Phase-Drift Limits and Adaptive Quadrature Readout in Programmable Photonic Processors

G\"okhan Elmas; Igor A. Litvin; Janis N\"otzel

REVIEW 2 major objections 5 minor 19 references

Phase-Drift Limits and Adaptive Quadrature Readout in Programmable Photonic Processors

T0 review · 2 major / 5 minor · reviewed 2026-08-07 · deepseek-v4-flash

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

Pith's one-line read This paper derives the phase-drift penalty of sequential sine/cosine readout in programmable photonic processors: fixed ordering costs $3Q_\tau/8$ in mean-square error, and a phase-predicted ordering rule reduces that to…

desk verdict Solid analytical paper with a missing empirical anchor: the central Qτ estimate is never reported, so the practical claims hang on an unstated number. read the letter →

arxiv 2608.02249 v2 pith:UKEOW2EM submitted 2026-08-03 physics.optics quant-ph

Gökhan Elmas , Igor A. Litvin , Janis Nötzel This is my paper · ORCID

classification physics.opticsquant-ph

keywords programmablephotonicprocessorphasedriftinstabilityquadraturereadoutestimationadaptivemeasurementFisherinformationfeedbackstabilization

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

This paper asks how much phase-estimation accuracy is lost when a programmable photonic processor reads out its sine and cosine quadratures sequentially instead of simultaneously, and when a temporal readout is actually worth using. Working from the four-port configuration of an eight-mode processor, the authors derive that, to first order in the phase increment $\delta_\tau$, the reconstruction error is $-\delta_\tau\sin^2\phi_0$ for cosine-then-sine order and $-\delta_\tau\cos^2\phi_0$ for the reverse, giving a uniform mean-square drift error of $3Q_\tau/8$, where $Q_\tau=\mathrm{Var}(\delta_\tau)$ is the phase-increment variance. They then show that if a predictor of the phase is available, measuring the locally less informative quadrature first lowers the uniform first-order penalty to $(3/8-1/\pi)Q_\tau$, an 84.9 percent reduction. The same ordering principle emerges from a state-space treatment in which the Fisher information of a stale measurement falls from $I$ to $I/(1+IQ_\tau)$. These laws matter because they turn a processor's measured phase statistics into a dimensionless crossover condition, expressed in the product of spatial Fisher information and phase-increment variance, that tells an operator when sequential readout beats simultaneous readout.

What carries the argument

The load-bearing object is the differential of the two-argument arctangent, $d\theta=(x\,dy-y\,dx)/(x^2+y^2)$, applied to a pair in which one quadrature is sampled before the drift and the other after. This identity turns a delayed quadrature sample into an explicit first-order phase error whose drift coefficient weights the locally insensitive quadrature: $-\sin^2\phi_0$ for cosine-first order and $-\cos^2\phi_0$ for sine-first order. A second mechanism is the phase-predicted ordering rule of Eq. (78), which compares $\sin^2\phi_p$ with $\cos^2\phi_p$ (or, with unequal noise, the full Fisher informations $I_C(\phi_p)$ and $I_S(\phi_p)$) and measures the less informative quadrature first; symmetry reduces the uniform average of the minimized coefficient to $3/8-1/\pi$. The third mechanism is the increment-aware marginalization of Eq. (95), which represents a stale observation as a fresh observation plus an extra Gaussian term of variance $Q_\tau$, so its effective information is $I/(1+IQ_\tau)$; this is what makes the state-space estimator stop trusting an old sample once the intervening phase uncertainty dominates. Finally, the ideal balanced Poisson calculation shows each quadrature carries Fisher information equal to its detected signal-photon number, which gives the physical resource scale for the architecture boundaries.

What would settle it

Take simultaneous sine and cosine records from the same processor, estimate $Q_\tau$ at a chosen reconfiguration interval via Eq. (29), then form sequential estimates with both fixed orders and with the adaptive order using a phase predictor whose error is small compared with the ordering boundary; if the ratio of adaptive to fixed mean-square error does not approach $(3/8-1/\pi)/(3/8)\approx0.151$ as $Q_\tau\to 0$, the first-order drift law is wrong.

