{"id":"d43eed1a-fe04-46f9-9b8d-f0e8b5f763c5","arxiv_id":"2412.04311","paper_version":2,"verdict":"ACCEPT","confidence":"MODERATE","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":0,"one_line_summary":"An unbounded generalization of Lorentzian metric spaces is defined, with Gromov-Hausdorff stability of the prelength and length properties and a canonical quasi-uniform structure.","lead":"The authors extend their earlier Lorentzian metric spaces framework, which models spacetime geometry using only a two-point distance function, to spacetimes that are unbounded and non-compact. The paper proves that such spaces retain key properties, including stability of the length-space property under Gromov-Hausdorff limits, and that they carry a canonical quasi-uniform structure.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"No significant objection identified.","rationale":"The reader's ACCEPT verdict is appropriate. The central claim (Thm. 6.17) is supported by a correct direct proof. The weakest assumption identified by the reader—countable generation—is a genuine domain restriction, but it is not a correctness risk for the results as stated. I noted a concern in the first proof's use of Thm. 5.11, but since the direct proof is valid and covers both the prelength and length cases (the latter via a short additivity argument), this does not undermine the theorem. The paper is a solid extension of the bounded theory, and no hidden assumption or circular step appears to invalidate the main conclusions.","tokens_in":51748,"tokens_out":36887,"duration_ms":327880,"concrete_test":"Verify whether the family F = {I_ε(p,q)} in the first proof of Theorem 6.17 satisfies the covering property of Theorem 5.11 for arbitrary finite sets; if it fails, the first proof is incomplete, but the direct proof stands independently.","verdict_should_be":"UNCHANGED","load_bearing_attack":"After a careful pass over the main theorem (Thm. 6.17) and its two proofs, I find no load-bearing flaw in the central claim. The direct proof of Thm. 6.17 is internally consistent: the quasi-correspondence data supply the points s^k_{j,n} needed for the approximate time functions, the bound (13) holds with ε_n = α(δ_{r_n} + 2^{-r_n}) → 0, and the diagonal construction produces an isocausal limit curve whose maximality follows from the additivity of maximal curves and the arbitrary-partition argument. The countable-generation restriction is a genuine scope limitation but is explicitly acknowledged and does not affect the validity of the stated results within that scope. A minor gap exists in the first proof of Thm. 6.17: the family F = {I_ε(p,q)} does not obviously satisfy the covering property required by Thm. 5.11 for arbitrary finite sets, so the invocation of Thm. 5.11 there is fragile. However, the direct proof is independent and complete, so this does not change the verdict.","agreement_with_reader":"partial"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper removes the boundedness assumption from the authors' earlier notion of Lorentzian metric space and develops a theory for unbounded Lorentzian metric spaces. The main objects are sets X with a Lorentzian distance d satisfying the reverse triangle inequality, a compactness/continuity condition on chronologically related sets, and a distinguishing condition. Adding countable generation gives Polish spaces (Prop. 3.20), time functions (Lemma 4.11), a limit curve theorem (Thm. 5.9), and a Gromov-Hausdorff theory for sequenced Lorentzian metric spaces based on quasi-correspondences. The central result is Theorem 6.17, stating that GH-limits of sequenced Lorentzian (pre)length spaces are again Lorentzian (pre)length spaces; two proofs are given. The paper also shows that every Lorentzian metric space carries a canonical quasi-uniformity whose associated order is the causal relation J (Thm. 7.10), and that sequenced spaces carry a canonical quasi-metric (Thm. 7.15). A comparison with the Braun--McCann framework is included.","tokens_in":51940,"tokens_out":10574,"duration_ms":100947,"significance":"If the results stand, this is an important step toward a minimalist, synthetic Lorentzian geometry that covers globally hyperbolic spacetimes and causets while supporting a notion of Gromov-Hausdorff convergence. The paper is careful and detailed, and the central GH-stability theorem is given two proofs, one of which is direct and essentially complete. The construction