{"id":"c05d9886-e6b2-45cc-a17b-23b024377c5d","arxiv_id":"2505.08785","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":7.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":0,"one_line_summary":"A two-state stochastic evolution is divisible between given times if and only if the earlier transition matrix lies in one of two explicitly described cone regions, with continuous curves crossing a critical diagonal necessarily becoming indivisible.","lead":"This paper gives a complete geometric map of when a two-state probabilistic process can be split into two consecutive steps and when it cannot. It uses cone-like regions in the space of transition matrices, offering a fresh visual language for quantum non-Markovianity and information erasure.","discovery_kind":"new_method","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Central cone inequalities (24)-(25) are asserted, not derived; the main theorem's correctness is unverified.","rationale":"I considered the SQC critique raised by the reader. The paper itself flags in Section V C that division events can coexist with nonzero coherences, so the quantum interpretation is already qualified; moreover the claim 'indivisible dynamics include quantum dynamics' can be supported by the direct embedding (3) of quantum channels into stochastic matrices, without relying on the full SQC. I therefore do not treat SQC as the most load-bearing issue. Instead, the central theorem's cone inequalities are asserted without proof. The whole geometric classification and all applications—the continuous-crossing criterion, the future cone, the information arrow—stand or fall with (24)-(25). Because no formal verification or derivation is supplied, this is the least secure condition. My spot-check of the algebra suggests the inequalities may be correct, but the manuscript has not demonstrated this; hence the reader's CONDITIONAL verdict is appropriate. The proposed numerical/algebraic test would settle whether the concern is merely presentational or substantive.","tokens_in":32681,"tokens_out":16614,"duration_ms":173493,"concrete_test":"Re-derive (24)-(25) from Eq. (23) by imposing nonnegativity of all four entries of Γ(t←t') under r+s-1>0 and simplifying; then independently verify with a dense numerical grid: sample 10^7 random (p,q,r,s) in (0,1)^4, compute Γ(t←t') by (23), and check that its stochasticity is exactly equivalent to (24a)-(24c) for r+s>1 and to (25a)-(25c) for r+s<1. Any mismatch—especially near p or q close to 0/1 or near r+s=1—would falsify the geometric characterization. Repeat the same check for the blue region condition (26).","verdict_should_be":"UNCHANGED","load_bearing_attack":"The central characterization depends entirely on the cone regions defined by inequalities (24) and (25) and the dual blue regions from (26). These are presented without derivation: after writing the transition matrix Γ(t←t') in (23), the text states 'The inequalities (24) and (25) correspond to (r,s) only in the gray regions of figure (5)' and gives no argument. Every subsequent result—the continuous-crossing criterion, the future cone (29), the information-time picture of Section IV—uses these regions. If (24)-(25) are incomplete or wrong, e.g., a missing boundary case, an implicit assumption about p,q near 0 or 1, or an algebra slip in reducing the four entrywise positivity constraints of (23) to the max/min form, the main classification collapses. The paper itself only handles the boundary p,q=0,1 in prose and treats the degenerate line r+s=1 separately, so the gap is not cosmetic. This is the least secure condition for the paper's central claim; the SQC quantum interpretation is secondary.","agreement_with_reader":"partial"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper studies divisibility of stochastic time-evolution for systems with two configurations. It parametrizes 2x2 left-stochastic matrices by diagonal entries (p,q), and asks, for a given pair Γ(t) and Γ(t'), whether a stochastic transition matrix Γ(t←t') exists with Γ(t)=Γ(t←t')Γ(t'). The central claim is a complete geometric criterion: for fixed Γ(t)=(p,q), the possible past matrices Γ(t')=(r,s) form two gray cone regions described by inequalities (24)-(25), the associated transition matrices form corresponding blue regions, and the possible futures under a division event at t are described by the region (29). The paper also shows that any continuous curve crossing the secondary diagonal p+q=1 must stop being divisible, introduces coordinates (X,T)=(p-q,1-p-q) in which the multiplication of stochastic matrices is (X,T)(χ,τ)=(X-χT,-Tτ), and interprets divisibility as progress toward the information-erasure line T=0. The last sections discuss symmetries, coarse graining and dilations, and applications to the Stochastic-Quantum