Pith. sign in

REVIEW 2 major objections 4 minor 69 references

A newly defined continuous-variable link product composes quantum circuit fragments directly in phase space, with a polynomial-time algorithm for Gaussian circuits.

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 08:07 UTC pith:EWHIUVLQ

load-bearing objection The Gaussian link-product algorithm is real and well-checked; the paper's wider claim to a general CV comb formalism rests on an unproved infinite-squeezing limit that could fail for non-Gaussian fragments. the 2 major comments →

arxiv 2607.21215 v1 pith:EWHIUVLQ submitted 2026-07-23 quant-ph

Quantum Circuit Fragments and Link Products in Continuous Variables

Amalina Lai , Graeme D. Berk , Minjeong Song , Mile Gu This is my paper
classification quant-ph
keywords continuous variablesquantum combslink productGaussian statescircuit fragmentsnon-Markovian processesChoi statesadaptive measurements
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.

Quantum circuits are usually drawn with fixed inputs and outputs, but many tasks—probing a non-Markovian noise source, inserting a subroutine, or running an adaptive strategy—involve only an open fragment of a circuit. In finite dimensions the link product assembles such fragments into composite processes; this paper brings that tool to continuous-variable systems, whose natural carriers are qumodes such as photonic modes. The central claim is that a CV link product exists, defined from the infinite-squeezing limit of Choi states, and that for Gaussian fragments the composition can be computed directly on covariance matrices in time that scales polynomially in the number of modes. A sympathetic reader would care because it supplies the compositional machinery of quantum combs—non-Markovian process models and adaptive agent-environment interactions—to continuous-variable and Gaussian quantum information.

Core claim

The paper defines CV circuit fragments via Choi states built from two-mode squeezed vacuum states with finite squeezing, then defines their CV link product ⋆_J as the limit, as the squeezing of the joined input modes tends to infinity, of a trace of the product of the two Choi states after a partial transpose (Result 1, Eq. (8)). It proves that when both fragments are Gaussian, this limit can be evaluated directly by a covariance-matrix algorithm, PrepCM followed by repeated MConnect Schur complements, with complexity O(m \bar m^2) where m is the number of linked mode pairs and \bar m is the total number of qumodes (Result 2). The worked examples show that the algorithm reproduces the concat

What carries the argument

The central object is the CV link product ⋆_J—the quantum-comb composition rule for continuous variables—together with its Gaussian realization. The link product is defined as an infinite-squeezing limit of traced Choi states (Eq. (8)); the Gaussian realization replaces the infinite-dimensional Choi states by covariance matrices and performs the limit as a sequence of Schur complements via the subroutines PrepCM and MConnect.

Load-bearing premise

The load-bearing premise is that the infinite-squeezing limit in Eq. (8) converges to a valid Choi state for every continuous-variable circuit fragment; the proof and all worked examples are restricted to Gaussian fragments, so the universal validity of the CV link product is not established beyond the Gaussian case.

What would settle it

Take a non-Gaussian circuit fragment (for instance, a single-mode cubic-phase gate or a photon-subtraction operation), compute the right-hand side of Eq. (8) for a finite squeezing, and test whether the limit exists, is normalized to unit trace, and is a positive Choi state. A fragment for which the limit diverges, is non-positive, or depends on the order of the partial trace would refute the blanket universal link-product claim; the Gaussian algorithm alone would remain intact.

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

If this is right

  • Non-Markovian continuous-variable processes can be represented as the Choi covariance of a chain of stitched fragments, enabling systematic tomographic characterization of such processes.
  • Adaptive agent-environment interactions, including a memory register, can be composed and analyzed as a link product, collapsing to an effective quantum channel on the system.
  • For Gaussian fragments the cost of composing circuits is polynomial in the number of modes, avoiding the exponential density-matrix blow-up and making multi-time Gaussian circuits tractable.
  • The examples show that stitching beamsplitter modules yields closed-form effective channels, providing modular design tools for CV quantum circuits.

