pith. sign in

arxiv: 2602.19120 · v2 · submitted 2026-02-22 · 🧮 math-ph · math.MP

Causal Architecture in Hidden Quantum Markov Models

Abdessatar Souissi , Abdessatar Barhoumi This is my paper

Pith reviewed 2026-05-15 20:58 UTC · model grok-4.3

classification 🧮 math-ph math.MP
keywords hidden quantum Markov modelscausal architecturequantum memoryqubit modelentangled liftingssequential processesdistinguishable processes
0
0 comments X

The pith

Reversing the order of emissions and transitions in hidden quantum Markov models produces distinguishable quantum processes at late times.

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

The paper introduces causal hidden quantum Markov models that swap the sequence of hidden-state updates and observations relative to standard models. In a qubit example with rotating hidden dynamics and sharp measurements, the two orders generate observable sequences that remain different no matter how the system is prepared and even at arbitrarily distant times. This separation disappears only when both architectures descend from entangled liftings of ordinary classical hidden Markov models, where the classical statistics coincide. The result supplies a concrete way to detect genuinely quantum memory effects in sequential data that cannot be removed by re-initialization or longer observation.

Core claim

The two causal architectures—emission then transition versus transition then emission—generally produce different quantum processes. These processes can be distinguished by measurements at arbitrarily late times, independent of the hidden-system initialization and even when the models begin in different states. The architectures become equivalent precisely when they arise from entangled liftings of classical hidden Markov models and therefore share identical classical statistics.

What carries the argument

Causal hidden quantum Markov models (cHQMMs) that reverse the conventional order of hidden updates and emissions.

If this is right

  • The two orders of operations cannot be reconciled by choosing special initial conditions or by waiting longer.
  • Quantum memory effects can be separated from classical ones by comparing the two causal architectures.
  • Entangled liftings of classical hidden Markov models erase the distinction and recover identical observable statistics.
  • Causal HQMMs provide a tool for identifying order-dependent quantum memory in sequential processes.

Where Pith is reading between the lines

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

  • The same reversal test could be applied to higher-dimensional or continuous-variable systems to check whether the distinction scales.
  • Experimental sequences generated by the two orders might serve as a benchmark for quantum simulators of hidden-state dynamics.
  • If the difference survives more general measurement schemes, it could constrain the design of quantum recurrent models used in time-series learning.

Load-bearing premise

The reported difference between the two architectures is shown only for a rotating qubit hidden state under sharp projective measurements.

What would settle it

Identical late-time output statistics for both architectures in the qubit model, regardless of initial state, would falsify the claimed distinguishability.

read the original abstract

We introduce a class of causal hidden quantum Markov models (cHQMMs) that reverse the usual order of hidden updates and emissions compared to conventional HQMMs. Using a simple qubit model with a rotating hidden state and sharp measurements, we show that these two architectures-emission then transition versus transition then emission-generally produce different quantum processes. They can be distinguished by measurements at arbitrarily late times, no matter how the hidden system is initialized, and even when the two models start from different initial states. This means that the two orders of operations lead to genuinely different observable behaviors that cannot be reconciled by waiting longer or by choosing special initial conditions. At the same time, we prove that the two architectures become equivalent when they arise from entangled liftings of classical hidden Markov models, sharing the same classical statistics. This identifies a clear dividing line between classical and genuinely quantum hidden memory. Our findings highlight causal HQMMs as a useful tool for studying and distinguishing quantum memory effects in sequential processes.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit. Tearing a paper down is the easy half of reading it; the pith above is the substance, this is the friction.

Referee Report

2 major / 2 minor

Summary. The manuscript introduces causal hidden quantum Markov models (cHQMMs) that reverse the conventional order of hidden-state transitions and emissions. Using an explicit qubit model with a rotating hidden state and projective measurements, it demonstrates that the two architectures (emission-then-transition versus transition-then-emission) produce observably distinct quantum processes that remain distinguishable at arbitrarily late times, independent of initialization. It further proves that the architectures become equivalent when realized as entangled liftings of classical hidden Markov models, thereby separating classical from genuinely quantum hidden memory.

