pith. sign in

arxiv: 2606.23372 · v1 · pith:3LIQEELTnew · submitted 2026-06-22 · 🪐 quant-ph

Bounding Classical and Quantum Correlations in Bayesian Networks with Quasiprobabilities

Paul Becsi , Matty J. Hoban This is my paper

Pith reviewed 2026-06-26 07:53 UTC · model grok-4.3

classification 🪐 quant-ph
keywords quasiprobabilitiesBayesian networksnon-signaling correlationscausal inferenceBell nonlocalitynested Markov modeltensor networks
0
0 comments X

The pith

Quasiprobabilities on unobserved nodes generate all non-signaling correlations in a broad class of Bayesian networks.

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

The paper examines quasiprobabilistic models of Bayesian networks in which distributions over hidden nodes obey normalization but may take negative values. It establishes that for many network structures these models produce every non-signaling distribution over the observed nodes, extending earlier results known only for Bell scenarios. The construction is linked to tensor-network factorizations of the joint distribution. This leads the authors to conjecture that the resulting set of observable marginals coincides with the nested Markov model that encodes the causal constraints of the network.

Core claim

Quasiprobabilistic models for Bayesian networks, obtained by replacing probability distributions on unobserved nodes with quasiprobabilities that respect only normalization, yield a set of observable correlations (the quasi set) that contains all non-signaling distributions for a broad class of networks; the same construction is conjectured to recover exactly the nested Markov model associated with the network.

What carries the argument

The quasi set formed by allowing quasiprobabilities (normalized but possibly negative) solely on unobserved nodes while retaining ordinary probabilities on observed nodes.

If this is right

  • The quasi set supplies an outer approximation to the quantum correlations permitted by any given network.
  • Tensor-network decompositions become a practical tool for describing the allowed correlations.
  • The conjecture equates the quasi set with the nested Markov model, thereby characterizing all non-signaling distributions compatible with the causal structure.

Where Pith is reading between the lines

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

  • The tensor-network representation may be reusable for analyzing other factorization problems in classical and quantum probability.
  • If the conjecture holds, linear programming over quasiprobabilities would compute the exact non-signaling bound for those networks without enumerating quantum strategies.

Load-bearing premise

That quasiprobabilities placed only on hidden nodes produce observable marginals that match or outer-bound the quantum set for the networks studied.

What would settle it

A non-signaling correlation over observed nodes in one of the considered networks that cannot be realized by any assignment of quasiprobabilities to the hidden nodes.

Figures

Figures reproduced from arXiv: 2606.23372 by Matty J. Hoban, Paul Becsi.