Where Pith is reading between the lines

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

  • A finite-squeezing experiment could quantify how large the squeezing must be before the ideal composition law holds, since the infinite-squeezing limit in Eq. (8) will carry a residual correction that decays with the squeezing parameter.
  • The covariance Schur-complement recipe could be dressed with a photon-number cutoff to approximate link products of non-Gaussian fragments, but the polynomial guarantee of Result 2 would be lost unless a different parameterization is found.
  • If the infinite-squeezing limit is shown to converge only for Gaussian fragments, the general CV-comb extension would need a separate regularization; the Gaussian algorithm itself would remain a self-contained contribution.

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

2 major / 4 minor

Summary. The paper develops a continuous-variable analog of the quantum-comb/link-product formalism. It defines CV circuit fragments through finite-squeezing Choi states (Definition 1), defines a CV link product via an infinite-squeezing limit (Definition 2 and Result 1, Eq. (8)), and then shows that for Gaussian fragments the link product can be computed directly in terms of covariance matrices and Schur complements (Result 2). The proposed Gaussian algorithm is illustrated on state-into-channel composition, channel concatenation, non-Markovian process construction, and adaptive agent–environment interactions. The stated goal is to extend the quantum-comb toolkit to continuous variables.

Significance. The Gaussian link-product algorithm is a genuinely useful contribution: it gives an efficient covariance-matrix method for composing Gaussian circuit fragments, with explicit verification against direct calculations in Appendices I, K, L, and M. The complexity claim O(m mbar^2) is plausible and well documented in Appendix F. The constructions build on established finite-squeezing Choi-state results [43,44], and no parameters are fitted to enforce the target outputs. However, the paper's central advertised generality—a CV link product for arbitrary CV circuit fragments—rests on the unproved convergence of the infinite-squeezing limit in Result 1. If that gap is closed, or if the claims are restricted to the Gaussian regime, the paper would be a solid contribution; as it stands, the general CV-comb framework is an assumption rather than an established result.

major comments (2)
  1. [Section II, Result 1 / Eq. (8)] The general CV link product is defined by the limit lim_{s_J→∞} K(s_J) tr_J[Υ_A^{T_JA} Υ_B(s_J ⊕ s̄_J)]. Appendix C's proof (Lemmas 1–2, and especially Eqs. (C13)–(C17)) moves the infinite-squeezing limit through partial traces and partial transposes without establishing uniform trace-class convergence or independence of the order of limits. For non-Gaussian fragments, particularly those with unbounded photon-number support in the Choi state, this limit need not exist or be unique. Since Definition 2 and the claimed general CV-comb extension rely on Result 1, this is a load-bearing gap that must be addressed—either by a convergence proof under explicit assumptions, or by explicitly restricting the general framework. The Gaussian algorithm (Result 2) is not affected, as it is justified directly in Appendices I, K, L, and M.
  2. [Abstract and Section II] The paper advertises a framework for 'continuous-variable circuit fragments' and a CV link product for arbitrary fragments, but every worked application and every theorem beyond the definition is Gaussian: Section IV and Appendices I–M treat only Gaussian channels, beam splitters, squeezers, and vacuum inputs. No non-Gaussian example is provided, and no convergence proof for the infinite-squeezing limit is given. The blanket statement that the formalism 'extends the quantum-comb toolkit to continuous variables' is therefore not currently supported. The robust, well-verified contribution is the Gaussian covariance-matrix link product; the general CV-comb claim should be either proven or appropriately scoped.
minor comments (4)
  1. [Section III] Typo: 'provides a means to to compute' should read 'provides a means to compute'.
  2. [Result 1 and Section II] The notation K(s_J) for the normalization constant is also used later for a circuit fragment K in Section IV; this overloading is confusing and should be disambiguated.
  3. [Appendix B] The reordering conventions for covariance matrices are described informally; a table or a small pseudocode block summarizing the mode order after PrepCM would improve reproducibility.
  4. [Appendix K] The text says 'we elect to also explicitly reorder... for pedagogical clarity' but the resulting block matrices are very dense; a concise summary of the final mode ordering would help the reader follow the calculation.