Significance. If the explicit distinction holds and the equivalence proof is complete, the work supplies a concrete, falsifiable signature of causal order in quantum sequential processes and a precise dividing line between classical and quantum memory effects. The machine-checked or fully explicit qubit construction and the lifting equivalence are strengths that could inform quantum process tomography and quantum machine-learning models of memory.

major comments (2)
  1. [§3 (qubit model and late-time distinguishability)] The assertion that the two architectures 'generally produce different quantum processes' (abstract and §3) rests solely on explicit computation for one qubit with a rotating hidden state and sharp measurements. No theorem establishes that the observable process tensors (or late-time statistics) must differ for arbitrary Hilbert-space dimension, general CPTP maps on the hidden system, or arbitrary POVMs. This makes the scope of the 'generally' claim unclear and load-bearing for the central distinction result.
  2. [Equivalence theorem (likely §4)] The equivalence proof for entangled liftings of classical HMMs is stated as preserving classical statistics under order reversal, but the manuscript does not detail how the lifting construction handles the reversed causal order while keeping the marginal process tensor identical; an expanded derivation or explicit map would confirm the result is not limited to the specific classical case.
minor comments (2)
  1. [Introduction / §2] Notation for the process tensor and the two causal orders should be introduced with a single consistent diagram or equation block early in the text to aid readability.
  2. [§3] The manuscript should state explicitly whether the qubit example uses a fixed initial state or averages over initial states, and whether the late-time distinguishability holds for mixed states.

Simulated Author's Rebuttal

2 responses · 0 unresolved

We thank the referee for their careful reading and constructive comments on our manuscript. We address each major comment below and indicate the revisions we will incorporate.

read point-by-point responses

  1. Referee: [§3 (qubit model and late-time distinguishability)] The assertion that the two architectures 'generally produce different quantum processes' (abstract and §3) rests solely on explicit computation for one qubit with a rotating hidden state and sharp measurements. No theorem establishes that the observable process tensors (or late-time statistics) must differ for arbitrary Hilbert-space dimension, general CPTP maps on the hidden system, or arbitrary POVMs. This makes the scope of the 'generally' claim unclear and load-bearing for the central distinction result.

    Authors: We agree that the distinction between the two architectures is demonstrated via explicit computation in a specific qubit model with a rotating hidden state and projective measurements, rather than via a general theorem for arbitrary Hilbert-space dimensions, CPTP maps, or POVMs. In the revised manuscript we will update the abstract and Section 3 to replace the term 'generally' with phrasing that explicitly limits the claim to the family of models considered (e.g., 'in this class of qubit models with rotating hidden states'). We will also add a short discussion noting that while the example illustrates a mechanism by which causal order produces distinguishable late-time statistics, a general proof for broader classes of maps and measurements is left for future work. These changes will make the scope of the central result precise. revision: yes

  2. Referee: [Equivalence theorem (likely §4)] The equivalence proof for entangled liftings of classical HMMs is stated as preserving classical statistics under order reversal, but the manuscript does not detail how the lifting construction handles the reversed causal order while keeping the marginal process tensor identical; an expanded derivation or explicit map would confirm the result is not limited to the specific classical case.

    Authors: We appreciate the request for additional detail on the equivalence result. In the revised manuscript we will expand the derivation in Section 4. We will supply an explicit lifting map from an arbitrary classical hidden Markov model to the corresponding entangled quantum model for both causal orders, and we will show step-by-step that the marginal process tensors coincide. This expanded derivation will confirm that the equivalence holds for the general class of entangled liftings of classical HMMs. revision: yes

Circularity Check

0 steps flagged

No circularity; explicit qubit construction and separate equivalence proof are self-contained

full rationale

The paper's central claims rest on an explicit qubit model computation showing distinguishability at late times and a separate proof that the architectures coincide for entangled liftings of classical HMMs. Neither reduces to a self-definition, fitted parameter renamed as prediction, or load-bearing self-citation chain. The derivation chain is independent of its own outputs and does not invoke prior author work as an unverified uniqueness theorem.