Figure 1
Figure 1. Figure 1: The bipartite Bell Scenario. the set of distributions consistent with it?". Of course, one needs to define mathematically what consistency means in this context. In the simplest setting where all variables are observed, consistency is given by the three equivalent Markov properties: the factorisation criterion, the local Markov property, and the global Markov property. If latent variables are present, clas… view at source ↗
Figure 2
Figure 2. Figure 2: Introducing noise variables, illustrated on the Triangle Scenario. [PITH_FULL_IMAGE:figures/full_fig_p008_2.png] view at source ↗
Figure 3
Figure 3. Figure 3: The correspondence between the 4-on-line Scenario and its underlying undirected graph. Edges become latent variables with the same domain size as the corresponding index. the 4-on-line Scenario. For some finite sets A, B, C, D we consider the order-4 real-valued tensors T(a, b, c, d), where a ∈ A, b ∈ B, c ∈ C, d ∈ D. We say such a tensor T decomposes with respect to T and bond dimensions (I, J, K) ∈ R 3 i… view at source ↗
Figure 4
Figure 4. Figure 4: Graph Comn. Set everything else in p ∗ new to behave the same as in p ∗ . Then p ∗ new also achieves p(xV ): p(xV ) = X xL Y v∈V ∪L p ∗ (xv|xpaG(v) ) = X xL\{l} Y l ′∈L\{l} p ∗ (xl ′) Y v∈V \ch(l) p ∗ (xv|xpaG(v) ) X xl p ∗ (xl) Y v∈ch(l) p ∗ (xv|xpaG(v) ) = X xL\{l} Y l ′∈L\{l} p ∗ new(xl ′) Y v∈V \ch(l) p ∗ new(xv|xpaG(v) )   X x + l p ∗ (x + l ) Y v∈ch(l) p ∗ (xv|xpaG(v)\{l} , x+ l ) + X x − l p ∗ (x… view at source ↗
Figure 5
Figure 5. Figure 5: The Triangle Scenario. By summing this equation over xV we obtain 1 = X {1≤ie≤re}e∈E Y e∈E ψe(ie) = Y e∈E X 1≤ie≤re ψe(ie). Therefore, p(xV ) = X {1≤ie≤re}e∈E Y v∈V p  xv|{xle = ie}e∈inc(v)  · Y e∈E ψe(ie) P 1≤ie≤re ψe(ie) . (27) Finally, we go back to the original correlation scenario G. Define the latent variable le corresponding to edge e in the graph G t to have domain {1, . . . , re} and quasiprobab… view at source ↗
Figure 6
Figure 6. Figure 6: The Triangle Scenario after extending the latents with a copy of the perfect correlation [PITH_FULL_IMAGE:figures/full_fig_p044_6.png] view at source ↗
Figure 7
Figure 7. Figure 7: The Common Ancestor Scenario. References 1. Sangmin Byeon, W.L.: Directed acyclic graphs for clinical research: a tutorial. Journal of Minimally Invasive Surgery 26(3), 97–107 (2023). doi: 10.7602/jmis.2023.26.3.97 2. García-Franco, J.D., Díez, F.J., Carrasco, M.Á.: Probabilistic graphical model for the evaluation of the emotional and dramatic personality disorders. Frontiers in Psychol￾ogy 13 (2022). doi:… view at source ↗
read the original abstract

Bell's theorem reveals that quantum theory is in tension with classical causal reasoning and, in particular, the notion of local causality. This is now understood as a particular example of non-classicality in the study of correlations in (Bayesian) networks with both unobserved and observed nodes: the correlations are probability distributions over the observed nodes. There is a great deal of work aiming to understand the bounds on quantum and classical correlations in such networks and one approach is to consider outer approximations to the former. Along these lines, we consider quasiprobabilistic models for Bayesian networks, which can be seen as classical models but the probability distributions involving unobserved nodes are "replaced" with quasiprobabilities that respect normalisation but not positivity. We denote the set of correlations resulting from these models as the quasi set. Such models have a history in the study of Bell-type non-classicality where it has been shown that they can produce all non-signalling correlations. We show a generalisation of this result for a broad class of networks, which motivates a conjecture that the quasi set recovers the so called nested Markov model. Our work utilises a connection to tensor network decompositions, which may be of independent interest.

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 paper introduces quasiprobabilistic models for Bayesian networks in which probability distributions on unobserved nodes are replaced by quasiprobabilities obeying normalization but not positivity. It shows that these models recover all non-signalling observable marginals for a broad class of networks, via an explicit connection to tensor-network decompositions. This generalization is used to motivate the conjecture that the resulting 'quasi set' exactly coincides with the nested Markov model.

Significance. If the conjecture holds, the quasiprobabilistic construction would supply a concrete outer approximation (or exact characterization) to the set of quantum correlations compatible with a given causal structure, extending the known recovery of the non-signalling set in Bell scenarios. The tensor-network link may be of independent interest for relating causal models to tensor decompositions. The approach is parameter-free once the network is fixed and directly falsifiable against explicit distributions.

major comments (2)
  1. [Abstract] Abstract and conjecture statement: the identification of the quasi set with the nested Markov model is presented only as a conjecture motivated by the generalization; no explicit construction is supplied showing that every distribution in the nested Markov model arises from some quasiprobability assignment on the hidden nodes (or vice versa).
  2. [Main text (generalization)] Generalization claim: while the tensor-network argument is said to establish recovery of all non-signalling marginals for a broad class of networks, the manuscript does not delineate the precise class of networks for which the result holds nor provide worked examples of networks and explicit quasiprobability assignments that realize the non-signalling set.
minor comments (2)
  1. [Introduction] The term 'quasi set' is used repeatedly before a formal definition appears; an early equation or boxed definition would improve readability.
  2. Notation for quasiprobability distributions on unobserved nodes should be distinguished typographically from ordinary probability distributions to avoid confusion with the classical case.

