valuable proposal: it identifies a realistic regime in which quantum interference plus photon-number-resolving detection outperforms direct detection for a practically important task, and it does so without fitting parameters to data or requiring idealized conditions. The supplementary derivation of the photon-counting statistics is detailed and the zero-diagonal HOM limit is checked. The central quantitative results, however, rest on two assumptions that are not fully supported as written: (i) the main text's likelihood-ratio definition is inconsistent with the Bayesian formulas it uses, and (ii) the log-normal approximation underlying Eq. (5) is validated only for direct measurement, not for the HOM protocols where all the reported speed-up numbers are computed. These issues are fixable but require additional validation and correction before the paper's quantitative conclusions can be accepted.
major comments (2)
- [Main text, 'Bayesian hypothesis testing' (Eq. (3) and Eq. (5))] The likelihood-ratio convention is internally inconsistent. The main text defines λjk = pjk(ξ)/pjk(0), but Eq. (3), Pe = 1/(1+Λ), and the decision rule Pe > 0.5 (equivalent to Λ < 1) are correct only for the inverse convention λjk = pjk(0)/pjk(ξ) used in Eq. (S19) of the supplement. With the main-text convention, Eq. (5) would predict confidence below 1/2 for the parameters studied. Since the reported N2-σ and speed-up values are evidently computed with the supplement's convention, the main text must be corrected and the two definitions aligned; otherwise the paper's central quantitative claims cannot be reproduced from the written equations.
- [Eqs. (4)-(5), Figs. 2-3, and Supplementary Fig. S2] The log-normal approximation for lnΛ is load-bearing: Eq. (5) is the basis for every N2-σ and speed-up value in Figs. 2 and 3. The validation in Fig. S2 and Fig. 1c is performed only for direct measurement. No finite-N simulation of log-likelihood distributions is reported for coherent or incoherent HOM. The HOM per-shot log-likelihood has a tail that grows like -2 ln|j-k| for large count imbalances (from Eq. (2)), and the distribution's skewness at the N values used in Figs. 2 and 3 is not demonstrated to be negligible. The mean-Pe check in Fig. 1c is a different functional (an average of 1/(1+Λ)) and does not validate the tail integral P(Λ<1). The assertion that higher noise randomises the distribution faster and thus accelerates convergence is plausible but unsupported for HOM. The authors should add finite-N simulations of the lnΛ distribution, or at minimum of the confidence tail, for both HOM variants at representative high-loss, high-background parameters. Without this, the quantitative speed-up factors may shift materially.
minor comments (5)
- [Abstract] The phrase 'this suggest that' should read 'this suggests that'.
- [Supplementary Material, Sections S1 and S3] The word 'Poissionian' is consistently misspelled; it should be 'Poissonian'.
- [Supplementary Fig. S4 caption] The word 'lelels' should be 'levels', and the axes are labelled 'nb' while the text refers to ¯ne; please make the notation consistent.
- [Main text, Ref. [31] / Note on Wilks' Theorem] The note that Wilks' Theorem gives a χ2 distribution for this problem is questionable, since the comparison here is between two simple hypotheses rather than nested models; please justify or remove the remark.
- [Main text, Eq. (1)] The mode ordering in the ket |jkpq⟩ is not defined until the supplement; please define it explicitly in the main text for readability.
Circularity Check
No significant circularity: all reported speed-ups are derived from stated quantum-optical assumptions and closed-form photon statistics, with no fitted parameters or self-citation chains bearing the central claim.
full rationale
The paper's derivation chain is self-contained. The central photon-counting distribution for extended HOM, Eq. (2), is derived from an explicit input state and linear optics propagation in the supplementary material (Eqs. S41–S60), not imported from prior work by fiat. The likelihood ratio is then defined from these distributions, and the confidence expression Eq. (5) follows from the central limit theorem plus a log-normal approximation for the overall likelihood ratio. No parameter is fitted to any subset of data; the coherent amplitude is optimized as a legitimate control parameter, and the speed-up factors are obtained by numerically inverting the derived confidence relation. The log-normal approximation is validated only for direct measurement in the supplement, but that is a correctness or robustness concern, not circularity: the approximation is an explicit, testable assumption, not an input secretly equivalent to the output. The paper cites prior work for the existence of extended HOM interference and its experimental feasibility, but the quantitative prediction of speed-up does not reduce to those citations. There is no self-definitional step, no fitted input renamed as a prediction, no load-bearing self-citation, no imported uniqueness theorem, and no renaming of a known empirical pattern. The derivation stands on its own equations, and the reported predictions are genuine consequences of the stated model.
Assumptions & free parameters
free parameters (3)
- emission probability xi =
0.1
- mode overlap epsilon =
0.9
- coherent state mean photon number nbar_c =
optimized per scenario
assumptions (5)
- domain assumption The emitter field is described by |e> = sqrt(1-xi)|0> + sqrt(xi)|1>, with a fixed phase relative to the coherent local oscillator for coherent HOM.
- domain assumption Background emission, detector dark counts, and un-overlapped coherent light are independent Poissonian noise processes whose means add.
- standard math The distribution of the log-likelihood ratio is log-normal for large N via the Central Limit Theorem.