Axiom & Free-Parameter Ledger

0 free parameters · 1 axioms · 1 invented entities

The central claim rests on standard quantum postulates for states and projective measurements together with the specific construction of a rotating qubit; no free parameters or new entities with independent evidence are introduced in the abstract.

axioms (1)
  • standard math Standard postulates of quantum mechanics for qubit states, unitary evolution, and sharp (projective) measurements
    Invoked to define the rotating hidden state and the emission process in the qubit example.
invented entities (1)
  • causal hidden quantum Markov models (cHQMMs) no independent evidence
    purpose: Models that reverse the order of hidden-state transition and emission
    Newly defined class whose properties are the subject of the paper

pith-pipeline@v0.9.0 · 5464 in / 1274 out tokens · 26097 ms · 2026-05-15T20:58:08.597191+00:00 · methodology

discussion (0)

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

Lean theorems connected to this paper

Citations machine-checked in the Pith Canon. Every link opens the source theorem in the public Lean library.

What do these tags mean?
matches
The paper's claim is directly supported by a theorem in the formal canon.
supports
The theorem supports part of the paper's argument, but the paper may add assumptions or extra steps.
extends
The paper goes beyond the formal theorem; the theorem is a base layer rather than the whole result.
uses
The paper appears to rely on the theorem as machinery.
contradicts
The paper's claim conflicts with a theorem or certificate in the canon.
unclear
Pith found a possible connection, but the passage is too broad, indirect, or ambiguous to say the theorem truly supports the claim.

Forward citations

Cited by 2 Pith papers

Reviewed papers in the Pith corpus that reference this work. Sorted by Pith novelty score.

  1. Quantum Viterbi Algorithm

    quant-ph 2026-05 unverdicted novelty 6.0

    A quantum Viterbi decoder for hidden quantum Markov models achieves strictly higher scores than any classical diagonal strategy on the same observations by optimizing over coherent quantum trajectories.

  2. Cocycle Actions on Hidden Quantum Markov Models: Symmetry Protection and Topological Order

    math-ph 2026-05 unverdicted novelty 5.0

    Symmetry actions on hidden quantum Markov models for 1D spin systems are classified by group-cohomology 2-cocycles, yielding a stochastic description of SPT order that matches the AKLT chain.

Reference graph

Works this paper leans on

45 extracted references · 45 canonical work pages · cited by 2 Pith papers

  1. [1]

    Accardi, L., Soueidy, E.G., Lu, Y.G., Souissi, A.: Hidden Quantum Markov processes. Infin. Dimens. Anal. Quantum Probab. Relat. Top. (2024)

    work page 2024

  2. [2]

    Annali di Matematica Pura ed Applicata (1923-), 184(3), 327-346 (2005)

    Accardi, L., Fidaleo, F., Entangled markov chains. Annali di Matematica Pura ed Applicata (1923-), 184(3), 327-346 (2005)

    work page 1923

  3. [3]

    Entangled Markov chains are indeed entangled

    Accardi, L., Matsuoka, T., Ohya, M. Entangled Markov chains are indeed entangled. Infinite Dimensional Analysis, Quantum Probability and Related Topics, 9(03), 379-390 (2006). 21

    work page 2006

  4. [4]

    Accardi, L.: On the noncommutative Markov property. Funct. Anal. Appl.9(1), 1–8 (1975)

    work page 1975

  5. [5]

    Accardi, L., Souissi, A., Soueidy, E.G.: Quantum Markov chains: A unification approach. Infin. Dimens. Anal. Quantum Probab. Relat. Top.23(2), 2050016 (2020)

    work page 2020

  6. [6]

    Degree of entanglement in Entangled Hidden Markov Models

    Accardi, L., Souissi, A., Rhaima, M. Degree of entanglement in Entangled Hidden Markov Models. Chaos, Solitons & Fractals, 196, (2025) 116389

    work page 2025

  7. [7]

    Barrett, J., Lorenz, R., Oreshkov, O.: Cyclic quantum causal models. Nat. Commun.12(1), 885 (2021)

    work page 2021

  8. [8]

    V., Pogosov, W

    Babukhin, D. V., Pogosov, W. V. The effect of quantum noise on algorithmic perfect quantum state transfer on NISQ processors. Quantum Information Processing, 21(1), (2022) 7

    work page 2022

  9. [9]

    Springer, Berlin Heidelberg New York (1986)

    Bratteli, O., Robinson, D.W.: Operator Algebras and Quantum Statistical Mechanics 1. Springer, Berlin Heidelberg New York (1986)

    work page 1986

  10. [10]