Watch

Extended reading notes

Core claim

The paper's central claim is that sequential quadrature readout has a definite, computable phase-drift penalty, and that this penalty can be reduced by ordering the two measurements according to a phase predictor. With the estimate defined at the completion of the second measurement, the exact atan2 reconstruction of the mixed-time pair obeys, to first order in the phase increment, $e_{C\to S}=-\delta_\tau\sin^2\phi_0+O(\delta_\tau^2)$ and $e_{S\to C}=-\delta_\tau\cos^2\phi_0+O(\delta_\tau^2)$. Averaging over a uniformly distributed initial phase gives a fixed-order drift mean-square error of $3Q_\tau/8$. When a predictor $\phi_p$ for the initial phase is available, the rule "measure the quadrature with the smaller local Fisher information first" gives $\mathbb{E}[\min(\sin^4\phi_0,\cos^4\phi_0)]=3/8-1/\pi$, an 84.9 percent reduction in the drift-induced mean-square error. The same ordering principle is recovered from a local state-space model: marginalizing the unknown phase increment reduces a stale observation's Fisher information from $I$ to $I/(1+IQ_\tau)$, so delaying a highly informative quadrature is more costly than delaying a weak one. In the ideal balanced Poisson benchmark, each quadrature's Fisher information equals its detected signal-photon number, which converts the whole comparison into dimensionless architecture boundaries in the plane of total spatial information $I_{\mathrm{sp}}$ and phase-increment variance $Q_\tau$.

Load-bearing premise

The load-bearing premise is that the phase drifts only slightly during the short gap between the two quadrature measurements, so the first-order error formulas and the choice of arctangent branch are valid, and that a predictor of the current phase is accurate enough to choose the right measurement order.

Editorial extensions

If this is right

For information gain $g=1.6$, temporal readout is favored over simultaneous readout only below the dimensionless boundaries $I_{\mathrm{sp}}Q_\tau=1.00$ (fixed order), $3.75$ (symmetric increment-aware), and $6.61$ (ideal adaptive order).
The adaptive advantage is robust to moderate predictor error: at predictor error $\sigma_p=0.1$ rad the reduction is still 83.2 percent, and it degrades to the fixed-order value only as the predictor becomes uninformative.
In the noisy receiver simulation, the increment-aware estimator pushes the spatial-to-temporal crossover from $Q\approx0.0100$ to $Q\approx0.0127$, and adaptive ordering pushes it to $Q\approx0.0470$, meaning each refinement extends the range over which sequential readout is preferable.
A stale quadrature can never carry more than $1/Q_\tau$ of information about the end-time phase, so even a perfect measurement of the initial phase cannot beat the uncertainty of the intervening process increment.
The architecture rules depend only on the product $I_{\mathrm{sp}}Q_\tau$ and the gain factor $g$, not on any particular delay-scaling law, so they can be applied at whatever reconfiguration interval an operator chooses.

Reading between the lines

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

The same "measure the locally less informative quantity first" principle should transfer to any tracking problem where two observations of a slowly drifting parameter have phase-dependent Fisher information; the sine-cosine quadrature pair is the symmetric special case.
For a practical controller, the useful quantity is not the perfect-prediction coefficient $3/8-1/\pi$ but the coefficient as a function of predictor error $\sigma_p$; the paper's $\sigma_p=0.1$ rad result suggests the gain degrades smoothly, so an operator could calibrate the operating point from the measured $\sigma_p$.
The empirical estimator $\hat{Q}_{\tau_m}$ built from synchronized simultaneous records gives a direct way to populate the architecture boundaries on a real processor: compute $I_{\mathrm{sp}}$ from the calibrated detection model, estimate $Q_\tau$ at the intended switching interval, and check which side of the boundary the operating point lies on.
One could test the increment-aware estimator's advantage outside the small-drift regime by comparing the circular posterior of Eq. (104) with fixed-order atan2 at larger $Q_\tau$; the paper's Monte Carlo already shows the first-order laws depart from exact results there, but the circular estimator's behavior at those drifts is presented for a specific noise model rather than as a closed-form law.