Circularity Check

0 steps flagged

No significant circularity: the CV link-product formula is anchored to external finite-squeezing Choi-state results and verified against independent direct calculations.

full rationale

The derivation chain is self-contained in the relevant sense. Definition 2 sets the CV link product as the operation producing the J-stitched composite's Choi state; Result 1 then derives the concrete limit expression (8) in Appendix C. The key input is Lemma 1, whose trace formula A(ρ) ∝ lim tr[ρ^T Υ_A] is cited from the external works [43,44] (Fiurášek; Giedke–Cirac), and the normalization constant K(r)=cosh^2(r/2) is computed inside the paper from the identity channel, not assumed. The Gaussian algorithm (Result 2) is derived independently from characteristic-function Gaussian integrals (Appendix D) and its outputs are checked against direct process evaluations: Appendix K states 'This verifies that the result of our link product indeed corresponds to the covariance of the Choi matrix of the process pictured in Fig. 6a,' and Appendices L–M perform the same cross-check for the bidirectional and comb-comb stitches. No parameter is fitted to force the target output; the examples are closed-form covariance results. Self-citations appear in the introduction and discussion for motivation only and are not load-bearing. The unproved infinite-squeezing convergence for non-Gaussian fragments is an analytic gap and a correctness risk, but it is not a circularity: the Gaussian regime, where the algorithm lives, is justified by direct Gaussian integrals and independent checks.

Axiom & Free-Parameter Ledger

0 free parameters · 3 axioms · 0 invented entities

The paper introduces no fitted parameters and no invented physical entities. Its central assumptions are the validity of the infinite-squeezing link-product limit, the decomposability of bidirectional links into commuting one-way contractions under causal order, and the faithfulness of finite-squeezing Choi states.

axioms (3)
  • domain assumption The infinite-squeezing limit defining the CV link product yields a well-defined Choi state for arbitrary CV circuit fragments.
    Result 1 / Eq. (8) defines the link product via lim_{s_J→∞}; convergence is not proved for non-Gaussian fragments; examples and the Gaussian algorithm assume it.
  • domain assumption Every circuit fragment has a well-defined global causal order, so any bidirectional stitching can be decomposed into commuting one-way contractions.
    Stated before Eq. (6) and used to extend Result 1 to bidirectional links; excludes causally indefinite fragments.
  • standard math Finite-squeezing Choi states fully characterize a CV circuit fragment.
    Asserted in Section II after Definition 1; follows from the full Schmidt rank of TMSV but is not proved in the paper.

pith-pipeline@v1.3.0-alltime-deepseek · 42261 in / 15152 out tokens · 169191 ms · 2026-08-01T08:07:31.368769+00:00 · methodology

0 comments
read the original abstract

Quantum circuits are often drawn as complete processes, with fixed inputs and outputs. In many quantum-information tasks, however, the natural object is only a fragment of such a circuit: an unknown source of non-Markovian noise to be probed, a subroutine to be inserted into a larger algorithm, or an agent implementing an adaptive strategy. In finite dimensions, the link product provides a systematic means to analyze how such circuit fragments interact and compose. Here, we develop the corresponding framework for continuous-variable systems. We introduce continuous-variable circuit fragments and associated link products that stitch such fragments together into larger processes. We show that the formalism simplifies substantially in the Gaussian regime, where link products can be evaluated efficiently using covariance-matrix representations. We use the formalism to construct non-Markovian processes and adaptive agent-environment interactions from modular components. This extends the quantum-comb toolkit to continuous variables, providing systematic methods for adaptive sensing, non-Markovian noise mitigation, and higher-order quantum circuit design.

Figures

Figures reproduced from arXiv: 2607.21215 by Amalina Lai, Graeme D. Berk, Mile Gu, Minjeong Song.

