{"id":"7139b713-1f7d-41a9-bef3-de96288a746d","arxiv_id":"2508.01891","paper_version":4,"verdict":"UNVERDICTED","confidence":"LOW","novelty_score":6.0,"correctness_risk":"unknown","formal_verification":"none","parameter_count":0,"one_line_summary":"A new extension of Markov renewal theory to multidimensional time is defined, with algebraic tools and an algorithm for solving its renewal equations.","lead":"This paper introduces a new class of stochastic processes, multi-time Markov Renewal chains, where time is multidimensional. The authors develop an algebraic framework based on the convolution product of multidimensional matrix sequences and provide an efficient algorithm for solving the associated renewal equations.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The framework's core depends on the convolutional inverse of I−Q; the abstract does not state conditions under which this inverse exists for arbitrary multidimensional sojourn distributions.","rationale":"The reader's weakest assumption correctly identified the convolutional inverse existence as the load-bearing condition. The abstract itself signals that existence is not a trivial given ('paying particular attention') and stops short of claiming a general result. In the absence of the full text, the safest verdict is UNVERDICTED, as the reader already concluded. Our concrete concern is not a demonstrated error but a missing condition that could invalidate the main results for a natural subclass of multidimensional Markov renewal chains. Since the paper is presented as a general extension, this gap matters; but because no full text is available, we cannot move the verdict to reject. The recommended concrete test—checking the theorem and running a degenerate zero-sojourn-time counterexample—would either confirm the assumption is harmless or expose a real limitation. The proposed test is feasible from the full paper and does not require external data.","tokens_in":673,"tokens_out":3176,"duration_ms":41987,"concrete_test":"Obtain the full text and locate the theorem or proposition that establishes existence of the convolutional inverse for I − Q. Check whether it states an explicit condition on Q(0) (e.g., spectral radius of Q(0) strictly less than 1) and whether that condition is satisfied for all allowed sojourn-time distributions. Then run the following counterexample: a two-state chain where each state jumps to the other with probability 1 and sojourn time identically equal to the zero vector in N^2. If the paper's framework either excludes this kernel without stating so or fails to produce a unique solution, the abstract's claim of 'arbitrarily selected' distributions is not supported.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The central construction solves the multi-time Markov renewal equation μ = ν + μ ∗ Q by applying the convolutional inverse (I − Q)^{−1}. For the theory to cover 'arbitrarily selected' multidimensional sojourn-time distributions, this inverse must exist and be computable for every kernel Q arising in the new class. The abstract says only that 'particular attention' is paid to existence, representation, and computation, but it does not assert a general existence theorem. In the convolution algebra of d-dimensional matrix sequences, a necessary condition for invertibility of I − Q is that the constant term I − Q(0) is invertible, where Q(0) is the matrix of probabilities of zero sojourn time. If the model permits positive probability of a zero vector sojourn time (e.g., a transition that occurs at the same time point in all dimensions), then Q(0) can have a row sum of 1, making I − Q(0) singular. In that case the renewal equation either has no solution or a non-unique one, and the algebraic method fails. The abstract gives no assumption excluding such kernels. Since the full text is unavailable, the existence proof cannot be checked; this is the single most load-bearing gap because every downstream result—the renewal equations and the Gauss–Jordan algorithm—presupposes that inverse.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper proposes a new class of stochastic processes called multi-time Markov Renewal chains, which extend classical Markov renewal theory by letting time evolve in multiple dimensions. The state space is taken to be finite, and sojourn times are allowed to follow multidimensional distributions. The development centers on an algebraic treatment of the convolution product of multidimensional matrix sequences, with particular attention to the existence, representation, and computation of the convolutional inverse. This inverse is then used to write down multi-time Markov renewal equations, and a novel adaptation of the Gauss-Jordan algorithm is proposed for efficient practical implementation. The abstract presents the framework as new and states that several algebraic properties of the convolution product are explored, but it does not