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.

Desk Editor's Note

read the letter

This one is a solid theory paper with an empirical anchor problem. The central derivation—the first-order atan2 drift formulas, the adaptive ordering coefficient (3/8 − 1/π), and the Fisher-information reduction I/(1+IQτ)—is correct and, as far as I can tell, not in the prior literature. The Monte Carlo checks line up with the analytic laws, and the dimensionless architecture boundaries are a useful addition. That part deserves a serious referee.

The soft spot is the missing bQτ. Section 3.1 defines an estimator from the 35 recordings, but the paper never reports the value. So the headline 84.9% reduction and the crossover boundaries are not tied to the actual processor that motivated the study. Figure 5 shows the first-order law breaks down as Qτ grows; without knowing whether the real device sits in the small-drift regime, the practical relevance is an open question. This is a fixable gap, not a fatal one, but it is a real gap. Also, the code and data are only on request; a public artifact would help.

I was a bit more charitable than the stress-test note: the missing Qτ doesn't undermine the algebra, and the paper explicitly states the limits of its local approximations in Sec. 8.5. But the empirical claim is the one place where the reader has to take the authors at their word, and the paper makes it easy to check—just report the number.

Who is this for? People designing temporal receivers for programmable photonic processors, especially those doing phase tracking with sequential quadrature measurements. It gives them design rules they can use. It won't reshape the field, but it's honest progress.

I'd send it to review. The theory is solid, the siting is appropriate, and the missing Qτ is addressable in a revision.

Referee Report

2 major / 5 minor

Summary. This paper analyzes the estimation error caused by sequential (temporal) acquisition of sine and cosine quadratures in programmable photonic processors, where the relative phase drifts between the two measurements. The authors define the target phase at the end of the second measurement, derive the first-order errors e_{C→S} = -δτ sin²φ0 + O(δτ²) and e_{S→C} = -δτ cos²φ0 + O(δτ²), and obtain the uniform fixed-order mean-square penalty 3Qτ/8. They then propose a phase-predicted adaptive ordering rule with uniform penalty (3/8 - 1/π)Qτ, an 84.9% MSE reduction, and an increment-aware state-space estimator whose effective Fisher information for a stale observation is I/(1+IQτ). Under an ideal balanced Poisson detection model, quadrature Fisher information equals detected photon number, yielding dimensionless architecture boundaries. Monte Carlo simulations validate the small-drift laws and quantify robustness to prediction error and nonlinear effects.

Significance. The analytical core is self-contained and the algebra checks: the atan2 perturbation, the uniform averages 3/8 and 3/8 - 1/π, and the effective-information formula follow from stated assumptions without fitted parameters. The paper explicitly separates ideal benchmarks (perfect phase prediction, symmetric local information) from nonlinear circular simulations, and it acknowledges detector nonidealities. The main value is a simple design rule for choosing between spatial and temporal readout, with explicit resource assumptions. The strengths—derivations grounded in standard estimation theory, Monte Carlo validation of the nonlinear estimators, and a clear statement of model scope—support the paper's usefulness if the missing empirical anchor is supplied.

major comments (2)