Simulated Author's Rebuttal

2 responses · 0 unresolved

We thank the referee for their careful reading and constructive feedback. We appreciate the positive assessment of the work's significance and address the major comments point by point below, with plans for revision where appropriate.

read point-by-point responses

  1. Referee: [Abstract] Abstract and conjecture statement: the identification of the quasi set with the nested Markov model is presented only as a conjecture motivated by the generalization; no explicit construction is supplied showing that every distribution in the nested Markov model arises from some quasiprobability assignment on the hidden nodes (or vice versa).

    Authors: We agree that the equivalence between the quasi set and the nested Markov model is presented strictly as a conjecture, motivated by the tensor-network generalization result but without a full bidirectional explicit construction. The manuscript establishes one direction (that quasiprobabilistic models recover all non-signalling marginals) for a broad class via the tensor-network link, which provides the motivation for the conjecture. We will revise the abstract and main text to state the conjectural status more explicitly and add a short discussion of the structural features (e.g., the ability of quasiprobabilities to match arbitrary marginals while respecting normalization) that support expecting the converse to hold. A complete proof remains an open question beyond the present scope. revision: partial

  2. Referee: [Main text (generalization)] Generalization claim: while the tensor-network argument is said to establish recovery of all non-signalling marginals for a broad class of networks, the manuscript does not delineate the precise class of networks for which the result holds nor provide worked examples of networks and explicit quasiprobability assignments that realize the non-signalling set.

    Authors: The referee is correct that the precise class of networks is described only as 'broad' without formal delineation and that no explicit worked examples with quasiprobability assignments are provided. In the revised manuscript we will define the class more rigorously as the set of Bayesian networks whose observed marginals admit a tensor-network decomposition allowing quasiprobability assignments on hidden nodes to reproduce any non-signalling distribution. We will also add at least two concrete examples: the Bell scenario (recovering the full non-signalling polytope) and a simple three-node chain with one hidden variable, including explicit quasiprobability values and the resulting observable marginals. revision: yes

Circularity Check

0 steps flagged

No significant circularity; quasiprobability models and tensor-network generalization are defined independently

full rationale

The paper defines the quasi set by replacing unobserved-node distributions with quasiprobabilities obeying only normalization (no positivity), a construction stated independently of any quantum or nested-Markov target. The generalization to a broad class of networks is shown via an explicit tensor-network connection that recovers all non-signalling marginals; this step does not reduce to a self-definition, fitted parameter renamed as prediction, or self-citation chain. The further claim that the quasi set recovers the nested Markov model is presented only as a conjecture motivated by the generalization, not as a proven identity. No load-bearing self-citations, ansatzes smuggled via citation, or renamings of known results appear in the derivation chain. The central results therefore remain self-contained against external benchmarks.

Axiom & Free-Parameter Ledger

0 free parameters · 1 axioms · 0 invented entities

Based on abstract only; the central construction rests on the domain assumption that quasiprobabilities suffice for hidden nodes.

axioms (1)
  • domain assumption Quasiprobabilities on unobserved nodes (normalization preserved, positivity dropped) can be used to generate observable correlations in Bayesian networks
    This is the defining step that creates the quasi set and enables the generalization claim.

pith-pipeline@v0.9.1-grok · 5740 in / 1121 out tokens · 24707 ms · 2026-06-26T07:53:40.158023+00:00 · methodology

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

61 extracted references · 33 canonical work pages

  1. [1]

    Journal of Minimally Invasive Surgery26(3), 97–107 (2023).doi: 10.7602/jmis.2023.26.3.97

    Sangmin Byeon, W.L.: Directed acyclic graphs for clinical research: a tutorial. Journal of Minimally Invasive Surgery26(3), 97–107 (2023).doi: 10.7602/jmis.2023.26.3.97

    work page doi:10.7602/jmis.2023.26.3.97 2023

  2. [2]