provide proofs, definitions, or numerical results.","tokens_in":928,"tokens_out":3439,"duration_ms":39943,"significance":"If the technical claims hold, this is a genuinely novel extension of Markov renewal theory to multidimensional time, which could be relevant for systems with multiple independent time scales, such as coupled physical processes or multi-rate queueing models. The algebraic approach using convolutional inverses of multidimensional matrix sequences is elegant and may be computationally attractive, as suggested by the claimed Gauss-Jordan adaptation. However, the significance is conditional: the abstract alone does not allow verification of the existence and representation theorems for the convolutional inverse, nor the correctness of the renewal equations or the algorithm. The paper would be a meaningful contribution if the full text provides rigorous proofs and demonstrates computability on concrete examples; at present, the contribution is promising but unverified.","major_comments":[{"comment":"The central renewal equation, which in the conventional form reads mu = nu + mu * Q, requires the convolutional inverse (I - Q)^{-1}. The abstract says only that 'particular attention' is paid to the existence, representation, and computation of this inverse, but it does not state the conditions under which the inverse exists for the class of multidimensional sojourn-time distributions considered. This is load-bearing: if the model allows a positive probability of a zero sojourn-time vector, the constant term I - Q(0) can become singular, and the renewal equation may fail to have a unique solution. The manuscript should explicitly state assumptions on the support of the sojourn-time distribution (e.g., strictly positive sojourn times) or provide a general existence theorem with nontrivial conditions, because the subsequent results and the algorithm all presuppose this invertibility.","section":"Abstract"},{"comment":"The phrase 'sojourn times in the different states of the system to be arbitrarily selected from a multidimensional distribution' is ambiguous. It could be read as allowing any joint distribution, including distributions with mass at the zero vector, which would raise the invertibility issue noted above. Alternatively, the model may implicitly require positive sojourn times in all dimensions, which is the usual assumption in Markov renewal theory. This distinction should be made explicit, as it determines whether the convolutional inverse exists without additional restrictions.","section":"Abstract"},{"comment":"Because only the abstract was available for review, the existence proofs, the detailed construction of the convolutional inverse, the derivation of the renewal equations, and the description of the Gauss-Jordan adaptation could not be checked. In particular, it is impossible to verify whether the inverse is explicitly representable in the considered algebra of multidimensional matrix sequences, or whether the algorithm has the claimed efficiency and termination properties. The reader is therefore left with an unverified central claim, and the recommendation reflects this uncertainty.","section":"Abstract (general)"}],"minor_comments":[{"comment":"The phrase 'Markov Renewal theory' should be written as 'Markov renewal theory' for consistency with standard terminology.","section":"Abstract"},{"comment":"The name of the class, 'multi-time Markov Renewal chains', should use lowercase 'renewal' to match the standard spelling of 'Markov renewal chain'.","section":"Abstract"},{"comment":"The sentence 'where the possible number of states of a physical system is finite' could be rephrased as 'where the state space of the physical system is finite' to be more precise.","section":"Abstract"},{"comment":"The abstract states that the practical implementation is 'achieved very efficiently' without giving any complexity measure, benchmark, or comparison; either provide such evidence in the paper or soften the claim in the abstract.","section":"Abstract"}],"recommendation":"uncertain","confidential_remarks":"This review is based solely on the 300-word abstract because the full text was not supplied. The central technical claims—the existence and representation of the convolutional inverse, the derivation of the renewal equations, and the correctness of the Gauss-Jordan adaptation—are not verifiable from the abstract alone. The existence concern identified by the stress-test is real in the sense that the abstract does not state the required conditions, but it may well be resolved in the full text. I recommend that the editor obtain the full manuscript before making a decision, and that the authors be asked to clarify the support conditions for sojourn times and the invertibility conditions for the convolutional kernel."