- domain assumption Equal Bayesian priors and a decision threshold at P(e)=0.5 define the detection confidence.
- domain assumption Incoherent emission is fully described by averaging over the phase theta, equivalent to setting cos(theta)=0.
Cite this review
Pith. "Pith review of Large speed-up of quantum emitter detection via quantum interference." pith.science (2026). https://pith.science/paper/JCRO62ND
@misc{pith2026250500950,
author = {Pith},
title = {Pith review of: Large speed-up of quantum emitter detection via quantum interference},
year = {2026},
howpublished = {\url{https://pith.science/paper/JCRO62ND}},
note = {Machine review of arXiv:2505.00950}
}
read the original abstract
Quantum emitters are a key resource in quantum technologies, microscopy, and other applications. The ability to rapidly detect them is useful both for quality control in engineered emitter arrays and for high-contrast imaging of naturally occurring emitters. Using full photon-counting statistics and optimal Bayesian hypothesis testing, we show that extended Hong-Ou-Mandel interference between quantum emission and a coherent field enables orders-of-magnitude speed-ups in emitter detection under realistic noise and loss. Strikingly, the performance advantage improves as loss and background noise increase, and persists for incoherent emission. Taken together with prior demonstrations of extended Hong-Ou-Mandel interference, this suggest that substantial performance gains are achievable with current technology under realistic, non-ideal conditions. This offers a new approach to fast, low-intensity imaging and for emitter characterization in large-scale quantum systems. Fundamentally, the discovery that quantum interference and measurements, used together, are more robust to both loss and noise than standard measurement techniques opens the possibility of broad applications across quantum metrology.
Figures
Reference graph
Works this paper leans on
-
[1]
Aharonovich, D
I. Aharonovich, D. Englund, and M. Toth, Nature Pho- tonics 10, 631 (2016)
2016
-
[2]
H. Defienne, W. P. Bowen, M. Chekhova, G. B. Lemos, D. Oron, S. Ramelow, N. Treps, and D. Faccio, Nature Photonics 18, 1024 (2024)
work page 2024
-
[3]
G. Vicidomini, P. Bianchini, and A. Diaspro, Nature Methods 15, 173 (2018)
work page 2018
-
[4]
B. Li, C. Wu, M. Wang, K. Charan, and C. Xu, Nature Methods 17, 163 (2020)
work page 2020
-
[5]
S. P. Fodor, Science 277, 393 (1997)
work page 1997
-
[6]
PsiQuantum Team, Nature https://doi.org/10.1038/s41586-025-08820-7 (2025)
-
[7]
Azuma, S
K. Azuma, S. E. Economou, D. Elkouss, P. Hilaire, L. Jiang, H.-K. Lo, and I. Tzitrin, Reviews of Modern Physics 95, 045006 (2023)
2023
-
[8]
H. Deschout, F. C. Zanacchi, M. Mlodzianoski, A. Di- 6 aspro, J. Bewersdorf, S. T. Hess, and K. Braeckmans, Nature Methods 11, 253 (2014)
work page 2014
Show all 42 references
-
[9]
M. J. Rust, M. Bates, and X. Zhuang, Nature Methods 3, 793 (2006)
2006
-
[10]
Birrittella, J
R. Birrittella, J. Mimih, and C. C. Gerry, Phys. Rev. A 86, 063828 (2012)
2012
-
[11]
P. M. Alsing, R. J. Birrittella, C. C. Gerry, J. Mimih, and P. L. Knight, Phys. Rev. A 105, 013712 (2022)
2022
-
[12]
P. M. Alsing, R. J. Birrittella, C. C. Gerry, J. Mimih, and P. L. Knight, (2024), The Hong-Ou-Mandel effect is really odd, arXiv:2410.11800 (2024)
2024 arXiv
-
[13]
Rarity, P
J. Rarity, P. Tapster, and R. Loudon, Journal of Optics B: Quantum and Semiclassical Optics 7, S171 (2005)
2005
-
[14]
Lyons, V
A. Lyons, V. Zickus, R. ´Alvarez-Mendoza, D. Triggiani, V. Tamma, N. Westerberg, M. Tassieri, and D. Faccio, Nature Communications 14, 8005 (2023)
2023
-
[15]
Eaton, A
M. Eaton, A. Hossameldin, R. J. Birrittella, P. M. Alsing, C. C. Gerry, H. Dong, C. Cuevas, and O. Pfister, Nature Photonics 17, 106 (2023)
2023
-
[16]
Cheng, Y
R. Cheng, Y. Zhou, S. Wang, M. Shen, T. Taher, and H. X. Tang, Nature Photonics 17, 112 (2023)
2023
-
[17]
Madonini, F