Correspondence.","tokens_in":32867,"tokens_out":15250,"duration_ms":156584,"significance":"If the central inequalities are rigorously established, the paper provides a genuinely explicit necessary and sufficient condition for divisibility of two-configuration stochastic dynamics, with a clean geometric interpretation that is likely to be useful for non-Markovianity and for comparing classical and quantum evolution. The manuscript contains several checkable algebraic identities, notably the multiplication law (43), the oscillator example (59)-(60), and the coarse-graining formulas of Section VI, which are strengths. The main limitations are that the central cone inequalities are asserted rather than derived, the treatment of boundary and degenerate cases is informal, and the advertised quantum conclusions depend on an externally imported correspondence that the paper itself shows to be non-equivalent in an important sense.","major_comments":[{"comment":"The central theorem is not proved in the manuscript. After Eq. (23), the text states that the four column-sum-reduced positivity inequalities split according to the sign of r+s-1 and then simply lists (24)-(25). No derivation is shown that reduces the entrywise positivity of (23) to the max/min form, and no systematic treatment is given of the boundary cases p=0, p=1, q=0, q=1, r+s=1, or p+q=1, even though the gray regions are later used as closed regions including those boundaries. Since the claimed complete characterization and all subsequent results (the future cone, the continuity criterion, and the information-time picture) rest on these inequalities, the paper needs a full derivation, ideally in an appendix, together with a boundary-case lemma.","section":"III A, Eqs. (23)-(25)"},{"comment":"The future-cone region □Γ(t) is introduced by the same method, but Eq. (29) is again presented without derivation, with only the remark that the inequalities are the reverse of (24)-(25). The derivation requires Γ(t) to be invertible, and the exact form of the region for detΓ(t)>0 and detΓ(t)<0, including the boundary cases used in the examples of Figure 8, should be stated and proved. As it stands, the future cone is as much an assumption as the past cone.","section":"III A, Eq. (29)"},{"comment":"The displayed equality appears to be incorrect as written. Taking p=q=0.4 and θ=0 in (54), one has Θ unitary, det(Θ⊙Θ*)=detΓ=-0.2, and det(ΘΘ*)=1, so the left-hand side equals -0.2, while the right-hand side equals 1. The claimed relation with Birkhoff's contraction coefficient should therefore be rederived and corrected; this is important because the section advertises new connections between the SQC and classical contraction coefficients.","section":"V B, Eq. (56)"},{"comment":"The advertised quantum-level conclusions are conditional on the Stochastic-Quantum Correspondence of Refs. [5,6], which is imported from external work and not proved here. Section V C itself shows that, under the SQC prescription, a stochastic evolution can have division events at times when the density matrix has nonzero coherences, so stochastic divisibility is not equivalent to the usual quantum decoherence or channel-divisibility picture. To make the title and abstract claims defensible, the paper should either prove the needed SQC statement for the two-configuration case or explicitly present the main theorem as a theorem about stochastic matrices, with the quantum statements clearly flagged as conditional on the SQC.","section":"V and Abstract/Title"}],"minor_comments":[{"comment":"The sentence \"These inequalities are simple the reverse of inequalities (24) and (24)\" should cite (24) and (25), not (24) twice.","section":"III A, after Eq. (29)"},{"comment":"The symbol □ is used for the unit square, for the map H→HΓ(t), and for the set □Γ(t); the notation should be defined explicitly to avoid confusion.","section":"III A, Eq. (28) and Figure 6"},{"comment":"The sentence \"Any continuous curve crossing the secondary diagonal ... corresponds to an indivisible stochastic dynamics\" is too strong as worded: a continuous curve starting from the identity is divisible before the crossing and becomes indivisible only after it. The later discussion in III B is more careful, but the earlier statement should be rephrased.","section":"III, statement near Figure 2"},{"comment":"The example is described as \"right continuous with left limits, or càdlàg, continuous, divisible and information-decreasing almost everywhere\"; càdlàg functions are not continuous, so the wording should be corrected to avoid the apparent contradiction.","section":"IV A, Eq. (46)"},{"comment":"There are several typographical errors, including \"past past\" in the caption of Figure 5, \"he evolution\" in the caption of Figure 6, \"explanains\" in Section III