Figure 1
Figure 1. Figure 1: ). The dynamics being probed may also be non￾Markovian, such that successive channel uses are corre￾lated and the effect of a later intervention can depend on earlier interventions [4–7]. Indeed, the input to a quan￾tum algorithm is often not encoded in an input state, but rather in a small quantum circuit to be processed by inserting it as a subroutine within a larger circuit [8– 11]. In all these setting… view at source ↗
Figure 2
Figure 2. Figure 2: Choi States of Circuit Fragments. Consider a circuit fragment L with N = 2 input qudit wires and M = 3 output qudit wires (a). Its black-box properties can then be fully captured by the corresponding N + M = 5 qudit Choi state ΥL prepared by the circuit in (b). Here, each blue disk segment denotes a maximally entangled qudit pair |Φ⟩, while the green disk segment denotes the initial state of an ancilla qud… view at source ↗
Figure 3
Figure 3. Figure 3: J-stitching circuit fragments. Two circuit frag￾ments can be composed together to form larger circuit frag￾ments via a J-stitching, where J specifies exactly how the inputs and outputs on one fragment connect to the other. In the example above, circuit fragments A and B are composed to form C = A ⋆J B, where each pair of modes in J share the same color. Formally, we can write J = {(o A 1 , iB 1 ),(o B 1 , … view at source ↗
Figure 4
Figure 4. Figure 4: Stitching a Non-Markovian Process. We can construct the Choi representation of a Non-Markovian process by stitching a sequence of circuit fragments Tk in series, such that each two consecutive circuit fragments Tk and Tk+1 are stitched together by the stitching (E k o , Ek+1 i ). Although Result 1 is stated for unidirectional stitching from A to B, it is sufficient for evaluating any valid bidi￾rectional c… view at source ↗
Figure 5
Figure 5. Figure 5: Linking two quantum channels. ΥA(r) and ΥB(r ′ ) are the Choi states of two single-input, single-output circuit fragments, i.e., quantum channels. We can find the Choi states of their concatenation ΥC via the mode-stitching y = (o A, iB ). where Fs is as before. Direct application of our algo￾rithm, using ΓA(r) and ΓB(r ′ ), with the total number of stitched pairs, m = 1 gives Γout = (ur ′F 2 s ⊕ urI2) − M… view at source ↗
Figure 7
Figure 7. Figure 7: Agent-Environmental Interactions over mul￾tiple timesteps can be represented by two interacting quan￾tum circuit fragments which we denote here by K (green) and L (blue). Setting j1 = (o K 1 , iL 1 ), j2 = (o L 1 , iK 2 ) and j3 = (o K 2 , iL 2 ) and J = {j1, j2, j3}, the link product C = K⋆J L then allows us to determine the overall effect of the resulting multi-time agent-environment interaction. The res… view at source ↗
Figure 8
Figure 8. Figure 8: Recasting Lemma 2. ρ ⋆J ΥA, which involves only a subset of modes of ρ stitched to a subset of modes of A (a) can be recast as a full all-output to all-input J ′ -stitching of ρ ⊗ I (whose Choi state is ρ ′ ) and A ′ (·) = I ⊗ A(·) (whose Choi state is ΥA′ ), such that we can apply Lemma 1 to evaluate this partial stitch (b). In this case, J ′ constitutes three ‘smaller’ stitches: our original J (green lin… view at source ↗
Figure 9
Figure 9. Figure 9: Recasting Result 1. Given two circuit fragments A and B that are to be J-stitched (a), they can be recast into C (b). The Choi state of C can be found (c) and re-expressed as the state of ρ (denoted in the blue box) as an input to B ⊗ IB. Equivalently, ΥC can be obtained by evaluating the stitchings in (d). We note here that all the lines can denote a collection of modes, rather than a single mode. Proof. … view at source ↗
Figure 10
Figure 10. Figure 10: (a) Building non-Markovianity. In this section, we expound on how to evaluate the stitching of oE to i A 3 . The mode labels and TMSV squeezing dependences are illustrated in the figure. (b) Interacting a comb with a channel. Here, we link the process from the previous example to a multimode channel via two mode-stitches, j1 = (o A 1 , i′ 1) and j2 = (o ′ 1, iA 2 ). As per all the previous examples, for a… 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

69 extracted references · 6 linked inside Pith

  1. [1]

    Yang, Physical review letters123, 110501 (2019)

    Y. Yang, Physical review letters123, 110501 (2019)

    2019

  2. [2]