[Sec. 3.1 and Sec. 9] The estimator bQτm defined in Eq. (29) is never evaluated. The paper motivates the receiver analysis with 35 free-running recordings from an eight-mode processor but reports no value (or even order of magnitude) for Qτ at any reconfiguration interval τ. This matters because the headline laws 3Qτ/8 and (3/8 - 1/π)Qτ, and the crossover conditions in Eqs. (122)-(127), are all linear in Qτ or IspQτ and are valid only in the small-drift regime; Fig. 5 shows the first-order law departing from exact Monte Carlo as Qτ grows. Without a reported bQτm, the reader cannot verify that the motivating processor actually operates in the regime where the 84.9% reduction and the stated crossover values are quantitatively accurate. The authors should report bQτm for at least the selected reconfiguration interval, along with a confidence interval or range, or explicitly relativize the motivating claims.
[Sec. 5, Sec. 6, and Sec. 8] The paper never states a quantitative criterion for 'sufficiently small δτ'. Equations (55), (69), (76), and (83) are derived under a first-order Taylor expansion and neglect atan2 branch crossings, but the architecture boundaries (122)-(127) use these first-order coefficients without a stated validity condition. The Monte Carlo comparison in Sec. 9 validates the fixed-order crossover at Q = 0.01, but no general threshold (e.g., Qτ below some fraction of rad²) is given. Since the boundaries are intended as design rules, the authors should add a smallness condition or an explicit statement of the range of Qτ over which the first-order coefficients are accurate to a given tolerance.

minor comments (5)

[Throughout] There are typographical issues: 'Recket al.' should be 'Reck et al.', and the author names 'G¨ okhan' and 'N¨ otzel' show encoding artifacts in the full text.
[Sec. 5] Equation (72) writes O(Qτ²), but the second-order corrections in Eqs. (57) and (71) are phase-dependent; the remainder notation could be clarified to indicate that the coefficient of Qτ² depends on φ0.
[Sec. 9] The crossover values (Q⋆,fixed = 0.0100, Q⋆,IA ≈ 0.0127, Q⋆,adaptive ≈ 0.0470) are reported without error bars or a quantitative agreement metric; a table comparing the Monte Carlo crossovers with the analytical expressions (122), (124), and (127) would substantially aid the reader.
[Sec. 3.1] The unwrapped phase reconstruction in Eq. (24) is sensitive to unwrapping errors, but no discussion of outlier handling or unwrap robustness is provided; a brief note on this would strengthen the definition of the estimator bQτm.
[Data availability] The statement that simulation code and data are 'available from the authors upon reasonable request' is acceptable but a permanent repository link would improve reproducibility.

Circularity Check

0 steps flagged · score 1.0 of 10

No significant circularity: the drift, adaptive-ordering, and Fisher-information results are self-contained; the only self-citations are motivational and the missing Qτ is an empirical gap, not a circular step.

full rationale

The derivation chain for the headline results is self-contained. The fixed-order errors follow directly from Taylor-expanding the delayed quadrature and applying the exact differential dθ = (x dy − y dx)/(x^2 + y^2): Eq. (55), e_C→S = −δτ sin²φ0 + O(δτ²), and Eq. (69), e_S→C = −δτ cos²φ0 + O(δτ²), are obtained from Eqs. (39)–(54) and (58)–(68), not assumed. The uniform MSE 3Qτ/8 (Eq. 76) is the elementary integral E[sin⁴φ0] = 3/8, and the adaptive coefficient 3/8 − 1/π (Eq. 83) is the same integral over the sector where min(sin⁴, cos⁴) applies; neither coefficient is fitted. The increment-aware Fisher reduction I/(1 + IQτ) (Eq. 95) follows from rewriting the stale observation as z_C = φ1 + (ε_C − w) with independent variances I_C⁻¹ and Qτ; this is a standard Gaussian marginalization, and the photon-counting identification I = ηΛ (Eqs. 112, 114) is derived from the Poisson likelihood. The architecture boundaries (Eqs. 122, 124, 127) are algebraic consequences of these expressions and are stated with their idealized assumptions. The main self-citations are to Ref. [9] for the motivating eight-mode processor and to related group work [10–13]; these supply the experimental context and the estimator definition Eq. (29), but no reported bQτ value is used to force the analytic or Monte Carlo results, and Ref. [9] is not invoked as a uniqueness theorem or as an ansatz. The Monte Carlo section uses the same generative model as the theory; this is a self-consistency check, not an independent external benchmark, and the paper itself flags the departure of the first-order law at larger Qτ (Fig. 5) and the erosion of the adaptive advantage with predictor error (Fig. 6). The failure to report bQτ from the 35 recordings is a missing empirical anchor for the motivating regime, not a circular step. No reduction of a prediction to its input by construction was found.