C, and \"closed to the center\" in Section III C. The references [5,6] are arXiv preprints; if published versions exist, they should be cited.","section":"Throughout"}],"recommendation":"major_revision","confidential_remarks":"The core stochastic-matrix result is plausible and likely correct, but it is presented as a theorem without a derivation of the central inequalities. I would be willing to accept after the author supplies a complete derivation of (24)-(25) and (29), handles the boundary cases in a lemma, and corrects or carefully rederives Eq. (56). The dependence of the advertised quantum claims on the external SQC is a scope and framing issue that should be addressed, possibly by making the stochastic result primary in the abstract and title."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"What you should know: this paper gives a complete geometric characterization of when a 2x2 stochastic dynamics is divisible at a given time. The picture is nice: in the square of stochastic matrices, the possible pasts of a matrix Γ(t) lie in cone-like regions, and the possible transition matrices lie in dual regions. I checked the main inequalities (24)-(25) on a few numerical examples and they work. The claim that any continuous curve crossing the secondary diagonal is indivisible follows from the cone structure and is correct given (24)-(25). This is genuinely new as far as I can tell — the cited literature on generating sets and information-decrease equivalence does not contain this geometric picture.\n\nThe main soft spot is real but narrow: inequalities (24) and (25) are stated without derivation. They come from requiring the eight entries of (23) to be nonnegative, and with column sums equal to one that should reduce to four inequalities. But the text just says “the following expressions valid at the interior” and moves on. Since every subsequent result — the past and future cones, the continuous-crossing theorem, the T-coordinate picture — uses these inequalities, the paper’s central claim sits on an unproven algebraic step. The boundary cases are also waved through in prose. A referee should ask for a full derivation, including the boundary cases, and probably a cleaner statement of the degenerate cases.\n\nThe quantum part is secondary. The title promises “Stochastic-Quantum dynamics,” but the quantum interpretation is imported from Barandes’ SQC, which is itself speculative. The paper honestly shows that division events can occur while the SQC density matrix has nonzero coherences, so the correspondence is not a strict equivalence. The stochastic results stand on their own regardless of SQC; I would encourage the author to present them that way and soften the quantum claims.\n\nThere are also minor issues: some typos, and the future-cone inequalities (29) are stated as “the reverse” of (24)-(25) with no derivation. The coarse-graining section has an open question, which is fine.\n\nWho is this for? People working on non-Markovianity, divisibility of classical stochastic processes, and the SQC program. It deserves a serious referee, but only after the derivation gap is fixed. I’d send it to review with a request for a major revision that supplies the missing algebra.","headline":"A genuinely new geometric criterion for 2-state stochastic divisibility, but the derivation of the central inequalities is missing; worth refereeing if the author supplies the algebra.","tokens_in":33364,"tokens_out":3594,"would_cite":true,"duration_ms":33268,"reading_group":"yes","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["15B51","60J20"],"pacs":[],"model":"deepseek-v4-flash","headline":"For two-configuration systems, divisibility of a stochastic evolution is fully fixed by cone-shaped regions in the space of transition matrices; any continuous curve crossing the determinant-zero line is indivisible.","keywords":["divisible stochastic dynamics","indivisible dynamics","stochastic matrices","two configurations","information erasure","Stochastic-Quantum Correspondence","non-Markovianity","cone structure"],"falsifier":"Build a continuous curve in the unit square starting at the identity $(1,1)$ that crosses the secondary diagonal $p+q=1$, and check numerically whether every point $(p(t'),q(t'))$ with $0\\le t'\\le t$ satisfies the divisibility inequalities (24)-(25); the paper's criterion predicts that no such fully divisible crossing curve exists, so finding one would falsify the central claim.","tokens_in":32497,"feed_emoji":"🎲","tokens_out":10233,"duration_ms":89667,"temperature":0.7,"pith_summary":"This paper sets out to prove that, for a system with two configurations, the question of whether a stochastic evolution can be split at an intermediate time into two stochastic steps has a complete geometric answer: the allowed past and future