    Frontiers in Psychol- ogy13(2022).doi: 10.3389/fpsyg.2022.996609

    García-Franco, J.D., Díez, F.J., Carrasco, M.Á.: Probabilistic graphical model for the evaluation of the emotional and dramatic personality disorders. Frontiers in Psychol- ogy13(2022).doi: 10.3389/fpsyg.2022.996609

    work page doi:10.3389/fpsyg.2022.996609 2022

  3. [3]

    Journal of Economic Literature 58(4), 1129–1179 (2020)

    Imbens, G.W.: Potential Outcome and Directed Acyclic Graph Approaches to Causal- ity: Relevance for Empirical Practice in Economics. Journal of Economic Literature 58(4), 1129–1179 (2020)

    2020

  4. [4]

    Aca- demic Press (2020)

    Theodoridis, S.: Machine Learning: A Bayesian and Optimization Perspective. Aca- demic Press (2020)

    2020

  5. [5]

    The Annals of Mathematical Statistics 5(3), 161–215 (1934).https://www.jstor.org/stable/2957502

    Wright, S.: The Method of Path Coefficients. The Annals of Mathematical Statistics 5(3), 161–215 (1934).https://www.jstor.org/stable/2957502

    arXiv 1934

  6. [6]

    Cambridge University Press (2009)

    Pearl, J.: Causality. Cambridge University Press (2009)

    2009

  7. [7]

    Clarendon Press (1996)

    Lauritzen, S.L.: Graphical Models. Clarendon Press (1996)

    1996

  8. [8]

    The Annals of Statistics49(5) (2021).doi: 10.1214/21-aos2064

    Bongers, S., Forré, P., Peters, J., Mooij, J.M.: Foundations of structural causal mod- els with cycles and latent variables. The Annals of Statistics49(5) (2021).doi: 10.1214/21-aos2064

    work page doi:10.1214/21-aos2064 2021

  9. [9]

    Nature Com- munications12(1) (2021).doi: 10.1038/s41467-020-20456-x

    Barrett, J., Lorenz, R., Oreshkov, O.: Cyclic quantum causal models. Nature Com- munications12(1) (2021).doi: 10.1038/s41467-020-20456-x

    work page doi:10.1038/s41467-020-20456-x 2021

  10. [10]

    In: Proceedings of the Sixth Annual Conference on Uncertainty in Artificial Intelligence

    Verma, T., Pearl, J.: Equivalence and synthesis of causal models. In: Proceedings of the Sixth Annual Conference on Uncertainty in Artificial Intelligence. UAI ’90, pp. 255–

  11. [11]

    Elsevier Science Inc., USA (1990)

    1990

  12. [12]

    Springer New York, NY (1993)

    Spirtes, P., Glymour, C., Scheines, R.: Causation, Prediction, and Search. Springer New York, NY (1993)

    1993

  13. [13]

    In: Proceedings of the Eighteenth Conference on Uncertainty in Artificial Intelligence

    Tian, J., Pearl, J.: On the testable implications of causal models with hidden variables. In: Proceedings of the Eighteenth Conference on Uncertainty in Artificial Intelligence. UAI’02, pp. 519–527. Morgan Kaufmann Publishers Inc., Alberta, Canada (2002)

    2002

  14. [14]

    The Annals of Statistics46(6A) (2018).doi: 10.1214/17-aos1631

    Evans, R.J.: Margins of discrete Bayesian networks. The Annals of Statistics46(6A) (2018).doi: 10.1214/17-aos1631

    work page doi:10.1214/17-aos1631 2018

  15. [15]

    Morgan Kaufmann Publishers Inc., San Francisco, CA, USA (1988)

    Pearl, J.: Probabilistic Reasoning in Intelligent Systems: Networks of Plausible Infer- ence. Morgan Kaufmann Publishers Inc., San Francisco, CA, USA (1988)

    1988

  16. [16]

    Annals of Statistics51(2023).doi: 10.1214/22- AOS2253 45

    Richardson, T.S., Evans, R.J., Robins, J.M., Shpitser, I.: Nested Markov Properties for Acyclic Directed Mixed Graphs. Annals of Statistics51(2023).doi: 10.1214/22- AOS2253 45

    work page doi:10.1214/22- 2023

  17. [17]