    Y. Xiao, Y. Yang, X. Wang, Q. Liu, and M. Gu, Phys. Rev. Lett.130, 240201 (2023)

    2023

  3. [3]

    Liu and Y

    Q. Liu and Y. Yang, Quantum8, 1571 (2024)

    2024

  4. [4]

    Di´ osi and L

    L. Di´ osi and L. Ferialdi, Physical review letters113, 200403 (2014)

    2014

  5. [5]

    R. R. Camasca and G. T. Landi, Physical Review A103, 022202 (2021)

    2021

  6. [6]

    Benedetti, M

    C. Benedetti, M. G. Paris, and S. Maniscalco, Physical Review A89, 012114 (2014)

    2014

  7. [7]

    Vasile, S

    R. Vasile, S. Maniscalco, M. G. A. Paris, H.-P. Breuer, and J. Piilo, Physical Review A84, 052118 (2011)

    2011

  8. [8]

    Chiribella, G

    G. Chiribella, G. M. D’Ariano, and P. Perinotti, EPL (Europhysics Letters)83, 30004 (2008)

    2008

  9. [9]

    Chiribella, G

    G. Chiribella, G. M. D’Ariano, and P. Perinotti, Physical review letters101, 060401 (2008)

    2008

  10. [10]

    Thompson, K

    J. Thompson, K. Modi, V. Vedral, and M. Gu, New Jour- nal of Physics20, 013004 (2018)

    2018

  11. [11]

    M. T. Quintino, Q. Dong, A. Shimbo, A. Soeda, and M. Murao, Physical Review A100, 062339 (2019)

    2019

  12. [12]

    Gutoski and J

    G. Gutoski and J. Watrous, inProceedings of the thirty- ninth annual ACM symposium on Theory of computing (2007) pp. 565–574

    2007

  13. [13]

    Chiribella, G

    G. Chiribella, G. M. D’Ariano, and P. Perinotti, Physical Review A—Atomic, Molecular, and Optical Physics80, 022339 (2009)

    2009

  14. [14]

    Taranto, S

    P. Taranto, S. Milz, M. Murao, M. T. Quintino, and K. Modi, arXiv preprint arXiv:2503.09693 (2025)

    Pith/arXiv arXiv 2025

  15. [15]

    K. Ried, M. Agnew, L. Vermeyden, D. Janzing, R. W. Spekkens, and K. J. Resch, Nature Physics11, 414 (2015)

    2015

  16. [16]

    J. Xing, T. Feng, Z. Fan, H. Ma, K. Bharti, D. E. Koh, and Y. Xiao, arXiv preprint arXiv:2306.04983 (2023)

    Pith/arXiv arXiv 2023

  17. [17]

    Kurdzia lek, P

    S. Kurdzia lek, P. Dulian, J. Majsak, S. Chakraborty, and R. Demkowicz-Dobrza´ nski, New Journal of Physics27, 013019 (2025)

    2025

  18. [18]

    S. Milz, F. A. Pollock, and K. Modi, Open Systems & Information Dynamics24, 1740016 (2017)

    2017

  19. [19]

    F. A. Pollock, C. Rodr ´ ıguez-Rosario, T. Frauenheim, M. Paternostro, and K. Modi, Physical Review A97, 012127 (2018)

    2018

  20. [20]

    Milz and K

    S. Milz and K. Modi, PRX Quantum2, 030201 (2021)

    2021

  21. [21]

    S. Milz, F. A. Pollock, and K. Modi, Physical Review A 98, 012108 (2018)

    2018

  22. [22]

    G. A. White, F. A. Pollock, L. C. Hollenberg, K. Modi, and C. D. Hill, PRX Quantum3, 020344 (2022)

    2022

  23. [23]

    G. A. White, K. Modi, and C. D. Hill, Physical Review Letters130, 160401 (2023)

    2023

  24. [24]

    Tanggara, M

    A. Tanggara, M. Gu, and K. Bharti, arXiv preprint arXiv:2405.17567 (2024)

    Pith/arXiv arXiv 2024

  25. [25]