    A., Huang, W., Barlow, T

    Clark, L. A., Huang, W., Barlow, T. M., Beige, A. (2015). Hidden quantum markov models and open quantum systems with instantaneous feedback. In ISCS 2014: Interdisciplinary Symposium on Complex Systems (pp. 143-151). Cham: Springer International Publishing

    work page 2015

  11. [11]

    J., Sinayskiy, I., Petruccione, F

    David, I. J., Sinayskiy, I., Petruccione, F. Digital simulation of convex mixtures of Markovian and non-Markovian single qubit Pauli channels on NISQ devices. EPJ Quantum Technology, 11(1), (2024) 1-26

    work page 2024

  12. [12]

    A survey of NISQ era hybrid quantum-classical machine learning research

    De Luca, G. A survey of NISQ era hybrid quantum-classical machine learning research. Journal of Artificial Intelligence and Technology, 2(1), (2022) 9-15

    work page 2022

  13. [13]

    Topological quantum memory

    Dennis, E., Kitaev, A., Landahl, A., Preskill, J. Topological quantum memory. Journal of Mathematical Physics, 43(9), (2002) 4452-4505

    work page 2002

  14. [14]

    Elliott, T. J. Memory compression and thermal efficiency of quantum implementations of nondeterministic hidden Markov models. Physical Review A, 103(5), (2021) 052615

    work page 2021

  15. [15]

    J., Gu, M., Garner, A

    Elliott, T. J., Gu, M., Garner, A. J., Thompson, J. Quantum adaptive agents with efficient long-term memories. Physical Review X, 12(1), (2022) 011007

    work page 2022

  16. [16]

    Fanizza, M., Lumbreras, J., Winter, A.: Quantum theory in finite dimension cannot explain every general process with finite memory. Commun. Math. Phys.405(2), 50 (2024)

    work page 2024

  17. [17]

    Quantum Mach

    Ghasemian, E.: Stationary states of a dissipative two-qubit quantum channel and their appli- cations for quantum machine learning. Quantum Mach. Intell.5(1), 13 (2023)

    work page 2023

  18. [18]

    A vailable at SSRN 5274549 (2025)

    Ghysels, E., Morgan, J., Mohammadbagherpoor, H.: On quantum and quantum-inspired max- imum likelihood estimation and filtering of stochastic volatility models. A vailable at SSRN 5274549 (2025)

    work page 2025

  19. [19]

    Jamio/suppress lkowski, A. (1972). Linear transformations which preserve trace and positive semidefi- niteness of operators. Reports on mathematical physics, 3(4), 275-278

    work page 1972

  20. [20]

    Choi, M. D. Completely positive linear maps on complex matrices. Linear algebra and its applications, 10(3), (1975) 285-290. 22

    work page 1975

  21. [21]

    npj Quantum Inf.4, 17 (2018)

    Giarmatzi, C., Costa, F.: A quantum causal discovery algorithm. npj Quantum Inf.4, 17 (2018)

    work page 2018

  22. [22]

    (Eds.): States, Effects, and Operations: Fundamental Notions of Quantum Theory

    Kraus, K., B¨ ohm, A., Dollard, J.D., Wootters, W.H. (Eds.): States, Effects, and Operations: Fundamental Notions of Quantum Theory. Springer, Berlin Heidelberg (1983)

    work page 1983

  23. [23]

    Quantum8, 1232 (2024)

    Li, X.Y., Zhu, Q.S., Hu, Y., Wu, H., Yang, G.W., Yu, L.H., Chen, G.: A new quantum ma- chine learning algorithm: split hidden quantum Markov model inspired by quantum conditional master equation. Quantum8, 1232 (2024)

    work page 2024

  24. [24]

    Moln´ ar, L.,ˇSemrl, P.: Spectral order automorphisms of the spaces of Hilbert space effects and observables. Lett. Math. Phys.80(3), 239–255 (2007)

    work page 2007

  25. [25]

    PRX Quantum2, 030201 (2021)

    Milz, S., Strasberg, P.: Quantum stochastic processes and quantum non-Markovianity. PRX Quantum2, 030201 (2021)

    work page 2021

  26. [26]

    Monras, A., Beige, A., Wiesner, K.: Hidden quantum Markov models and non-adaptive read- out of many-body states. Appl. Math. Comput. Sci.3(1), 93–122 (2011)

    work page 2011

  27. [27]