    Behaviormetrika41, 3–39 (2014)

    Shpitser,I.,Evans,R.J.,Richardson,T.S.,Robins,J.M.:IntroductiontonestedMarkov models. Behaviormetrika41, 3–39 (2014)

    2014

  18. [18]

    The Annals of Statistics42(4), 1452–1482 (2014)

    Evans, R.J., Richardson, T.S.: Markovian Acyclic Directed Mixed Graphs For Discrete Data. The Annals of Statistics42(4), 1452–1482 (2014)

    2014

  19. [19]

    Bernoulli25(2) (2019).doi: 10.3150/17-bej1005

    Evans, R.J., Richardson, T.S.: Smooth, identifiable supermodels of discrete DAG mod- els with latent variables. Bernoulli25(2) (2019).doi: 10.3150/17-bej1005

    work page doi:10.3150/17-bej1005 2019

  20. [20]

    arXiv:1207

    Shpitser, I., Richardson, T.S., Robins, J.M., Evans, R.: Parameter and Structure Learning in Nested Markov Models, (2012). arXiv:1207 . 5058 [stat.ML].https : //arxiv.org/abs/1207.5058

    Pith/arXiv arXiv 2012

  21. [22]

    Physics Physique Fizika1, 195– 200 (1964).doi: 10.1103/PhysicsPhysiqueFizika.1.195.https://link.aps.org/doi/ 10.1103/PhysicsPhysiqueFizika.1.195

    Bell, J.S.: On the Einstein Podolsky Rosen paradox. Physics Physique Fizika1, 195– 200 (1964).doi: 10.1103/PhysicsPhysiqueFizika.1.195.https://link.aps.org/doi/ 10.1103/PhysicsPhysiqueFizika.1.195

    work page doi:10.1103/physicsphysiquefizika.1.195.https://link.aps.org/doi/ 1964

  22. [23]

    Freedman, S.J., Clauser, J.F.: Experimental Test of Local Hidden-Variable Theories. Phys. Rev. Lett.28, 938–941 (1972).doi: 10.1103/PhysRevLett.28.938.https : / / link.aps.org/doi/10.1103/PhysRevLett.28.938

    work page doi:10.1103/physrevlett.28.938.https 1972

  23. [26]

    arXiv:quant - ph / 0404076 [quant-ph].https : / / arxiv

    Cleve, R., Hoyer, P., Toner, B., Watrous, J.: Consequences and Limits of Nonlocal Strategies, (2010). arXiv:quant - ph / 0404076 [quant-ph].https : / / arxiv . org / abs/quant-ph/0404076

    Pith/arXiv arXiv 2010

  24. [27]

    arXiv:0712.0921 [quant-ph].https://arxiv.org/abs/0712.0921

    Greenberger, D.M., Horne, M.A., Zeilinger, A.: Going Beyond Bell’s Theorem, (2007). arXiv:0712.0921 [quant-ph].https://arxiv.org/abs/0712.0921

    Pith/arXiv arXiv 2007

  25. [28]

    Americal Journal of Physics58, 1131–1143 (1990).doi: 10.1119/1.16243

    Greenberger, D.M., Horne, M.A., Shimony, A., Zeilinger, A.: Bell’s theorem without inequalities. Americal Journal of Physics58, 1131–1143 (1990).doi: 10.1119/1.16243

    work page doi:10.1119/1.16243 1990

  26. [29]

    Physical Review Letters 82(7), 1345–1349 (1999).doi: 10.1103/physrevlett.82.1345

    Bouwmeester, D., Pan, J.-W., Daniell, M., Weinfurter, H., Zeilinger, A.: Observation of Three-Photon Greenberger-Horne-Zeilinger Entanglement. Physical Review Letters 82(7), 1345–1349 (1999).doi: 10.1103/physrevlett.82.1345

    work page doi:10.1103/physrevlett.82.1345 1999

  27. [30]

    Nature403, 515–519 (2000)