    Kobayashi, H

    F. Kobayashi, H. Manabe, G. A. White, T. Far- relly, K. Modi, and T. M. Stace, arXiv preprint arXiv:2412.13739 (2024)

    Pith/arXiv arXiv 2024

  26. [26]

    S. L. Braunstein and P. van Loock, Reviews of Modern Physics77, 513 (2005)

    2005

  27. [27]

    Grosshans, G

    F. Grosshans, G. Van Assche, J. Wenger, R. Brouri, N. J. Cerf, and P. Grangier, Nature421, 238 (2003)

    2003

  28. [28]

    S.-H. Tan, B. I. Erkmen, V. Giovannetti, S. Guha, S. Lloyd, L. Maccone, S. Pirandola, and J. H. Shapiro, Physical review letters101, 253601 (2008). 10

    2008

  29. [29]

    Nair and M

    R. Nair and M. Gu, Optica7, 771 (2020)

    2020

  30. [30]

    M. G. Genoni, M. G. A. Paris, G. Adesso, H. Nha, P. L. Knight, and M. S. Kim, Phys. Rev. A87, 012107 (2013)

    2013

  31. [31]

    Boulebnane, M

    S. Boulebnane, M. P. Woods, and J. M. Renes, Phys. Rev. Lett.127, 010502 (2021)

    2021

  32. [32]

    Y. Guo, P. Taranto, B.-H. Liu, X.-M. Hu, Y.-F. Huang, C.-F. Li, and G.-C. Guo, Phys. Rev. Lett.126, 230401 (2021)

    2021

  33. [33]

    S. Zhu, X. You, A. Romanenko, and A. Grassellino, Two- tooth bosonic quantum comb for temporal-correlation sensing (2026), arXiv:2601.10916 [quant-ph]

    arXiv 2026

  34. [34]

    Groeblacher, A

    S. Groeblacher, A. Trubarov, N. Prigge, G. Cole, M. As- pelmeyer, and J. Eisert, Nature communications6, 7606 (2015)

    2015

  35. [35]

    Chen, C.-C

    P.-W. Chen, C.-C. Jian, and H.-S. Goan, Physical Review B—Condensed Matter and Materials Physics83, 115439 (2011)

    2011

  36. [36]

    Weedbrook, S

    C. Weedbrook, S. Pirandola, R. Garcia-Patron, N. J. Cerf, T. C. Ralph, J. H. Shapiro, and S. Lloyd, Reviews of Modern Physics84, 621 (2012), arXiv:1110.3234 [quant- ph]

    Pith/arXiv arXiv 2012

  37. [37]

    Vaidman, Physical Review A49, 1473 (1994)

    L. Vaidman, Physical Review A49, 1473 (1994)

    1994

  38. [38]

    Furusawa, J

    A. Furusawa, J. L. Sørensen, S. L. Braunstein, C. A. Fuchs, H. J. Kimble, and E. S. Polzik, science282, 706 (1998)

    1998

  39. [39]

    N. C. Menicucci, P. Van Loock, M. Gu, C. Weedbrook, T. C. Ralph, and M. A. Nielsen, Physical review letters 97, 110501 (2006)

    2006

  40. [40]

    For those interested, please see [9]

  41. [41]

    Adesso, S

    G. Adesso, S. Ragy, and A. R. Lee, Open Systems & Information Dynamics21, 1440001 (2014)

    2014

  42. [42]

    The Weyl operator displaces states in phase space, such that ˆD(⃗ α)† ˆXj ˆD(⃗ α) =ˆXj +x j and ˆD(⃗ α)† ˆPj ˆD(⃗ α) =ˆPj + pj

  43. [43]

    Fiurasek, Physical Review Letters89, 137904 (2002), arXiv:quant-ph/0204069

    J. Fiurasek, Physical Review Letters89, 137904 (2002), arXiv:quant-ph/0204069

    Pith/arXiv arXiv 2002

  44. [44]

    Giedke and J

    G. Giedke and J. I. Cirac, Physical Review A66, 032316 (2002)

    2002

  45. [45]

    A. S. Holevo, Journal of Mathematical Physics52(2011)

    2011

  46. [46]