of a canonical quasi-uniformity and the explicit comparison with Braun--McCann are valuable contributions. The main limitations are the countable-generation restriction, which is explicitly acknowledged, and several local proof gaps that are repairable and do not undermine the central claims.","major_comments":[{"comment":"The family F = {I_epsilon(p,q) : p,q in X, epsilon > 0} does not satisfy the covering hypothesis of Theorem 5.11, which requires that for every finite set {p_1,...,p_m} there is F in F with I(p_1,...,p_m) subset F. In a standard chronological diamond, points arbitrarily close to the boundary of I(p,q) have d(p,x) or d(x,q) arbitrarily small, so no single epsilon > 0 covers I(p,q). Thus the first proof of Theorem 6.17 has a genuine gap. Since the direct proof is independent and complete, this does not affect the validity of the theorem, but the first proof should be withdrawn or amended.","section":"Section 6.5, first proof of Theorem 6.17"},{"comment":"The proof asserts that 'The set X'_r is compact', where X'_r = I_R(p'_1,...,p'_r) is the union of open chronological diamonds plus finitely many points; this set is not compact in general. The diagonal subsequence argument can be repaired by using the relative compactness of the individual diamond I(p'_i,p'_j) that contains all phi_m(x) for a fixed x, but as written the proof is invalid at this step. This gap affects the uniqueness of GH limits (Theorem 6.13), not the GH-stability theorem.","section":"Section 6.1, Proposition 6.9"},{"comment":"The statement 'By Proposition 2.8 the sets I(p_1,...,p_n) are compact' is inaccurate: Proposition 2.8 yields relative compactness only. The sigma-compactness conclusion is still correct because X = union_m closure(I(p_1,...,p_m)) and each closure is compact, but the proof should be corrected to say this explicitly.","section":"Section 3.3, Proposition 3.20"}],"minor_comments":[{"comment":"After deriving w <= z and tau(w) < tau(z), the text writes 'tau(zeta(w)) = tau(zeta(z)) = r'; the intended statement is tau(w) = tau(z) = r, since w and z are limits of the points zeta(q_{n_k}) and zeta(q_{m_k}).","section":"Lemma 5.5, proof"},{"comment":"The set X_m = I_R(p_1,...,p_m) is not compact in general; several arguments (e.g. Proposition 6.9 and the direct proof of Theorem 6.17) should explicitly invoke the compactness of the closed sets I_epsilon(...) or of the closures of chronological diamonds rather than compactness of X_m.","section":"Section 6, notation and subsequent uses"},{"comment":"The sentence 'the existence of an isochronal curve passing through every point implies J = I' should read J = closure(I), since J is closed while I is generally open.","section":"Corollary 5.27, proof"},{"comment":"The notation zeta_n(t) in sigma_n(t) is ambiguous because sigma_n(t) is a point of the quotient space B_{F_n}; it should say 'choose a representative zeta_n(t) in the equivalence class sigma_n(t)'.","section":"Theorem 5.11, proof, second part"},{"comment":"In the first case the proof concludes 'I(p,q) is compact', but the argument only establishes relative compactness; the final sentence of the second case correctly says 'relatively compact.' Please harmonize the wording.","section":"Proposition 2.8, proof"}],"recommendation":"minor_revision","confidential_remarks":"The paper is a substantial continuation of the authors' bounded theory, and the heavy reliance on [27] is natural for this line of work. The local proof gaps identified above are repairable without changing the main results; the direct proof of Theorem 6.17 appears sound. I recommend minor revision rather than acceptance in the current form because a few stated proofs contain false assertions about compactness."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Two things to know. First, the central claim—GH-stability of the (pre)length property for unbounded Lorentzian metric spaces (Thm 6.17)—holds up. The direct proof is consistent, and the stress-test objection about the first proof is real but minor. Second, this is an extension of the authors' own bounded theory [27], not a new framework. Most proofs are adaptations, but the unbounded version is nontrivial and the new quasi-uniformity material is substantial.