transition matrices form cone-like regions in the unit square of $2\\times 2$ stochastic matrices. If the answer is yes at a given instant, that instant is called a division event; if no, the evolution is indivisible at that instant. The paper shows, as a direct corollary, that any continuous curve crossing the secondary diagonal $p+q=1$ corresponds to an indivisible stochastic dynamics, and that divisible dynamics point along a time coordinate associated with information erasure. This matters because indivisible stochastic dynamics are the ones that, under the Stochastic-Quantum Correspondence, can reproduce quantum-like phenomena while working only with probabilities.","feed_headline":"Cone geometry decides if a two-state process is divisible","feed_subtitle":"New bounds on 2x2 stochastic matrices mark where evolution splits into erasable, Markovian steps","key_machinery":"The load-bearing object is the parametrisation of a $2\\times 2$ stochastic matrix by its diagonal entries, identified with the point $(p,q)$ in the unit square. The argument then works with the coordinates $X=p-q$ and $T=1-(p+q)$, in which the determinant is simply $-T$, the identity sits at $T=-1$, the permutation at $T=1$, and matrix multiplication takes the compact form $(X,T)(\\chi,\\tau)=(X-\\chi T,-T\\tau)$. The cone inequalities (24)-(25) are obtained by imposing that $\\Gamma(t\\leftarrow t')=\\Gamma(t)\\Gamma^{-1}(t')$ in equation (23) be entrywise non-negative, which splits according to the sign of $\\det\\Gamma(t')=r+s-1$. This machinery converts the existence problem of a divisibility equation into checking whether a point lies in an explicit convex region, and it reveals the $T$ axis as an information-erasure time direction.","core_discovery":"On the paper's own terms, the discovery is a necessary and sufficient geometric criterion for divisibility of two-configuration stochastic dynamics. Writing the transition matrix as $\\Gamma(t)$ with diagonal entries $(p,q)$ in the unit square, the paper proves that a hypothetical intermediate matrix $\\Gamma(t')=(r,s)$ can appear in a division $\\Gamma(t)=\\Gamma(t\\leftarrow t')\\Gamma(t')$ if and only if $(r,s)$ lies in the gray regions defined by inequalities (24) and (25), with the corresponding $\\Gamma(t\\leftarrow t')$ lying in the associated blue regions of Figure 5. The criterion holds for open and closed systems under only the assumptions that there are two configurations, that initial probabilities can be freely ascribed, and that later probabilities are linear functions of the initial ones. A central corollary is that any continuous curve in the square that crosses the secondary diagonal $p+q=1$, where the determinant vanishes, corresponds to an indivisible stochastic dynamics. The paper also derives the reachable-future region, relates the geometry to the information-decrease criterion for divisibility, and gives examples of discontinuous dynamics whose divisible blocks are not themselves divisible, contrasting with quantum channels.","pith_inferences":["The cone picture suggests a natural quantitative measure of indivisibility: the distance of a point $(p(t),q(t))$ from the closest allowed past cone, which could be compared with existing non-Markovianity and channel-divisibility quantifiers.","The two-configuration criterion is likely the lowest layer of a hierarchy for $N\\ge 3$ inside the Birkhoff polytope; the paper's coarse-graining and dilation results imply that indivisibility of a large system can survive or disappear under coarse graining, so higher-dimensional cone structures will have to be stated relative to a chosen coarse graining.","A testable experimental corollary is that any two-level system whose transition probabilities trace a continuous curve across the secondary diagonal must contain an indivisible epoch, regardless of whether its density matrix shows coherences; this could be checked in any platform with single-shot state preparation and measurement."],"forward_implications":["A continuous stochastic evolution can be divisible for the whole path from the identity only while it stays in the closed gray region containing the identity; once it crosses the determinant-zero line, it is indivisible for the remainder of the path.","Divisibility at a time $t'$ for a given $\\Gamma(t)$ becomes a membership test: $\\Gamma(t')$ must lie in the gray cone, and $\\Gamma(t\\leftarrow t')$ in the blue cone, with the two regions related by the permutation symmetry.","Divisible dynamics have a defined arrow of time in matrix space, pointing toward the erasure line $T=0$; indivisible dynamics either move against that arrow or behave tachyonic relative to the cones.","Discontinuous stochastic dynamics can possess divisible blocks of evolution that are not themselves divisible at the block boundaries, a behaviour the paper contrasts with strictly bidivisible quantum channels.","For any number of configurations, continuity restricts allowed symmetry transformations of a stochastic dynamics to relabellings of the configurations, and each pair of divisors generates $N!