    Explicitly,J A = (IA ∪O A)∩(I J ∪O J ) andJ B = (IB ∪ OB)∩(I J ∪O J )

  47. [47]

    Since no qumode is stitched more than once, the elemen- tary contractions act on disjoint mode pairs and hence commute

  48. [48]

    Altherr and Y

    A. Altherr and Y. Yang, Physical Review Letters127, 060501 (2021)

    2021

  49. [49]

    Huang, L

    Z. Huang, L. Lami, and M. M. Wilde, PRX Quantum5, 020354 (2024)

    2024

  50. [50]

    Yadin, F

    B. Yadin, F. C. Binder, J. Thompson, V. Narasimhachar, M. Gu, and M. Kim, Physical Review X8, 041038 (2018)

    2018

  51. [51]

    G. A. White, P. Jurcevic, C. D. Hill, and K. Modi, Phys- ical Review X15, 021047 (2025)

    2025

  52. [52]

    N. C. Menicucci, Physical review letters112, 120504 (2014)

    2014

  53. [53]

    Fukui, A

    K. Fukui, A. Tomita, A. Okamoto, and K. Fujii, Physical review X8, 021054 (2018)

    2018

  54. [54]

    M. T. Quintino, Q. Dong, A. Shimbo, A. Soeda, and M. Murao, Physical review letters123, 210502 (2019)

    2019

  55. [55]

    Yoshida, A

    S. Yoshida, A. Soeda, and M. Murao, Physical Review Letters131, 120602 (2023)

    2023

  56. [56]

    Y. Mo, L. Zhang, Y.-A. Chen, Y. Liu, T. Lin, and X. Wang, npj Quantum Information11, 32 (2025)

    2025

  57. [57]

    L. Hu, L. Zhang, X. Chen, W. Ye, Q. Guo, and H. Fan, Annalen der Physik533, 2000589 (2021)

    2021

  58. [58]

    re-order

    A. Serafini,Quantum continuous variables: a primer of theoretical methods(CRC press, 2023). 11 Appendix A: Re-ordering qumodes Recall that we originally set ˆR= ( ˆX1, ˆP1, . . . ,ˆXn, ˆPn) and⃗ α= (x 1, p1, . . . xn, pn)T ∈R 2n. In this notation, the qumodes are ordered in a specific manner according to the form of ˆRand⃗ α. As a consequence, the covaria...

    2023

  59. [59]

    Consider a quantum stateρ, with output qumodesO ρ, in the spaceD( N o∈Oρ Ho)that is equivalent toD( N i∈IA Hi)

    Lemma 1 Lemma 1(Stitching a state into a channel with link product (all outputs to all inputs)).Consider a CV channel Awith input and output qumodesI A andO A respectively, with Choi stateΥ A(r)onD( N i∈IA Hi ⊗ N o∈OA Ho). Consider a quantum stateρ, with output qumodesO ρ, in the spaceD( N o∈Oρ Ho)that is equivalent toD( N i∈IA Hi). Then, the J-stitching ...

  60. [60]

    Lemma 2 We note that Lemma 1 would still apply to the most general case where only a subsetO ρ is fed into a subset ofI A by considering insteadA ′(ρ′) whereρ ′ =ρ⊗Φ(r ′) andA ′(·) =I ⊗ A(·). We elucidate an explicit version of this case in the following: Lemma 2(Stitching a state into a channel with link product formula (partial outputs to partial inputs...

  61. [61]

    The idea is to first construct the channelCas per Fig

    Result Proof We return to the proof for Result 1. The idea is to first construct the channelCas per Fig. 9b, then find its Choi state, Υ C as per Fig. 9c, using the definition of the Choi state. We also know that Υ A ⋆J ΥB = Υ C by definition. From there, we make use of Lemma 1 to re-express the form of Υ C to obtain the form of Result 1. 17 Figure 9.Reca...