F. Madonini, F. Severini, F. Zappa, and F. Villa, Ad- vanced Quantum Technologies 4, 2100005 (2021)
2021
-
[18]
Hong, Z.-Y
C.-K. Hong, Z.-Y. Ou, and L. Mandel, Physical Review Letters 59, 2044 (1987)
1987
-
[19]
Loredo, C
J. Loredo, C. Ant´ on, B. Reznychenko, P. Hilaire, A. Harouri, C. Millet, H. Ollivier, N. Somaschi, L. De Santis, A. Lemaˆ ıtre,et al. , Nature Photonics 13, 803 (2019)
2019
-
[20]
Maillette de Buy Wenniger, S
I. Maillette de Buy Wenniger, S. Wein, D. Fioretto, S. Thomas, C. Ant´ on-Solanas, A. Lemaitre, I. Sagnes, A. Harouri, N. Belabas, N. Somaschi,et al., Optica Quan- tum 2, 404 (2024)
2024
-
[21]
Eshun, B
A. Eshun, B. Gu, O. Varnavski, S. Asban, K. E. Dorf- man, S. Mukamel, and T. Goodson III, Journal of the American Chemical Society 143, 9070 (2021)
2021
-
[22]
Ndagano, H
B. Ndagano, H. Defienne, D. Branford, Y. D. Shah, A. Lyons, N. Westerberg, E. M. Gauger, and D. Faccio, Nature Photonics 16, 384 (2022)
2022
-
[23]
Y. Li, M. Baba, and M. Matsuoka, Physical Review A 55, 3177 (1997)
1997
-
[24]
Jachura and R
M. Jachura and R. Chrapkiewicz, Optics Letters40, 1540 (2015)
2015
-
[25]
C. A. Casacio, L. S. Madsen, A. Terrasson, M. Waleed, K. Barnscheidt, B. Hage, M. A. Taylor, and W. P. Bowen, Nature 594, 201 (2021)
2021
-
[26]
Kielpinski, C
D. Kielpinski, C. Monroe, and D. J. Wineland, Nature 417, 709 (2002)
2002
-
[27]
J. C. Waters, Journal of Cell Biology 185, 1135 (2009)
2009
-
[28]
Piekarek, D
M. Piekarek, D. Bonneau, S. Miki, T. Yamashita, M. Fu- jiwara, M. Sasaki, H. Terai, M. G. Tanner, C. M. Natara- jan, R. H. Hadfield, et al., Optics Letters 42, 815 (2017)
2017
-
[29]
Supplementary Information: Large speed-up of quantum emitter detection via extended Hong–Ou–Mandel inter- ference
-
[30]
I. J. Good, Weight of evidence and the (Bayesian) likeli- hood ratio, in The Use Of Statistics In Forensic Science (CRC Press, 1991) Chap. 3
1991
-
[31]
(4) in our high-measurement-number limit
Note: by Wilks’ Theorem, a χ2 distribution is generally more accurate at low measurement numbers, but agrees with Eq. (4) in our high-measurement-number limit
-
[32]
3 b there is a maximum in N2−σ at ¯nc ∼ 0.42
Note: for incoherent HOM with saturation-free detectors, in Fig. 3 b there is a maximum in N2−σ at ¯nc ∼ 0.42. A vacuum local oscillator would outperform a coherent local oscillator that had lower mean photon number
-
[33]
Polacchi, F
B. Polacchi, F. Hoch, G. Rodari, S. Savo, G. Carva- cho, N. Spagnolo, T. Giordani, and F. Sciarrino, arXiv preprint arXiv:2310.20521 (2023)
2023 arXiv
-
[34]
Bozzio, M
M. Bozzio, M. Vyvlecka, M. Cosacchi, C. Nawrath, T. Seidelmann, J. C. Loredo, S. L. Portalupi, V. M. Axt, P. Michler, and P. Walther, npj Quantum Information8, 104 (2022)
2022
-
[35]
J. J. Renema, Physical Review A 101, 063840 (2020)
2020
-
[36]
Drahi, D
D. Drahi, D. V. Sychev, K. K. Pirov, E. A. Sazhina, V. A. Novikov, I. A. Walmsley, and A. Lvovsky, Quantum 5, 416 (2021)
2021
-
[37]
Bartolucci, P
S. Bartolucci, P. Birchall, H. Bombin, H. Cable, C. Daw- son, M. Gimeno-Segovia, E. Johnston, K. Kieling, N. Nickerson, M. Pant, et al. , Nature Communications 14, 912 (2023)
2023
-
[38]
M. A. Nielsen, Reports on Mathematical Physics 57, 147 (2006)
2006
-
[39]
Mandracchia, X
B. Mandracchia, X. Hua, C. Guo, J. Son, T. Urner, and S. Jia, Nature Communications 11, 94 (2020)
2020
-
[40]
K. E. Kubow and A. R. Horwitz, Nature Cell Biology 13, 3 (2011)
2011
-
[41]
Note: while µy andσy parametrise the log-normal distri- bution, they should not be interpreted as its mean and standard deviation
-
[42]
Large speed-up of quantum emitter detection via quantum interference
S. Algeri, J. Aalbers, K. D. Mor˚ a, and J. Conrad, Nature Reviews Physics 2, 245 (2020). S1 Supplementary Material for: “Large speed-up of quantum emitter detection via quantum interference” W arwick P . Bowen Australian Research Council Centre of Excellence in Quantum Biotec...
2020
Reviewed August 16, 2026 · model on record in the stance chip above.
Discussion (0). Continue with ORCID to comment.