Pith. sign in

REVIEW 2 major objections 5 minor 17 references

Non-Hermitian photon number filtering using N00N state Bloch oscillations

T0 review · 2 major / 5 minor · reviewed 2026-08-06 · deepseek-v4-flash

Pith's one-line read A Bloch-oscillating waveguide array with an asymmetric loss cut can sort N=1 and N=2 N00N states, switching between ~80% two-photon and ~90% one-photon output by tuning input phase.

desk verdict The filtering idea is sound, but the phase inconsistency between Eq. (5) and Eq. (9) undercuts the headline switching claims until clarified. read the letter →

arxiv 2608.01484 v1 pith:QSYFOQPB submitted 2026-08-02 quant-ph

classification quant-ph MSC 81V80 PACS 42.50.-p
keywords N00NstatesBlochoscillationsphotonnumberfilteringasymmetriclosswaveguidearraysnon-Hermitianphotonicsphase-dependentswitchingquantumstatepreparation
verification ladder T0 review T1 audit T2 compute T3 formal

The pith

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

The reading

The paper proposes a way to separate the one- and two-photon parts of a photonic state without destroying it. In an array of waveguides whose effective index grows linearly, a one-photon N00N state (an entangled superposition of |1,0> and |0,1>) Bloch-oscillates with its bright fringe on one side fixed by input phase, while a two-photon N00N state spreads symmetrically on both sides. Removing half the waveguides at half a Bloch period deletes whichever sector sits on the lossy side; changing input phase switches between about 80% two-photon and 90% one-photon conditional output. Such a reconfigurable, non-destructive photon-number filter would be useful for preparing and routing states in photonic circuits.

What carries the argument

The argument is carried by the Green's function $U_{\mu,\nu}(z)=e^{i\pi(\mu-\nu)/2}e^{iBz(\mu+\nu)/2}J_{\mu-\nu}\big((4C/B)\sin(Bz/2)\big)$, which gives the single-photon propagation amplitudes and hence the photon-density profiles for N00N inputs. The cut is modeled as a partial trace over the modes on the lossy half of the array at $z_c=\lambda_B/2$, followed by reintroducing those waveguides as vacuum modes and evolving under the unchanged Hamiltonian to $z_{\text{out}}=\lambda_B$. The ratio $C/B$, proportional to $\lambda_B/L_t$, controls the transverse spread of the bright fringes and is the main optimization parameter.

What would settle it

Measure the output photon-number statistics of an integrated waveguide array with a physically implemented cut (for example a Bragg scatterer or siphoned waveguides) at $z=\lambda_B/2$; if the $P_1/P_2$ ratio does not track the input phase with peaks near $\phi=\pi$ and $\phi\approx\pi/2,3\pi/2$, or if reflected light from the cut repopulates the supposedly removed sector, the idealized trace-and-vacuum-reinsertion model fails.

Watch

Extended reading notes

Core claim

The paper claims that at the half Bloch period $z_c = \lambda_B/2$, the N=1 sector of a N00N input localizes on one side of the waveguide array according to the input phase $\phi$, while the N=2 sector spreads evenly across both sides. Imposing total loss on one half of the array at that point removes whichever sector sits on the lossy side. With equal N=1 and N=2 inputs, $\phi=\pi$ yields a conditional output that is more than 80% two-photon, while $\phi\approx\pi/2,3\pi/2$ yields roughly 90% one-photon output. The output is read from the two central waveguides at the full Bloch period $z_{\text{out}}=\lambda_B$, and the filtering is optimized by the ratio $C/B$ of coupling to index gradien

Load-bearing premise

The filtering effect is contingent on modeling the cut as instantaneous total loss in half the waveguides: the lost modes neither reflect nor re-couple, and the surviving array evolves under the same Hamiltonian; a real siphoning or Bragg-scattering cut of finite length could behave differently.