    Pan, J.-W., Bouwmeester, D., Daniell, M., Weinfurter, H., Zeilinger, A.: Experimental test of quantum nonlocality in three-photon Greenberger–Horne–Zeilinger entangle- ment. Nature403, 515–519 (2000)

    2000

  28. [32]

    New Journal of Physics17(3) (2015).doi: 10.1088/1367-2630/17/3/033002 46

    Wood, C.J., Spekkens, R.W.: The lesson of causal discovery algorithms for quantum correlations: causal explanations of Bell-inequality violations require fine-tuning. New Journal of Physics17(3) (2015).doi: 10.1088/1367-2630/17/3/033002 46

    work page doi:10.1088/1367-2630/17/3/033002 2015

  29. [34]

    New Journal of Physics14 (2012).doi: 10.1088/1367-2630/14/10/103001

    Fritz, T.: Beyond Bell’s Theorem: Correlation Scenarios. New Journal of Physics14 (2012).doi: 10.1088/1367-2630/14/10/103001

    work page doi:10.1088/1367-2630/14/10/103001 2012

  30. [35]

    Yagdjian, A

    Fritz, T.: Beyond Bell’s Theorem II: Scenarios with Arbitrary Causal Structure. Com- munications in Mathematical Physics341(2), 391–434 (2015).doi: 10.1007/s00220- 015-2495-5

    work page doi:10.1007/s00220- 2015

  31. [36]

    arXiv:0710

    Laskey, K.B.: Quantum Causal Networks, (2007). arXiv:0710 . 1200 [quant-ph]. https://arxiv.org/abs/0710.1200

    Pith/arXiv arXiv 2007

  32. [38]

    arXiv:1906

    Barrett, J., Lorenz, R., Oreshkov, O.: Quantum Causal Models, (2020). arXiv:1906. 10726 [quant-ph].https://arxiv.org/abs/1906.10726

    arXiv 2020

  33. [39]

    Quantum Information and Computation18(11 & 12) (2018)

    Rosset, D., Gisin, N., Wolfe, E.: Universal bound on the cardinality of local hidden variables in networks. Quantum Information and Computation18(11 & 12) (2018). doi: 10.26421/qic18.11-12

    work page doi:10.26421/qic18.11-12 2018

  34. [40]

    Journal of Causal Inference7(2) (2019).doi: 10.1515/jci-2017-0020

    Wolfe,E.,Spekkens,R.W.,Fritz,T.:TheInflationTechniqueforCausalInferencewith Latent Variables. Journal of Causal Inference7(2) (2019).doi: 10.1515/jci-2017-0020

    work page doi:10.1515/jci-2017-0020 2019

  35. [41]

    Journal of Causal Inference8(1), 70–91 (2020).doi: 10.1515/jci- 2018-0008

    Navascués, M., Wolfe, E.: The Inflation Technique Completely Solves the Causal Com- patibility Problem. Journal of Causal Inference8(1), 70–91 (2020).doi: 10.1515/jci- 2018-0008

    work page doi:10.1515/jci- 2020

  36. [42]

    Physical Review X11.doi: 10.1103/PhysRevX.11.021043

    Wolfe, E., Pozas-Kerstjens, A., Grinberg, M., Rosset, D., Acín, A., Navascués, M.: Quantum Inflation: A General Approach to Quantum Causal Compatibility. Physical Review X11.doi: 10.1103/PhysRevX.11.021043

    work page doi:10.1103/physrevx.11.021043

  37. [43]

    Clauser, J.F., Horne, M.A., Shimony, A., Holt, R.A.: Proposed Experiment to Test Lo- calHidden-VariableTheories.Phys.Rev.Lett.23,880–884(1969).doi:10.1103/Phys- RevLett.23.880.https://link.aps.org/doi/10.1103/PhysRevLett.23.880

    work page doi:10.1103/phys- 1969

  38. [44]

    arXiv:quan t-ph/0101005 [quant-ph].https://arxiv.org/abs/quant-ph/0101005

    Brassard, G.: Quantum Communication Complexity (A Survey), (2001). arXiv:quan t-ph/0101005 [quant-ph].https://arxiv.org/abs/quant-ph/0101005

    arXiv 2001

  39. [45]