Assumptions & free parameters 3 free parameters · 6 assumptions · 0 invented entities

The central results are cleanly derived from standard calculus and Gaussian state-space assumptions. Qτ and g are inputs, not fitted to make a specific outcome appear. No ad hoc entities are introduced.

free parameters (3)

Qτ (phase-increment variance)
Central input; the paper provides an estimator in Eq. (29) but reports no measured value from the 35 recordings, so the crossover boundaries remain undemonstrated on real data.
g (information gain factor) = 1.6 in Monte Carlo
Chosen resource ratio for the architecture comparison; the boundaries depend on it.
vsp, vt, σp (simulation noise and prediction parameters) = vsp=0.01, vt=6.25e-3, σp=0.1 rad
Monte Carlo design parameters for the receiver comparison and adaptive robustness study.

assumptions (6)

standard math Differential identity for atan2: dθ = (x dy - y dx)/(x² + y²)
Used in Sec. 5 to derive first-order drift errors.
domain assumption Local zero-mean Gaussian increment model: δτ ~ N(0, Qτ)
Assumed in Sec. 3 for the receiver analysis and Monte Carlo; the paper notes it is a local model, not a global diffusion fit.
domain assumption Uniform phase prior for φ0 when averaging drift penalties
Used in Sec. 5.3 and Sec. 6 to compute E[sin4φ0] and the adaptive coefficient.
domain assumption Independent Gaussian quadrature noise with phase-independent variances
Assumed in Sec. 6.1 and Sec. 7 for Fisher information and the linearized phase observations.
domain assumption Ideal balanced Poisson photon-counting detection
Used in Sec. 8.1 to show each quadrature's Fisher information equals its detected signal-photon number; acknowledged as a benchmark, not a detector model for the processor data.
ad hoc to paper Perfect phase prediction for the ideal adaptive ordering analysis
The analytic 84.9% reduction assumes the predictor φp equals φ0; the robustness study with σp > 0 is a separate simulation, and Figure 6 shows the benefit degrades with prediction error.

how reviews work

0 comments

Cite this review

Pith. "Pith review of Phase-Drift Limits and Adaptive Quadrature Readout in Programmable Photonic Processors." pith.science (2026). https://pith.science/paper/UKEOW2EM

@misc{pith2026260802249,
  author       = {Pith},
  title        = {Pith review of: Phase-Drift Limits and Adaptive Quadrature Readout in Programmable Photonic Processors},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/UKEOW2EM}},
  note         = {Machine review of arXiv:2608.02249}
}

abstract

Phase fluctuations between optical inputs limit programmable photonic processors because their output powers depend on coherent interference. We study the phase-drift penalty that arises when sine and cosine quadratures are measured sequentially rather than simultaneously. The analysis is motivated by measurements from an eight-mode programmable photonic processor, including 35 free-running recordings of 300 s acquired at approximately 125 samples per second per channel. These recordings provide an empirical route for estimating the phase-increment variance at a selected reconfiguration interval. The estimate is defined at the time of the second measurement. For fixed quadrature order, perturbation of the atan2 reconstruction gives $e_{C\to S}=-\delta_\tau\sin^2\phi_0+O(\delta_\tau^2)$ and $e_{S\to C}=-\delta_\tau\cos^2\phi_0+O(\delta_\tau^2)$. Writing $Q_\tau=\operatorname{Var}(\delta_\tau)$, uniform phase averaging gives the first-order drift mean-square error $3Q_\tau/8$. A phase-predicted ordering rule measures the locally less informative quadrature first and the more informative quadrature second. Its uniform first-order penalty is $(3/8-1/\pi)Q_\tau$, which is 84.9 percent below the fixed-order value. We also derive an increment-aware estimator from a local state-space model. Marginalizing the unknown phase increment increases the variance of a stale phase observation by $Q_\tau$, reducing its Fisher information from $I$ to $I/(1+IQ_\tau)$. For ideal balanced Poisson detection, the Fisher information of each quadrature equals its detected signal-photon number. This yields dimensionless architecture boundaries in spatial information and phase-increment variance. Nonlinear Monte Carlo simulations validate the perturbative laws, quantify robustness to prediction error, and compare simultaneous, fixed-order, increment-aware, and adaptive receivers under a common noise model.