Editorial extensions

If this is right

  • For equal N=1/N=2 inputs with $C/B=0.3$ and $\phi=\pi$, the conditional output is more than 80% two-photon; tuning $\phi$ near $\pi/2$ or $3\pi/2$ gives roughly 90% one-photon output.
  • The filter is most effective when the input is N=1-dominated: with 90% N=1 input the relative two-photon efficiency stays high while the one-photon sector is strongly suppressed, at the cost of about 94% total intensity loss.
  • Only $C/B$ matters for the normalized dynamics, so the same design curves apply across different $B$ and $C$ values; a SiN/SiO2 platform with $L_t=200\ \mu$m, $C=0.008\ \mu$m$^{-1}$, $B=0.027\ \mu$m$^{-1}$ and $\Delta n_{\text{eff}}\sim0.007$ is within a feasible parameter range.
  • The scheme is a phase-controlled switch between photon-number subspaces distributed over two output waveguides, offering demultiplexing and non-destructive routing rather than detection-triggered state preparation.

Reading between the lines

Editorial extensions of the paper, not claims the author makes directly.

  • Because the filter exploits phase coherence between N=1 and N=2 sectors, a mixed initial state will degrade it; quantifying the tolerable phase noise or mixing would be a direct extension of the paper's density-matrix method.
  • The same trace-and-reinsert cut idea could be applied to staggered or partial loss profiles and to N>2 photon-number sectors, where numerical simulation is already the route the paper anticipates.
  • The appendix's beam-splitter coherent-state example implies this is a quantum-state filter, not a classical intensity filter; a natural application is cleaning or routing heralded non-classical photon sources.
  • A testable extension is to place output couplers on both central waveguides and measure photocount correlations; the predicted phase-dependent switch between one- and two-photon statistics should appear as a corresponding change in coincidence rates.
Share X Bluesky LinkedIn Reddit HN

Editorial analysis

A structured set of objections, weighed in public.

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

Referee Report

2 major / 5 minor

Summary. The manuscript studies Bloch oscillations of N=1 and N=2 photonic N00N states in an array of evanescently coupled waveguides with a linearly graded propagation constant. It proposes placing total loss on one half of the waveguide array at the half Bloch period and simulates the resulting evolution with a reduced density matrix. The central claim is that tuning the input phase phi dynamically switches the output between a 2-photon-dominated subspace (about 80% conditional probability) and a 1-photon-dominated subspace (about 90% conditional probability). The paper includes an analytic Green's-function framework, density-matrix equations for the cut, and appendices with partial analytical checks.

Significance. If correct, the proposed scheme is a concrete, passive approach to photon-number filtering and routing that does not rely on destructive detection. The physical mechanism, based on the phase-dependent transverse localization of the N=1 N00N sector at the half Bloch period while the N=2 sector remains symmetric, is clear and falsifiable. The manuscript also provides useful analytical scaffolding and a plausible experimental parameter range. However, the quantitative claims currently rest on an internal phase-factor inconsistency that must be resolved before the results can be accepted.