    Mor, B., Garhwal, S., Kumar, A.: A systematic review of hidden Markov models and their applications. Arch. Comput. Methods Eng.28(3), 1947–1979 (2021)

    work page 1947

  28. [28]

    Nechita, I., Pucha/suppress la, Z., Pawela, /suppress L.,˙Zyczkowski, K.: Almost all quantum channels are equidis- tant. J. Math. Phys.59(5), 052201 (2018)

    work page 2018

  29. [29]

    Cambridge University Press, Cambridge (2010)

    Nielsen, M.A., Chuang, I.L.: Quantum Computation and Quantum Information. Cambridge University Press, Cambridge (2010)

    work page 2010

  30. [30]

    npj Quantum Inf.5(1), 50 (2019)

    Pirandola, S., Laurenza, R., Lupo, C., Pereira, J.L.: Fundamental limits to quantum channel discrimination. npj Quantum Inf.5(1), 50 (2019)

    work page 2019

  31. [31]

    Rabiner, L.R.: A tutorial on hidden Markov models and selected applications in speech recog- nition. Proc. IEEE77(2), 257–286 (1989)

    work page 1989

  32. [32]

    Doctoral dissertation, University of California, Davis (2024)

    Ragone, M.:SO(n) AKLT chains as symmetry protected topological quantum ground states. Doctoral dissertation, University of California, Davis (2024)

    work page 2024

  33. [33]

    arXiv:2601.16126 (2026)

    Sundar, R., Elliott, T.: Quantum dimension reduction of hidden Markov models. arXiv:2601.16126 (2026)

    work page arXiv 2026

  34. [34]

    Souissi, A.: Matrix product states as observations of entangled hidden Markov models. J. Stat. Phys.192(7), 88 (2025)

    work page 2025

  35. [35]

    arXiv:2512.18642 (2025)

    Souissi, A., Andolsi, A.: A hidden quantum Markov model framework for entanglement and topological order in the AKLT chain. arXiv:2512.18642 (2025)

    work page arXiv 2025

  36. [36]

    Soueidi, E.G.: Entangled hidden Markov models

    Souissi, A. Soueidi, E.G.: Entangled hidden Markov models. Chaos Solitons Fractals174, 113804 (2023)

    work page 2023

  37. [37]

    Srinivasan, S., Gordon, G., Boots, B.: Learning hidden quantum Markov models. In: Int. Conf. Artif. Intell. Statist., Lanzarote, Spain (2017)

    work page 2017

  38. [38]

    Taranto, P., Elliott, T.J., Milz, S.: Hidden quantum memory: Is memory there when somebody looks? Quantum7, 991 (2023). 23

    work page 2023

  39. [39]

    Forney, G.D.: The Viterbi algorithm. Proc. IEEE61(3), 268–278 (1973)

    work page 1973

  40. [40]

    IEEE Trans

    Viterbi, A.: Error bounds for convolutional codes and an asymptotically optimum decoding algorithm. IEEE Trans. Inf. Theory13(2), 260–269 (1967)

    work page 1967

  41. [41]

    Quantum5, 440 (2021)

    Wechs, J., Milz, S., Pollock, F., Modi, K.: Separating causal influence and correlations in quantum dynamics. Quantum5, 440 (2021)

    work page 2021

  42. [42]

    Quantum5, 435 (2021)

    Wechs, J., Milz, S., Pollock, F., Modi, K.: Quantum causal models: Markovian and non- Markovian processes. Quantum5, 435 (2021)

    work page 2021

  43. [43]

    Yan, F., Iliyasu, A.M., Liu, Z.T., Salama, A.S., Dong, F., Hirota, K.: Bloch sphere-based representation for quantum emotion space. J. Adv. Comput. Intell. Intell. Informatics19(1), 134–142 (2015)

    work page 2015

  44. [44]

    A., Pollock, F

    White, G. A., Pollock, F. A., Hollenberg, L. C., Hill, C. D., Modi, K. (2025). What can unitary sequences tell us about multi-time physics?. Quantum, 9, 1695

    work page 2025

  45. [45]

    Zonnios, M., Boyd, A., Binder, F.C.: Quantum generation of stochastic processes: spectral invariants and memory bounds. New J. Phys.27(6), 064507 (2025). 24

    work page 2025