\n\nWhat's actually new: the Polish property for countably generated spaces (Prop 3.20), the unbounded GH-convergence setup via sequenced spaces and quasi-correspondences, and the canonical quasi-uniformity/quasi-metric (Thms 7.10, 7.15). The paper also does an honest comparison with Braun-McCann and Müller. The writing is dense but careful; the proofs are explicit. Credit for giving two proofs of Thm 6.17—the first relies on a fragile covering property for the family F = {I_ε(p,q)}, but the second is self-contained and complete. That is the right way to handle a subtle point.\n\nSoft spots: the countable-generation assumption (Def 3.17) is the real scope condition. If a spacetime has a non-separable boundary, none of the sequenced-space machinery, Polish property, or GH-stability applies. The paper acknowledges this, but it means the title is slightly broader than the content. Also, the GH limit depends on the chosen generating sequence (Example 6.3); they show it, but anyone wanting a canonical limit will need extra structure. The self-citation load is heavy—many lemmas are lifted from [27] with minor modifications. That is acceptable when the base is solid, and [27] is solid, but it makes independent verification more work.\n\nWho should read: anyone working on synthetic Lorentzian geometry, causal structure, or GH-convergence of spacetimes. It is a solid within-subfield advance. It deserves a serious referee; I would send it to peer review and ask the referee to check the direct proof of Thm 6.17 carefully, but not to block on the first proof.","headline":"A careful, workmanlike extension of the authors' bounded Lorentzian metric space program; Thm 6.17 holds up, the countable-generation caveat is real but acknowledged.","tokens_in":52457,"tokens_out":2150,"would_cite":true,"duration_ms":21637,"reading_group":"maybe","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["53C50","53C23","54E15"],"pacs":[],"model":"deepseek-v4-flash","headline":"By dropping boundedness, the paper defines Lorentzian metric spaces from just three axioms on the Lorentzian distance and proves that Gromov-Hausdorff limits of Lorentzian (pre)length spaces remain Lorentzian (pre)length spaces.","keywords":["Lorentzian metric space","reverse triangle inequality","Gromov-Hausdorff convergence","Lorentzian length space","quasi-uniformity","time function","global hyperbolicity","countably generated space"],"falsifier":"A direct test is the paper's own example, $X=(0,\\infty)$ with $d(x,y)=(x-y)^+$ (Section 6.3): one can verify computationally that two different generating sequences for the approximating spaces give non-isomorphic GH-limits, which shows convergence is sequence-dependent. A stronger test of Theorem 6.17 would be to find any sequence of sequenced Lorentzian prelength spaces whose GH-limit satisfies the three axioms yet contains a chronologically related pair $x\\ll y$ with no isocausal curve connecting them; the paper's limit-curve theorem and diagonal construction are precisely what rule out such a limit, and exhibiting one would refute the stability claim.","tokens_in":51548,"feed_emoji":"⏳","tokens_out":11980,"duration_ms":111541,"temperature":0.7,"pith_summary":"This paper removes the boundedness assumption from the authors' earlier Lorentzian metric space theory and argues that three conditions already capture what is needed: the reverse triangle inequality for chronologically related pairs, continuity of the Lorentzian distance with relative compactness of chronological diamonds, and a distinguishing property for the distance function. The paper shows that adding a countable generating set makes every such space Polish, endowed with time functions, and suitable for a Gromov-Hausdorff theory in which balls and basepoints are replaced by truncated chronological regions and generating sequences. Its central result is that the Gromov-Hausdorff limit of a sequence of sequenced Lorentzian (pre)length spaces is again a Lorentzian (pre)length space, so the (pre)length property is stable under the convergence being defined. A canonical quasi-uniformity—in the sequenced case a canonical quasi-metric—is constructed directly from $d$, and it encodes both the topology and the extended causal relation $J$. This matters because it gives a coordinate-free, metric-like language for rough spacetimes, covering causets and smooth globally hyperbolic spacetimes, with global-hyperbolicity-type compactness built into the axioms.","feed_headline":"Gromov-Hausdorff limits preserve Lorentzian length structure","feed_subtitle":"Dropping boundedness, three distance axioms still give stable limits, time