    Physical Review Let- ters98(2007).doi: 10.1103/PhysRevLett.98.230501

    Acín,A.,Brunner,N.,Gisin,N.,Massar,S.,Pironio,S.,Scarani,V.:Device-Independent Security of Quantum Cryptography against Collective Attacks. Physical Review Let- ters98(2007).doi: 10.1103/PhysRevLett.98.230501

    work page doi:10.1103/physrevlett.98.230501 2007

  40. [46]

    Physical Review A71(2005).doi: 10.1103/PhysRevA.71.022101

    Barrett, J., Linden, N., Massar, S., Pironio, S., Popescu, S., Roberts, D.: Non-local correlations as an information theoretic resource. Physical Review A71(2005).doi: 10.1103/PhysRevA.71.022101

    work page doi:10.1103/physreva.71.022101 2005

  41. [47]

    Supplement8(4), 329–345 (1993)

    Tsirelson,B.S.:SomeresultsandproblemsonquantumBell-typeinequalities.Hadronic Journal. Supplement8(4), 329–345 (1993)

    1993

  42. [48]

    arXiv: 2001.04383 [quant-ph].https://arxiv.org/abs/2001.04383

    Ji, Z., Natarajan, A., Vidick, T., Wright, J., Yuen, H.: MIP*=RE, (2022). arXiv: 2001.04383 [quant-ph].https://arxiv.org/abs/2001.04383

    arXiv 2022

  43. [49]

    arXiv:1703.08618 [quant-ph].https://arxiv.org/abs/1703.08618

    Slofstra, W.: The set of quantum correlations is not closed, (2017). arXiv:1703.08618 [quant-ph].https://arxiv.org/abs/1703.08618. 47

    Pith/arXiv arXiv 2017

  44. [50]

    Communications in Mathematical Physics406(5) (2025).doi: 10.1007/s00220-024-05229-7

    Fu, H., Miller, C.A., Slofstra, W.: The Membership Problem for Constant-Sized Quan- tum Correlations is Undecidable. Communications in Mathematical Physics406(5) (2025).doi: 10.1007/s00220-024-05229-7

    work page doi:10.1007/s00220-024-05229-7 2025

  45. [51]

    doi: 10.1103/physrevlett.111.170403

    Al-Safi, S.W., Short, A.J.: Simulating all Nonsignaling Correlations via Classical or QuantumTheorywithNegativeProbabilities.PhysicalReviewLetters111(17)(2013). doi: 10.1103/physrevlett.111.170403

    work page doi:10.1103/physrevlett.111.170403 2013

  46. [52]

    EnhancedMolecularYieldfromaCryogenic Buffer Gas Beam Source via Excited State Chemistry

    Abramsky, S., Brandenburger, A.: The sheaf-theoretic structure of non-locality and contextuality. New Journal of Physics13(11), 113036 (2011).doi: 10.1088/1367- 2630/13/11/113036

    work page doi:10.1088/1367- 2011

  47. [53]

    arXiv: quant-ph/0508211 [quant-ph].https://arxiv.org/abs/quant-ph/0508211

    Barrett, J.: Information processing in generalized probabilistic theories, (2006). arXiv: quant-ph/0508211 [quant-ph].https://arxiv.org/abs/quant-ph/0508211

    Pith/arXiv arXiv 2006

  48. [54]

    Wigner, E.: On the Quantum Correction For Thermodynamic Equilibrium. Phys. Rev. 40, 749–759 (1932).doi: 10.1103/PhysRev.40.749.https://link.aps.org/doi/10. 1103/PhysRev.40.749

    work page doi:10.1103/physrev.40.749.https://link.aps.org/doi/10 1932

  49. [55]

    In: Quantum Implications: Essays in Honour of David Bohm

    Feynman, R.P.: Negative Probability. In: Quantum Implications: Essays in Honour of David Bohm. Ed. by B.J. Hiley and D. Peat, pp. 235–248. Methuen (1987)

    1987

  50. [56]

    arXiv:2212.11834 [cs.FL].https://arxiv.org/abs/2212.11834

    Yakaryılmaz,A.:Real-valuedaffineautomatacomputebeyondTuringmachines,(2022). arXiv:2212.11834 [cs.FL].https://arxiv.org/abs/2212.11834

    arXiv 2022

  51. [57]