$ further pairs via permutation."],"supporting_citations":[{"why":"Supplies the original notion of indivisible quantum channels that this paper transfers to stochastic dynamics.","marker":"[4]"},{"why":"Provides the Stochastic-Quantum Correspondence prescription used to map stochastic matrices to density matrices in Section V.","marker":"[5]"},{"why":"States the Stochastic-Quantum Theorem that every generalized stochastic system embeds in a unitary Hilbert-space dynamics.","marker":"[6]"},{"why":"Establishes the equivalence between divisibility and monotonic information decrease that the cone geometry translates.","marker":"[9]"},{"why":"Gives the divisibility theory of qubit channels, including Kraus-rank and bidivisibility results contrasted in Section V.","marker":"[10]"},{"why":"Baseline result that every $2\\times 2$ stochastic matrix is divisible, which the paper extends by showing the time-evolution need not be.","marker":"[15]"},{"why":"Provides the dilation-by-coarse-graining method that Section VI generalises to time-dependent and uncertain groupings.","marker":"[18]"},{"why":"Supplies the relative-entropy and contraction inequalities behind the information-erasure time coordinate $T$.","marker":"[26]"}],"fun_headline_variants":["Stochastic matrix cones reveal when evolution is irreducible","Two-state dynamics: cones decide if time splits into Markovian steps","Quantum dynamics are tachyonic in stochastic cone picture","Geometry of 2×2 matrices marks divisible vs indivisible dynamics","Cone bounds separate divisible and indivisible stochastic dynamics"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The cone test itself requires only freely ascribable initial probabilities and a linear relation between initial and later probabilities; the quantum reading of the title rests on the imported Stochastic-Quantum Correspondence, which the paper shows is not a strict equivalence because division events can occur while the associated density matrix still has nonzero coherences.","fun_headline_variants_meta":{"raw":{"variants":["Stochastic matrix cones reveal when evolution is irreducible","Two-state dynamics: cones decide if time splits into Markovian steps","Quantum dynamics are tachyonic in stochastic cone picture","Geometry of 2×2 matrices marks divisible vs indivisible dynamics","Cone bounds separate divisible and indivisible stochastic dynamics"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.001192,"raw_usage":{"total_tokens":4980,"prompt_tokens":1067,"completion_tokens":3913,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":683,"completion_tokens_details":{"reasoning_tokens":3833}},"tokens_in":683,"tokens_out":3913,"duration_ms":28256,"temperature":1.0,"reasoning_tokens":3833,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-15T21:47:41.160568+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Build a continuous curve in the unit square starting at the identity $(1,1)$ that crosses the secondary diagonal $p+q=1$, and check numerically whether every point $(p(t'),q(t'))$ with $0\\le t'\\le t$ satisfies the divisibility inequalities (24)-(25); the paper's criterion predicts that no such fully divisible crossing curve exists, so finding one would falsify the central claim.","supporting_citations":[{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Supplies the original notion of indivisible quantum channels that this paper transfers to stochastic dynamics."},{"cited_title":"c`adl`ag","cited_arxiv_id":null,"evidence_quote":"States the Stochastic-Quantum Theorem that every generalized stochastic system embeds in a unitary Hilbert-space dynamics."},{"cited_title":"Equivalence between divisibility and monotonic decrease of information in classical and quantum stochastic processes","cited_arxiv_id":"1408.7062","evidence_quote":"Establishes the equivalence between divisibility and monotonic information decrease that the cone geometry translates."},{"cited_title":"Generating Sets of Stochastic Matrices","cited_arxiv_id":"2411.18946","evidence_quote":"Baseline result that every $2\\times 2$ stochastic matrix is divisible, which the paper extends by showing the time-evolution need not be."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Provides the dilation-by-coarse-graining method that Section VI generalises to time-dependent and uncertain groupings."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Supplies the relative-entropy and contraction inequalities behind the information-erasure time coordinate $T$."}],"review_version":1}