Figures

Figures reproduced from arXiv: 2608.02249 by the authors.

**Figure 2.** Spatial and temporal quadrature acquisition. The temporal example shows the order [PITH_FULL_IMAGE:figures/full_fig_p005_2.png] view at source ↗

**Figure 3.** Phase-dependent drift coefficients for the two [PITH_FULL_IMAGE:figures/full_fig_p006_3.png] view at source ↗

Figures from the paper (4 more)

**Figure 4.** Figure 4: Architecture-selection boundaries for informa [PITH_FULL_IMAGE:figures/full_fig_p009_4.png]

**Figure 6.** Figure 6: Effect of prior phase uncertainty on adaptive [PITH_FULL_IMAGE:figures/full_fig_p009_6.png]

**Figure 6.** Figure 6: Effect of prior phase uncertainty on adaptive [PITH_FULL_IMAGE:figures/full_fig_p010_6.png]

**Figure 7.** Figure 7: Resource-normalized receiver comparison for [PITH_FULL_IMAGE:figures/full_fig_p010_7.png]

Review history (2 revisions) →

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

19 extracted references · 5 canonical work pages

[1]

Experimental realization of any discrete unitary op- 10 erator,

M. Reck, A. Zeilinger, H. J. Bernstein, and P. Bertani, “Experimental realization of any discrete unitary op- 10 erator,”Physical Review Letters73, 58–61 (1994), doi: 10.1103/PhysRevLett.73.58

doi:10.1103/physrevlett.73.58 1994
[2]

Optimal design for universal multiport interferometers,

W. R. Clements, P. C. Humphreys, B. J. Metcalf, W. S. Kolthammer, and I. A. Walmsley, “Optimal design for universal multiport interferometers,”Optica3, 1460–1465 (2016), doi: 10.1364/OPTICA.3.001460

doi:10.1364/optica.3.001460 2016
[3]

Large-scale quantum photonic circuits in silicon,

N. C. Harris, D. Bunandar, M. Pant, G. R. Stein- brecher, J. Mower, M. Prabhu, T. Baehr-Jones, M. Hochberg, and D. Englund, “Large-scale quantum photonic circuits in silicon,”Nanophotonics5, 456– 468 (2016), doi: 10.1515/nanoph-2015-0146

doi:10.1515/nanoph-2015-0146 2016
[4]

Quantum circuits with many photons on a pro- grammable nanophotonic chip,

J. M. Arrazola, V. Bergholm, K. Br´ adl´ er,et al., “Quantum circuits with many photons on a pro- grammable nanophotonic chip,”Nature591, 54–60 (2021), doi: 10.1038/s41586-021-03202-1

doi:10.1038/s41586-021-03202-1 2021
[5]

Silicon photonics codesign for deep learning,

Q. Cheng, J. Kwon, M. Glick, M. Bahadori, L. P. Car- loni, and K. Bergman, “Silicon photonics codesign for deep learning,”Proceedings of the IEEE108, 1261– 1282 (2020), doi: 10.1109/JPROC.2020.2968184

work page arXiv 2020
[6]

Pro- grammable photonic circuits,

W. Bogaerts, D. P´ erez, J. Capmany,et al., “Pro- grammable photonic circuits,”Nature586, 207–216 (2020), doi: 10.1038/s41586-020-2764-0

doi:10.1038/s41586-020-2764-0 2020
[7]