    Approximate quantum fourier transform with o(n log(n)) t gates

    Barrett, J., de Beaudrap, N., Hoban, M.J., Lee, C.M.: The computational landscape of general physical theories. npj Quantum Information5(1) (2019).doi: 10.1038/s41534- 019-0156-9

    work page doi:10.1038/s41534- 2019

  52. [58]

    In: Proceedings of the 21st AAAI Conference on Artificial Intelligence (2006)

    Shpitser, I., Pearl, J.: Identification of Joint Interventional Distributions in Recursive Semi-Markovian Causal Models. In: Proceedings of the 21st AAAI Conference on Artificial Intelligence (2006)

    2006

  53. [59]

    Evans,R.J.:GraphsforMarginsofBayesianNetworks.ScandinavianJournalofStatis- tics43(3), 625–648 (2015).doi: 10.1111/sjos.12194

    work page doi:10.1111/sjos.12194 2015

  54. [60]

    Scandinavian Journal of Statistics30(1), 145–157 (2003).doi: 10.1111/1467-9469.00323

    Richardson, T.: Markov Properties for Acyclic Directed Mixed Graphs. Scandinavian Journal of Statistics30(1), 145–157 (2003).doi: 10.1111/1467-9469.00323

    work page doi:10.1111/1467-9469.00323 2003

  55. [61]

    arXiv:2411.11614 [quant-ph].https://arxiv

    Zhang, X., Wang, Y.: On the physics of nested Markov models: a generalized proba- bilistic theory perspective, (2024). arXiv:2411.11614 [quant-ph].https://arxiv. org/abs/2411.11614

    arXiv 2024

  56. [62]

    Information andInference:AJournaloftheIMA8(2),273–288(2018).doi:10.1093/imaiai/iay009

    Robeva, E., Seigal, A.: Duality of graphical models and tensor networks. Information andInference:AJournaloftheIMA8(2),273–288(2018).doi:10.1093/imaiai/iay009. eprint:https://academic.oup.com/imaiai/article- pdf/8/2/273/28864933/ iay009.pdf

    work page doi:10.1093/imaiai/iay009 2018

  57. [63]

    Foundations of Computational Mathematics16, 1423–1472 (2016).doi: 10.1007/s10208-016-9317-9

    Bachmayr, M., Schneider, R., Uschmajew, A.: Tensor Networks and Hierarchical Ten- sors for the Solution of High-Dimensional Partial Differential Equations. Foundations of Computational Mathematics16, 1423–1472 (2016).doi: 10.1007/s10208-016-9317-9

    work page doi:10.1007/s10208-016-9317-9 2016

  58. [64]

    Letters in Mathematical Physics112(4) (2022).doi: 10.1007/s11005-022- 01552-z

    Barthel, T., Lu, J., Friesecke, G.: On the closedness and geometry of tensor network state sets. Letters in Mathematical Physics112(4) (2022).doi: 10.1007/s11005-022- 01552-z

    work page doi:10.1007/s11005-022- 2022

  59. [65]

    arXiv:1801

    Ye, K., Lim, L.-H.: Tensor network ranks, (2019). arXiv:1801 . 02662 [math.NA]. https://arxiv.org/abs/1801.02662. 48

    Pith/arXiv arXiv 2019

  60. [66]

    Electronic Proceedings in Theoretical Computer Science171, 84–89 (2014).doi: 10.4204/eptcs.171.8

    Janotta, P., Lal, R.: Non-locality in theories without the no-restriction hypothesis. Electronic Proceedings in Theoretical Computer Science171, 84–89 (2014).doi: 10.4204/eptcs.171.8

    work page doi:10.4204/eptcs.171.8 2014

  61. [67]

    Journal of Algorithms11, 644–654 (1990)

    Håstad, J.: Tensor rank is NP-complete. Journal of Algorithms11, 644–654 (1990). doi: 10.1016/0196-6774(90)90014-6 49

    work page doi:10.1016/0196-6774(90)90014-6 1990