functions, and a canonical quasi-uniformity.","key_machinery":"The object that carries the argument is the chronological diamond $I(p,q)=\\{x\\in X : p\\ll x\\ll q\\}$, with $x\\ll y$ meaning $d(x,y)>0$; relative compactness of these diamonds is the unbounded substitute for the old boundedness condition and is what makes the topology locally compact and unique when the chronological boundary is empty (Propositions 2.8 and 3.3). Around this, the argument organizes around (a) the extended causal relation $J$, defined by $d(p,y)\\ge d(p,x)$ and $d(x,p)\\ge d(y,p)$ for all $p$, whose compactness on causal hulls expresses global hyperbolicity (Theorems 4.5–4.6); (b) the $(m,\\epsilon)$ quasi-correspondence between truncated regions $X_m=I(p_1,\\dots,p_m)\\cup\\{p_1,\\dots,p_m\\}$, which defines GH-convergence for sequenced spaces and is the tool through which the limit-curve and stability arguments run; and (c) the canonical quasi-uniformity generated by the maps $d_p,d^p:X\\to\\mathbb{R}$ with $\\mathbb{R}$'s order quasi-uniformity, which has $J$ as its associated order (Theorem 7.10) and, for sequenced spaces, is induced by an explicit quasi-metric (Theorem 7.15).","core_discovery":"The central claim, stated on the paper's own terms, is that a Lorentzian metric space needs no auxiliary topology, metric, or boundedness data: a set $X$ with $d:X\\times X\\to[0,\\infty)$ satisfying (i) the reverse triangle inequality $d(x,z)\\ge d(x,y)+d(y,z)$ when $d(x,y),d(y,z)>0$, (ii) continuity of $d$ with compactness of $I_\\epsilon\\cap(I(x,y)\\times I(x,y))$, and (iii) point-distinguishing by the functions $d_p$ and $d^p$, is already a well-behaved spacetime object. On such spaces, without chronological boundary, property (ii) is equivalent to relative compactness of all chronological diamonds, and the extended causal relation $J$ (the largest relation compatible with $d$ and the reverse triangle inequality) is closed, transitive, and antisymmetric. For countably generated spaces the topology is $\\sigma$-compact, second-countable, and Polish (Proposition 3.20), time functions exist (Lemma 4.11), and the paper's defined Gromov-Hausdorff convergence for sequenced spaces—via $(m,\\epsilon)$ quasi-correspondences—has the property that limits of Lorentzian (pre)length spaces remain Lorentzian (pre)length spaces (Theorem 6.17).","pith_inferences":["A reader could push the canonical quasi-metric of Theorem 7.15 toward a Gromov precompactness criterion: families of sequenced Lorentzian metric spaces with uniformly controlled quasi-metric diameters at each truncation level would be expected to have convergent subsequences, a statement the paper gestures at in its conclusions but does not prove.","The countable-generation restriction suggests where the theory should break: a Lorentzian metric space whose chronological boundary structure forces uncountable generating sets would fall outside the Polish, time-function, and GH-stability results, so physical models with non-separable boundary structure would need a separate convergence formalism.","Because the GH-limit depends on the chosen generating sequence (Section 6.3), the paper's convergence is really a convergence of spaces pointed by a countable causal net; one could try to gauge-invariantize the notion by quantifying over all generating sequences, at the cost of recovering the full isometry class rather than a single limit."],"forward_implications":["Any GH-limit of sequenced Lorentzian (pre)length spaces is again a Lorentzian (pre)length space, so the (pre)length property is a closed condition under the paper's convergence notion (Theorem 6.17).","Countably generated Lorentzian metric spaces are Polish and carry bounded time functions, so arguments needing complete metrizability and order-preserving potentials apply without an auxiliary Riemannian metric (Proposition 3.20, Lemma 4.11).","The canonical quasi-uniformity (and quasi-metric in the sequenced case) encodes both topology and causal order from the distance function alone, giving a uniform notion of curve convergence in which pointwise and uniform convergence of isocausal curves coincide (Theorem 7.10, Corollary 5.8).","Away from chronological boundaries the three axioms are equivalent to relative compactness of chronological diamonds with continuous $d$ (Theorem 2.14), so the definition is checkable in practice and reproduces global-hyperbolicity-type compactness.","The class of