Impact of me- chanical vibrations on laser stability and carrier phase estimation in coherent receivers,

M. Kuschnerov, K. Piyawanno, M. S. Alfiad, B. Spinnler, A. Napoli, and B. Lankl, “Impact of me- chanical vibrations on laser stability and carrier phase estimation in coherent receivers,”IEEE Pho- tonics Technology Letters22, 1114–1116 (2010), doi: 10.1109/LPT.2010.2050472

work page arXiv 2010
[8]

Seimetz,High-Order Modulation for Optical Fiber Transmission

M. Seimetz,High-Order Modulation for Optical Fiber Transmission. Berlin, Germany: Springer, 2009

work page 2009

Show all 19 references

[9]

Mod- eling and analysis of phase instability in a photonic processor,

G. Elmas, I. A. Litvin, P. Kohl, and J. N¨ otzel, “Mod- eling and analysis of phase instability in a photonic processor,”Applied Optics64, 3995–4003 (2025), doi: 10.1364/AO.560370

2025 doi
[10]

Stable signal processing with a photonic processor,

I. A. Litvin, G. Elmas, P. Kohl, and J. N¨ otzel, “Stable signal processing with a photonic processor,”Journal of Lightwave Technology43, 9981–9990 (2025), doi: 10.1109/JLT.2025.3606027

2025
[11]

Robust calibration and en- ergy optimization in reconfigurable photonic proces- sors,

I. A. Litvin, G. Elmas, K. H. El-Safty, S. Chaud- hary, and J. N¨ otzel, “Robust calibration and en- ergy optimization in reconfigurable photonic proces- sors,”Optics Express33, 35011–35027 (2025), doi: 10.1364/OE.566817

2025 doi
[12]

Multi- input signal phase stabilization in photonic proces- sors with on-chip feedback control,

I. A. Litvin, G. Elmas, P. Kohl, and J. N¨ otzel, “Multi- input signal phase stabilization in photonic proces- sors with on-chip feedback control,”Optics Express 34, 11244–11258 (2026), doi: 10.1364/OE.579969

2026 doi
[13]

Joint detection on reconfigurable photonic processors with multi-input phase stabilization,

I. A. Litvin, G. Elmas, and J. N¨ otzel, “Joint detection on reconfigurable photonic processors with multi-input phase stabilization,”IEEE Photonics Journal18(3), 1–9 (2026), doi: 10.1109/JPHOT.2026.3683204

2026
[14]

Linear programmable nanophotonic processors,

N. C. Harris, J. Carolan, D. Bunandar, M. Prabhu, M. Hochberg, T. Baehr-Jones, M. L. Fanto, A. M. Smith, C. C. Tison, P. M. Alsing, and D. Englund, “Linear programmable nanophotonic processors,”Optica5, 1623–1631 (2018), doi: 10.1364/OPTICA.5.001623

2018 doi
[15]

8 ×8 reconfigurable quantum photonic processor based on silicon nitride waveguides,

C. Taballione, T. A. W. Wolterink, J. M. Renema, M. S. de Goede, B. J. Metcalf, P. P. Rohde, H. S. M. T. Yung, M. A. de Dood, E. J. Klein, D. J. Broeke, and K.-J. Boller, “8 ×8 reconfigurable quantum photonic processor based on silicon nitride waveguides,”Optics Express27, 268...

2019 doi
[16]

Large-scale silicon quantum photonics implementing arbitrary two-qubit processing,

X. Qiang, X. Zhou, J. Wang, C. M. Wilkes, T. Loke, S. O’Gara, L. Kling, G. D. Marshall, R. Santagati, T. C. Ralph, J. B. Wang, J. L. O’Brien, M. G. Thomp- son, and J. C. F. Matthews, “Large-scale silicon quantum photonics implementing arbitrary two-qubit processing,”Nature Pho...