major comments (2)
  1. [III.1, Eq. (9); Figs. 3-4] The initial state Eq. (5) defines the N=1 sector as alpha(|1_mu 0_nu> + e^{i phi}|0_mu 1_nu>), but Eq. (9) writes the evolved N=1 term as alpha( a_dagger_mu(-z) + e^{2 i phi} a_dagger_nu(-z) )|0>. These differ: at phi=pi, Eq. (5) gives e^{i phi}=-1 while Eq. (9) gives e^{2 i phi}=+1. Since the cut simulations are stated to use Eq. (9), the reported phi=pi optimum for 2-photon filtration and the phi ~ pi/2, 3pi/2 optimum for 1-photon filtration may have been computed for a state whose N=1 phase is doubled. The authors must state which expression was implemented, correct Eq. (9) if it is a typo, and rerun or adjust Figs. 3-4 and the derived efficiencies accordingly. If the code used the erroneous factor, the headline 80%/90% numbers do not follow from the declared input state.
  2. [III.1, Eqs. (12)-(13)] The cut model traces out the left modes at z_c and then reintroduces them as vacuum modes for subsequent evolution under the same Hamiltonian. This is not equivalent to a 'total loss' boundary: after z_c, light can tunnel back into the supposedly cut waveguides and return to the output modes. The paper acknowledges the idealized nature of the cut, but the repopulation effect can change the conditional probabilities in Figs. 3-5. Please provide an estimate or a separate simulation with the cut modes removed (or with explicit loss terms) to show that the filtering numbers are robust, or clearly state this limitation in the abstract and conclusions.
minor comments (5)
  1. [II, Eq. (1)] The sum over nu from -N/2 to N/2 contains N+1 terms for even N, inconsistent with the text 'sum over N waveguides' and with the 6-waveguide simulation. Please clarify the indexing used.
  2. [Appendix B and C] The analytical N=1 checks effectively treat the phase-symmetric state (phi=0) and do not include the e^{i phi} factor of Eq. (5). Since the phase dependence is the central mechanism, it would be valuable to show the analytical formula for general phi, or at least state the phi=0 restriction explicitly.
  3. [III.1, Eq. (12)] The notation sum_{n_i<1} is unclear; presumably the trace should sum over all Fock states of the cut modes. Please write the sum explicitly over n_{-N/2},...,n_0.
  4. [References] Reference [15] has a malformed DOI ('10.1103/grwm-1kj3'); please verify the bibliographic data.
  5. [Conclusion] The statement 'N=2 transmission is currently ~40%' can be confused with the 80% conditional filtering claim. Please define whether the 40% is absolute transmission efficiency and clarify the relation to the conditional probabilities P1 and P2.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: the filtering results are computed from an external unitary Green's function and an explicit loss model, with no fitted prediction or self-citation chain.

full rationale

The derivation chain is self-contained. The initial state (Eq. 5), the tight-binding Hamiltonian (Eq. 1), and the Green's function solution (Eq. 7, taken from the prior external work of Bromberg et al., Ref. [13]) completely determine the lossless evolution. The loss cut is implemented by an explicit partial trace over half the waveguides at the half Bloch period, followed by reintroduction of vacuum modes (Eqs. 12-13), and the output probabilities are computed by direct density-matrix simulation (Eqs. 16-17). No parameter is fitted to the claimed 80%/90% filter efficiencies; the C/B ~ 0.3 optimum is selected by scanning a design parameter in Fig. 5, and the phase-switching effect follows from the known phase-dependent fringe localization of N=1 N00N states, which is an external result. The apparent factor-of-two phase inconsistency between Eq. (5) (e^{iφ} for the N=1 term) and Eq. (9) (e^{2iφ} for the N=1 term) is a potential correctness issue, not a circularity: even if the simulation followed Eq. (9), the outputs would still be computed consequences of a stated, albeit different, input state. There are no load-bearing self-citations, no imported uniqueness theorems, and no definitional identification of prediction with input.

Assumptions & free parameters 3 free parameters · 4 assumptions · 0 invented entities

The main free parameters are the input state amplitudes alpha and beta, the input phase phi, and the C/B ratio, which is optimized to 0.3. The key axioms are the standard coupled-mode Hamiltonian and Green's function from [13], the specialized initial state Eq. (5) with no |11> component, and the idealized reflectionless total-loss cut model. No new physical entities are introduced.