examples is large: smooth globally hyperbolic spacetimes and causets both satisfy the axioms (Proposition 2.4, Proposition 3.18), and under the standing assumptions of the time-separation literature the paper's Lorentzian length spaces correspond exactly to length metric spacetimes with every point on an isochronal curve (Corollary 5.27)."],"supporting_citations":[{"why":"The authors' companion bounded-case theory; supplies the distance-based formalism, the distinction metric, and the bounded GH-stability results that the present proofs extend.","marker":"[27]"},{"why":"The original Lorentzian length spaces definition; the main comparison target whose auxiliary-metric dependence the present axioms remove.","marker":"[21]"},{"why":"The time-separation setting whose length metric spacetimes, under its first standing assumption, are shown to correspond one-to-one with the paper's Lorentzian length spaces.","marker":"[7]"},{"why":"The metric-geometry reference for pointed Gromov-Hausdorff convergence that motivates replacing balls and basepoints with truncated chronological regions and generating sequences.","marker":"[8]"},{"why":"The earlier quasi-uniformizability result for globally hyperbolic smooth spacetimes that Theorem 7.10 generalizes to the synthetic setting.","marker":"[24]"},{"why":"Provides the quasi-uniformity and closed-preorder theory used to construct and interpret the canonical structure of Section 7.","marker":"[33]"},{"why":"The origin of using Lorentzian distance (maximal proper time) as the metric-geometry analogue that the paper's axioms formalize.","marker":"[10]"}],"fun_headline_variants":["Lorentzian spaces unbounded: Polish and GH-stable","Dropping boundedness: Lorentzian GH convergence works","Three axioms, no bounds: Lorentzian length spaces persist","GH limits preserve unbounded Lorentzian metric structure","Unbounded Lorentzian metrics: stable under Gromov-Hausdorff"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The load-bearing premise is the countable-generation condition: every structural result—Polish topology, time functions, and Gromov-Hausdorff stability—applies only to spaces admitting a countable generating set $G$ with $X=I(G)$, and the paper itself notes that not every Lorentzian metric space admits any generating set at all.","fun_headline_variants_meta":{"raw":{"variants":["Lorentzian spaces unbounded: Polish and GH-stable","Dropping boundedness: Lorentzian GH convergence works","Three axioms, no bounds: Lorentzian length spaces persist","GH limits preserve unbounded Lorentzian metric structure","Unbounded Lorentzian metrics: stable under Gromov-Hausdorff"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000256,"raw_usage":{"total_tokens":1609,"prompt_tokens":1014,"completion_tokens":595,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":630,"completion_tokens_details":{"reasoning_tokens":509}},"tokens_in":630,"tokens_out":595,"duration_ms":5999,"temperature":1.0,"reasoning_tokens":509,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-11T21:32:28.320036+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"A direct test is the paper's own example, $X=(0,\\infty)$ with $d(x,y)=(x-y)^+$ (Section 6.3): one can verify computationally that two different generating sequences for the approximating spaces give non-isomorphic GH-limits, which shows convergence is sequence-dependent. A stronger test of Theorem 6.17 would be to find any sequence of sequenced Lorentzian prelength spaces whose GH-limit satisfies the three axioms yet contains a chronologically related pair $x\\ll y$ with no isocausal curve connecting them; the paper's limit-curve theorem and diagonal construction are precisely what rule out such a limit, and exhibiting one would refute the stability claim.","supporting_citations":[{"cited_title":"Burago, Y","cited_arxiv_id":null,"evidence_quote":"The metric-geometry reference for pointed Gromov-Hausdorff convergence that motivates replacing balls and basepoints with truncated chronological regions and generating sequences."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Provides the quasi-uniformity and closed-preorder theory used to construct and interpret the canonical structure of Section 7."},{"cited_title":"Busemann","cited_arxiv_id":null,"evidence_quote":"The origin of using Lorentzian distance (maximal proper time) as the metric-geometry analogue that the paper's axioms formalize."}],"review_version":1}