2018 doi
[17]

Silicon quantum pho- tonics,

J. W. Silverstone, D. Bonneau, J. L. O’Brien, and M. G. Thompson, “Silicon quantum pho- tonics,”IEEE Journal of Selected Topics in Quantum Electronics22, 390–402 (2016), doi: 10.1109/JSTQE.2016.2573218

2016
[18]

Phase-controlled inte- grated photonic quantum circuits,

B. J. Smith, D. Kundys, N. Thomas-Peter, P. G. R. Smith, and I. A. Walmsley, “Phase-controlled inte- grated photonic quantum circuits,”Optics Express 17, 13516–13525 (2009), doi: 10.1364/OE.17.013516

2009 doi
[19]

Sub-0.1 degree phase locking of a single-photon inter- ferometer,

V. Svarc, M. Nov´ akov´ a, M. Dudka, and M. Jeˇ zek, “Sub-0.1 degree phase locking of a single-photon inter- ferometer,”Optics Express31, 12562–12571 (2023), doi: 10.1364/OE.487414. 11

2023 doi

Pith tools

Reviewed August 7, 2026 · model on record in the stance chip above.

[1] [1]

Experimental realization of any discrete unitary op- 10 erator,

M. Reck, A. Zeilinger, H. J. Bernstein, and P. Bertani, “Experimental realization of any discrete unitary op- 10 erator,”Physical Review Letters73, 58–61 (1994), doi: 10.1103/PhysRevLett.73.58

doi:10.1103/physrevlett.73.58 1994

[2] [2]

Optimal design for universal multiport interferometers,

W. R. Clements, P. C. Humphreys, B. J. Metcalf, W. S. Kolthammer, and I. A. Walmsley, “Optimal design for universal multiport interferometers,”Optica3, 1460–1465 (2016), doi: 10.1364/OPTICA.3.001460

doi:10.1364/optica.3.001460 2016

[3] [3]

Large-scale quantum photonic circuits in silicon,

N. C. Harris, D. Bunandar, M. Pant, G. R. Stein- brecher, J. Mower, M. Prabhu, T. Baehr-Jones, M. Hochberg, and D. Englund, “Large-scale quantum photonic circuits in silicon,”Nanophotonics5, 456– 468 (2016), doi: 10.1515/nanoph-2015-0146

doi:10.1515/nanoph-2015-0146 2016

[4] [4]

Quantum circuits with many photons on a pro- grammable nanophotonic chip,

J. M. Arrazola, V. Bergholm, K. Br´ adl´ er,et al., “Quantum circuits with many photons on a pro- grammable nanophotonic chip,”Nature591, 54–60 (2021), doi: 10.1038/s41586-021-03202-1

doi:10.1038/s41586-021-03202-1 2021

[5] [5]

Silicon photonics codesign for deep learning,

Q. Cheng, J. Kwon, M. Glick, M. Bahadori, L. P. Car- loni, and K. Bergman, “Silicon photonics codesign for deep learning,”Proceedings of the IEEE108, 1261– 1282 (2020), doi: 10.1109/JPROC.2020.2968184

work page arXiv 2020

[6] [6]

Pro- grammable photonic circuits,

W. Bogaerts, D. P´ erez, J. Capmany,et al., “Pro- grammable photonic circuits,”Nature586, 207–216 (2020), doi: 10.1038/s41586-020-2764-0

doi:10.1038/s41586-020-2764-0 2020

[7] [7]

Impact of me- chanical vibrations on laser stability and carrier phase estimation in coherent receivers,

M. Kuschnerov, K. Piyawanno, M. S. Alfiad, B. Spinnler, A. Napoli, and B. Lankl, “Impact of me- chanical vibrations on laser stability and carrier phase estimation in coherent receivers,”IEEE Pho- tonics Technology Letters22, 1114–1116 (2010), doi: 10.1109/LPT.2010.2050472

work page arXiv 2010

[8] [8]

Seimetz,High-Order Modulation for Optical Fiber Transmission

M. Seimetz,High-Order Modulation for Optical Fiber Transmission. Berlin, Germany: Springer, 2009

work page 2009