  62. [62]

    − 1 4 ⃗ αT A\J ⃗ αT B\J ΩT γiA 1 0 0γ oB 1 ! Ω ⃗ αA\J ⃗ αB\J !# ·exp

    Correspondence with Gaussian Link Product Integral Going a step further, we can show that the result obtained by using the link product algorithm yields exactly the same result as if we were to evaluate the integral in (D10), and that this accurately provides an expression for the resulting combination of two channels. Below, we proceed with evaluating th...

  63. [63]

    Correspondence with State-into-channel Evaluation Next, we wish to show that this resulting covariance matrix correctly represents the output state of feeding a single TMSV copy, Φ(r) throughB◦A. In this case whereA,Bare both Gaussian channels, their Choi states are characterized by ⃗dA = ⃗0 ⃗ νA ! (I11) ⃗dB = ⃗ νB ⃗0 ! (I12) and ΓA(r) = γiA 1 γiA 1 ,oA 1...

  64. [64]

    This circuit stitching is nowbidirectional, withJ={(o A 1 , i′ 1),(o ′ 1, iA 2 )}

    For both these fragments,ρ 0 =ρ L =ρ E are taken to be the vacuum state such that its covariance matrix isI 2. This circuit stitching is nowbidirectional, withJ={(o A 1 , i′ 1),(o ′ 1, iA 2 )}. We start with the resulting covariance matrix of the beam splitter-beam splitter process calculated in the previous section, given by ˜Γout in (K7). In this exampl...

  65. [65]

    As per all the previous examples, for all information to pass through the join,r ′, r′′ → ∞, corresponding toIJ ={i ′ 1, iA 2 }

    andj 2 = (o′ 1, iA 2 ). As per all the previous examples, for all information to pass through the join,r ′, r′′ → ∞, corresponding toIJ ={i ′ 1, iA 2 }. The Choi states of the circuit fragments involved are given in (c). ΓρE ⋆A⋆A =   Γ00 Γ02 Γ03 Γ04 Γ01 Γ† 02 Γ22 Γ23 0Γ † 12 Γ† 03 Γ† 23 Γ33 Γ34 Γ† 13 Γ† 04 0Γ † 34 Γ44 0 Γ† 01 Γ12 Γ13 0Γ 11  ...

  66. [66]

    We then follow the iterations of the link product algorithm

    - where we note the stitched modes together. We then follow the iterations of the link product algorithm. In the first iteration, forMConnect, the matrix is subdivided as γtot =   γ11 0 0 0−γ 12Λ −γ13Λ 0Γ 00 Γ02 Γ03 Γ04 Γ01 0Γ † 02 Γ22 Γ23 0 Γ† 12 0Γ † 03 Γ† 23 Γ33 Γ34 Γ† 13 −Λγ† 12 Γ† 04 0Γ † 34 Γ44 + Λγ22Λ Λγ23Λ −Λγ† 13 Γ† 01 Γ12 Γ13 Λγ† 23Λ Γ1...

  67. [67]

    =γ ¯X (2) −γ ¯X,y=2 (γy=2)−1γ† ¯X,y=2 . This will correspond to the second mode-stitching of the modes (o ′ 1, iA 2 ), and γres(y= 2) as outlined here will have a form corresponding to the qumode order (o ′ E, oE, iA 1 , oA 2 ). Evaluating the second iteration ofMConnectto findγ res(y= 2), and discarding the environment subsystemso E, o′ E (corresponding ...

  68. [68]

    Different Ordering What if we had used a different order ofJ-stitching? Previously, we evaluated the mode-stitching of (o A 1 , i′

  69. [69]

    first, followed by that of (o ′ 1, iA 2 ). If instead, we wanted to evaluate the J-stitching according toJ={(o ′ 1, iA 2 ),(o A 1 , i′ 1)}we will obtain fromPrepCMa covariance matrix of the form: γtot =   γ11 0 0 0 −γ13Λ−γ 12Λ 0Γ 00 Γ02 Γ03 Γ01 Γ04 0Γ † 02 Γ22 Γ23 Γ† 12 0 0Γ † 03 Γ† 23 Γ33 Γ† 13 Γ34 −Λγ† 13 Γ† 01 Γ12 Γ13 Γ11 + Λγ33Λ Λγ † 23Λ −Λγ†...