free parameters (3)
  • Input state amplitudes alpha, beta = alpha=beta=1 (equal probability), also varied (90%, 99%, 99.9% N=1 dominated)
    Define the input superposition of N=1 and N=2 N00N states. The filtering performance depends on this mix (Fig. 6), and these are chosen by hand rather than derived.
  • Input phase phi = phi=pi for 2-photon filtration; phi~pi/2, 3pi/2 for 1-photon filtration
    Control parameter that switches which side the N=1 sector fringe is on. Varied to achieve the filtering, not fitted to data.
  • Coupling-to-gradient ratio C/B = C/B ~ 0.3 (optimal for 2-photon filtration)
    Determines the transverse spread of the Bloch oscillation fringes. The paper optimizes this ratio (Fig. 5) to maximize the 2-photon conditional probability, making it a hand-chosen design parameter.
assumptions (4)
  • domain assumption The coupled-mode Hamiltonian Eq. (1) and the Green's function Eq. (7) from Bromberg et al. [13] accurately describe the waveguide array.
    The paper relies on this prior result for the evolution of photon creation operators. The Green's function is not re-derived in this work.
  • ad hoc to paper The initial state Eq. (5) is a coherent superposition of N=1 and N=2 N00N states with no |11> component and with phase factors e^{i phi} and e^{2 i phi}.
    This specific input state is what makes the filtering mechanism work, since the N=2 sector has no |11> interference. The paper acknowledges (Appendix A, Conclusion) that realistic sources produce |11> components, which degrade the filtering.
  • ad hoc to paper The cut at z_c = lambda_B/2 acts as instantaneous, total, reflectionless loss on half the waveguides, and the cut modes can be reintroduced as vacuum modes for subsequent evolution.
    Section III.1: tracing over cut modes and then reintroducing them as vacuum modes assumes no reflections and allows back-coupling into the cut region. The paper does not compare this to a physical cut that removes the waveguides entirely.
  • domain assumption The waveguide platform is lossless except for the engineered cut.
    The simulations ignore propagation loss, fabrication imperfections, and other decoherence mechanisms. This is an idealization common in theoretical proposals.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Non-Hermitian photon number filtering using N00N state Bloch oscillations." pith.science (2026). https://pith.science/paper/QSYFOQPB

@misc{pith2026260801484,
  author       = {Pith},
  title        = {Pith review of: Non-Hermitian photon number filtering using N00N state Bloch oscillations},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/QSYFOQPB}},
  note         = {Machine review of arXiv:2608.01484}
}
abstract

We explore Bloch oscillations of $N=1$ and $N=2$ photonic N00N states, using an array of linearly growing effective index waveguides with a simulated asymmetric loss profile. Starting with equal probability input $N=1$ and $N=2$ N00N states, and siphoning off a portion of the waveguides near the half Bloch period, we selectively output specific photon number states. Tuning the input phase, we achieve dynamic switching between 2-photon dominated (80\% of the output) and 1-photon dominated (90\% of the output) cases. This offers a path to improved state preparation in photonic circuits used in computing and networking.

Figures

Figures reproduced from arXiv: 2608.01484 by the authors.

Figure 2
Figure 2. FIG. 2. Example of an asymmetric loss platform (not to [PITH_FULL_IMAGE:figures/full_fig_p002_2.png] view at source ↗
Figure 1
Figure 1. FIG. 1. Photon density evolution for input N00N states - [PITH_FULL_IMAGE:figures/full_fig_p002_1.png] view at source ↗
Figure 3
Figure 3. FIG. 3. Conditional probabilities (Eq. (17)) with and without [PITH_FULL_IMAGE:figures/full_fig_p003_3.png] view at source ↗
Figures from the paper (3 more)
Figure 4
Figure 4. Figure 4: FIG. 4. Plot of conditional probabilities with and without [PITH_FULL_IMAGE:figures/full_fig_p004_4.png]
Figure 5
Figure 5. Figure 5: FIG. 5. Plot of conditional probabilities with and without [PITH_FULL_IMAGE:figures/full_fig_p004_5.png]
Figure 7
Figure 7. Figure 7: FIG. 7. Top: normalized efficiency ratio [PITH_FULL_IMAGE:figures/full_fig_p007_7.png]

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

17 extracted references · 14 canonical work pages

  1. [1]