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Quick one, since we only have the abstract. My take: this is a real idea, not a repackaging. Multidimensional time for Markov renewal chains, with a convolution algebra of matrix sequences and a Gauss–Jordan adaptation to compute the inverse—that is a genuine extension, and if the algebraic machinery works out it gives a clean way to handle systems with several internal clocks. I like that they keep the state space finite on purpose and stay close to the applied setting. That's the right first move for a new class of processes. The abstract promises properties of the convolution, the inverse, renewal equations, and an algorithm. If those are delivered, it's a solid theoretical contribution to applied probability.\n\nThe soft spot is exactly where you'd guess. The whole framework leans on the convolutional inverse of I−Q, and the abstract says only that 'particular attention' is paid to existence, representation, and computation—not that a general existence theorem holds. If the sojourn-time distribution permits zero vector sojourn times, I−Q(0) can be singular and the inverse may not exist or may not be unique. The paper may well assume positivity or handle zero sojourn times separately, but the abstract doesn't say. That's the first thing I'd look for in the full text. The rest of the theory—the renewal equations and the algorithm—is downstream of that inverse, so if the existence conditions are narrow, the generality shrinks a lot. But I won't manufacture a fatal flaw from an abstract. The concern is real and testable, not a contradiction.\n\nI can't verify proofs or the algorithm without the full text, and neither can you. So the honest position is: promising, novel, unverifiable at this stage. That's exactly the situation where a serious referee is useful. I'd send it out, with instructions to focus the referee on the invertibility conditions and on whether the Gauss–Jordan adaptation actually terminates for the claimed class. The citation pattern I can't judge from an abstract, so no comment there.\n\nFor your reading group: maybe, once the full preprint is out. I wouldn't cite it until I've seen the inverse existence proof. But it deserves peer review, not a desk reject.","headline":"Abstract-only, so everything is provisional, but the multidimensional Markov renewal idea is genuinely novel and the inverse-existence question is the one load-bearing thing to check.","tokens_in":1365,"tokens_out":944,"would_cite":false,"duration_ms":11799,"reading_group":"maybe","serious_thinker":"unclear","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["60K15","60K05"],"pacs":[],"model":"deepseek-v4-flash","headline":"The paper introduces multi-time Markov renewal chains, where time is multidimensional, and develops algebraic renewal equations whose convolutional inverse is computed by a Gauss-Jordan adaptation.","keywords":["Markov renewal chains","multidimensional time","convolution product","convolutional inverse","multidimensional matrix sequences","renewal equations","Gauss-Jordan algorithm","stochastic processes"],"falsifier":"Take a finite-state chain whose sojourn-time distribution is supported only on a strictly positive multi-index, so the zero-index term of the convolution sequence is zero; then the convolutional inverse cannot exist, and the paper's Gauss-Jordan adaptation must either fail or fail to terminate, showing where the claimed framework's boundary lies.","tokens_in":497,"feed_emoji":"⏳","tokens_out":6631,"duration_ms":71001,"temperature":0.7,"pith_summary":"This paper introduces a new class of stochastic processes, multi-time Markov renewal chains, in which the sojourn times of a finite-state system are drawn from an arbitrary multidimensional distribution, so time advances along several axes at once. The central move is algebraic: multidimensional matrix sequences are combined by a convolution product, and the key operation is the convolutional inverse, whose existence, representation, and computation the paper analyzes. The multi-time Markov renewal equations are then written down and solved efficiently through a novel adaptation of the Gauss-Jordan algorithm to multidimensional sequences. A sympathetic reader would care because this offers a uniform, exact way to model and compute renewal systems in which multiple clocks—age, duration, coordinate time—run simultaneously, without resorting to simulation.","feed_headline":"New theory makes multidimensional-time renewal chains solvable","feed_subtitle":"Sojourn times can follow any multidimensional distribution, with renewal equations solved by a Gauss-Jordan variant.","key_machinery":"The central machinery is the convolution product of multidimensional matrix sequences and its associated convolutional inverse. A multidimensional matrix sequence is an array of matrices indexed by a vector of nonnegative integers, one entry per time dimension; convolution is the natural generalization of power-series multiplication, and the convolutional inverse is the sequence $\\mathbb{B}$ satisfying $\\mathbb{A} \\ast \\mathbb{B} = \\mathbb{I}$, where $\\mathbb{I}$ is the identity sequence taking the identity matrix at the zero multi-index and zero elsewhere. This inverse is what carries the argument: it reduces the multi-time Markov renewal equations to algebraic inversion instead of iterative or Monte Carlo solution. The accompanying novelty is a Gauss-Jordan elimination procedure adapted to these multidimensional sequences, which supplies the computational route to the inverse and therefore to the renewal solution.","core_discovery":"The paper's central claim is that Markov renewal theory extends coherently to multiple time dimensions while preserving its flexibility: each state sojourn time is drawn from an arbitrary multidimensional distribution, and the laws of the process are governed by multi-time Markov renewal equations built from the convolution product of multidimensional matrix sequences. The load-bearing algebraic result is the convolutional inverse of such a sequence, studied here in terms of existence, representation, and computation, and it is what makes the renewal equations explicitly solvable. The paper supplements the algebra with a practical method: an adaptation of the Gauss-Jordan elimination algorithm to multidimensional matrix sequences, giving an efficient implementation for finite state spaces. If the framework is right, systems whose evolution depends on several simultaneous time dimensions acquire a deterministic algebraic solution rather than an approximation.","pith_inferences":["A natural extension the authors do not pursue: the same convolutional inverse algebra should work over multivariate generating functions, where the inverse corresponds to division of formal power series, potentially simplifying proofs of existence via the zero-index term.","The existence boundary is probably a zero-index invertibility condition: if the matrix at the zero multi-index is invertible, the inverse should exist; if it is singular or zero, no inverse exists, which would delimit exactly where the Gauss-Jordan adaptation applies.","A direct validation experiment would instantiate a bivariate exponential sojourn-time distribution on a two-state chain, compute the renewal function by the Gauss-Jordan variant, and compare with simulation; close agreement would demonstrate the implementation, and any divergence would expose the inverse's computational limit."],"forward_implications":["Finite-state multi-time Markov renewal equations become solvable in closed algebraic form whenever the convolutional inverse exists, giving exact transition probabilities rather than simulation estimates.","The Gauss-Jordan adaptation offers a deterministic, efficient algorithm for computing sojourn-time and reliability characteristics of systems with multidimensional duration distributions.","Models with several concurrent clocks, such as age and calendar time or two spatial coordinates, can be expressed and solved within a single renewal equation framework.","Uniqueness and stability of the renewal solution follow from invertibility of the convolution sequence, so the algebraic structure takes over the role usually played by renewal-theoretic assumptions."],"supporting_citations":[],"fun_headline_variants":["Multi-dimensional time meets Markov renewal chains","Markov renewal chains extend to multi-time settings","New algebra solves multi-time renewal equations","Gauss-Jordan variant cracks multi-time renewal chains"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The entire framework depends on the assumption that the convolutional inverse of the multidimensional sojourn-time matrix sequence exists and is computable; if that inverse is absent for a relevant distribution, the renewal equations have no algebraic solution.","fun_headline_variants_meta":{"raw":{"variants":["Multi-dimensional time meets Markov renewal chains","Markov renewal chains extend to multi-time settings","New algebra solves multi-time renewal equations","Gauss-Jordan variant cracks multi-time renewal chains"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.00019,"raw_usage":{"total_tokens":1298,"prompt_tokens":862,"completion_tokens":436,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":478,"completion_tokens_details":{"reasoning_tokens":380}},"tokens_in":478,"tokens_out":436,"duration_ms":5340,"temperature":1.0,"reasoning_tokens":380,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-06T05:17:58.766659+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Take a finite-state chain whose sojourn-time distribution is supported only on a strictly positive multi-index, so the zero-index term of the convolution sequence is zero; then the convolutional inverse cannot exist, and the paper's Gauss-Jordan adaptation must either fail or fail to terminate, showing where the claimed framework's boundary lies.","supporting_citations":[],"review_version":1}