    Jacobson, G

    J. Jacobson, G. Bj¨ ork, I. Chuang, and Y. Yamamoto, Phys. Rev. Lett.74, 4835 (1995)

  2. [2]

    Walther, J.-W

    P. Walther, J.-W. Pan, M. Aspelmeyer, R. Ursin, S. Gas- paroni, and A. Zeilinger, Nature429, 158–161 (2004)

  3. [3]

    H. F. Hofmann and T. Ono, Phys. Rev. A76, 031806 (2007)

  4. [4]

    J. P. Dowling, Contemporary Physics49, 125–143 (2008)

  5. [5]

    I. Afek, O. Ambar, and Y. Sil- berberg, Science328, 879 (2010), https://www.science.org/doi/pdf/10.1126/science.1188172

  6. [6]

    Giovannetti, S

    V. Giovannetti, S. Lloyd, and L. Maccone, Nature Pho- tonics5, 222–229 (2011)

  7. [7]

    A. N. Boto, P. Kok, D. S. Abrams, S. L. Braunstein, C. P. Williams, and J. P. Dowling, Physical Review Letters85, 2733–2736 (2000)

  8. [8]

    T. Ono, R. Okamoto, and S. Takeuchi, Nature Commu- nications4, 10.1038/ncomms3426 (2013)

Show all 17 references
  1. [9]

    M. Fink, F. Steinlechner, J. Handsteiner, J. P. Dowling, T. Scheidl, and R. Ursin, New Journal of Physics21, 053010 (2019)

  2. [10]

    Silvestri, H

    R. Silvestri, H. Yu, T. Str¨ omberg, C. Hilweg, R. W. Peterson, and P. Walther, Sci- ence Advances10, eado0215 (2024), https://www.science.org/doi/pdf/10.1126/sciadv.ado0215

  3. [11]

    M. A. Selim, M. Ehrhardt, Y. Ding, H. M. Dinani, Q. Zhong, A. Perez-Leija, S. K. ¨Ozdemir, M. Heinrich, A. Szameit, D. N. Christodoulides, and M. Khajavikhan, Science387, 1424–1428 (2025)

  4. [12]

    Bromberg, Y

    Y. Bromberg, Y. Lahini, R. Morandotti, and Y. Silber- berg, Phys. Rev. Lett.102, 253904 (2009)

  5. [13]

    Bromberg, Y

    Y. Bromberg, Y. Lahini, and Y. Silberberg, Physi- cal Review Letters105, 10.1103/physrevlett.105.263604 (2010)

  6. [14]

    Lebugle, M

    M. Lebugle, M. Gr¨ afe, R. Heilmann, A. P´ erez-Leija, S. Nolte, and A. Szameit, Nature Communications6 (2015)

  7. [15]

    R. Hui, T. Shit, M. Di Liberto, D. Sen, and S. Mukherjee, Physical Review A113, 10.1103/grwm-1kj3 (2026)

  8. [16]

    H. Kim, O. Kwon, and H. S. Moon, Scientific Reports 11, 10.1038/s41598-021-99804-w (2021). Appendix A: 2-photon sector interference We revisit the Bloch oscillations for the zero loss case in section II. Including the|11⟩state in Eq. (5), |ψ0⟩= 1 N n α |1µ0ν⟩+e iϕ|0µ1ν⟩ (A1) +...

  9. [17]

    This is accomplished by tracing over these degrees of freedom

    Now, atz=z c, we cut waveguides on the left side of the platform (µ≤0). This is accomplished by tracing over these degrees of freedom. We have: ⟨n−N/2...n0|ψ⟩(C2) = 1√ 2 (δµα +δ να)U αµ′(−zc)⟨n−N/2...n0|a† µ′|0⊗N ⟩ = 1√ 2 (δµα +δ να)U αµ′(−zc)δn−N/20...δnµ′ 1...δn00 ×(1−Θ(µ ′)...

Pith tools

Reviewed August 6, 2026 · model on record in the stance chip above.