{"id":"5f7bde95-437f-4de8-a2e6-836b445f0cd9","arxiv_id":"2602.13183","paper_version":3,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":7.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":0,"one_line_summary":"Annihilation probabilities for colliding particles on a line are expressed exactly as a determinant with ghost-pair formal variables, collapsing to a Pfaffian for complete annihilation.","lead":"The paper proves exact determinant and Pfaffian formulas for the probability that annihilating random walks on a line produce a given number of collisions, survivors, and ghost positions. The tool is a 'ghost pair' method that keeps destroyed particles walking as invisible walkers, restoring the fixed-dimension structure that determinantal formulas need.","discovery_kind":"new_application","skeptic_critique":{"model":"deepseek-v4-flash","headline":"P2 (consecutive collision) is not just unverified: as literally stated it fails for the discrete-time random walk graph used in Corollary 5.5, so the advertised application is outside the proof's scope.","rationale":"The reader correctly identified P2 as the weakest assumption. I agree that Theorem 3.1's internal sign-reversing involution is coherent for abstract planar DAGs and that the Pfaffian section, while tersely proved, is plausible. However, the reader framed the issue as 'verification deferred to the companion paper'; the situation is sharper. The current paper itself contains the discrete random walk graph in Corollary 5.5, and on that graph the literal statement of P2 is violated by an elementary triple of paths. Consequently, either P2 is intended with an extra ordering/geometric qualification that is absent from Definition 2.4, or the main theorem does not apply to the discrete biased random walk advertised in Corollary 5.5. This is a concrete correctness risk in a named application, not a mere lack of reference. A small exhaustive check can decide the matter; if the counterexample is admissible, the proof is incomplete for a central claimed scope. I do not see a separate flaw in the Pfaffian sign bookkeeping that would independently change the verdict, so I keep the reader's conditional verdict rather than escalating to rejection before the test is run.","tokens_in":20189,"tokens_out":25217,"duration_ms":238526,"concrete_test":"Run an exact enumeration on the finite DAG with vertices (t,z), 0≤t≤3, z∈{-3,...,3}, edges to (t+1,z±1), and sources (-1,0),(0,0),(1,0). Check Definition 2.4(P2) for all triples of directed paths. The tuple P1:(-1,0)->(1,0), P3:(1,0)->(1,0), P2:(0,0)->(1,1)->(2,2) should be flagged as a violation if P2 is interpreted literally. If the authors intend an extra condition (e.g., targets ordered y1≺y2≺y3), rerun with exactly that condition; this settles whether the counterexample is admissible. If the violation persists, Corollary 5.5 lacks a valid proof and the abstract's claim for discrete random walks must be restricted.","verdict_should_be":"CONDITIONAL","load_bearing_attack":"The proof of Theorem 3.1 depends on the consecutive collision property P2 of Definition 2.4 in exactly the places where the involution needs the first crossing to involve adjacent particles: Lemma 4.15 and Proposition 4.18. The paper asserts P2 holds for random walks on Z and cites the companion paper [Śni26a], but no verification appears here. More concretely, on the spacetime graph used in Corollary 5.5 — vertices (t, z) with edges (t, z) -> (t+1, z±1) and transition weights (5.4) — the literal P2 appears false. Take sources x=-1, x'=0, x''=1; paths P1: -1 -> 0 and P3: 1 -> 0 meet at v=(1,0), while the path P2: 0 -> 1 -> 2 neither passes through v nor shares a vertex with P1 or P3 before time 1. Unless an implicit condition on target ordering or on the notion of 'meet' is added, this is a direct counterexample to P2. Since Theorem 5.3 and Corollary 5.5 inherit the annihilation theorem's axioms, the central claim's advertised application to discrete random walks on Z is not covered by the proof as written. This is a load-bearing gap, not a stylistic one.","agreement_with_reader":"partial"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper develops a combinatorial 'ghost pair' method for annihilating particles on a line. When two particles annihilate, both trajectories continue as invisible ghosts, so the number of trajectories remains n and a square determinant can be written. Theorem 3.1 expresses the weight of a fixed final state—survivor positions and ghost-pair positions—as a coefficient of a determinant whose ghost columns carry formal variables. The proof is organized through castings, attribution, rehearsal, and a sign-reversing involution. Section 5 uses a cancellative labeling to convert pairwise coalescence into complete annihilation and derives a Pfaffian formula, with applications to biased random walks on Z. The paper is self-consciously combinatorial and positions itself as complementary to the analytic Pfaffian results of Tribe–Zaboronski and Garrod et al.","tokens_in":20448,"tokens_out":25812,"duration_ms":245059,"significance":"If the central theorem is correct, it gives exact finite-time probabilities for prescribed annihilation outcomes, generalizes the Karlin–McGregor/LGV determinant to a setting with a varying number of visible particles, and provides a purely combinatorial explanation of the Pfaffian structure of annihilating systems. The manuscript has real strengths: the casting/rehearsal/involution architecture is spelled out in detail, the two- and four-particle examples are helpful, there are no fitted parameters, and the paper is honest about the role of ghost anonymity and about the computational nature of Appendix A. However, the consecutive-collision axiom P2 is both essential to the proof and, as stated, false for a graph used in the paper's own applications. Until that axiom is corrected and verified, the advertised scope is not covered by the proof.","major_comments":[{"comment":"The consecutive collision property (P2) is false as stated for the spacetime graph used in Corollary 5.5. Let vertices be (t,z) with edges (t,z) -> (t+1,z±1). Take sources x=-1, x'=0, x''=1. The paths P1: (0,-1)->(1,0) and P3: (0,1)->(1,0) meet at v=(1,0). The path P2: (0,0)->(1,1)->(2,2) neither passes through v nor intersects P1 or P3 before time 1. Thus P2 fails. The proof uses P2 precisely where the involution needs the first crossing to be between adjacent active particles: Lemma 4.15 assumes 'I<J adjacent in the active set', and Proposition 4.18 says 'By the consecutive collision property (P2), they are adjacent in the active set'. Without a correct P2, the sign-reversing involution on a non-adjacent crossing can break candidacy for a ghost pair involving an intervening particle, so the cancellation argument in Sections 4.6-4.8 is not valid for that graph. Since Theorem 5.3 and Cor","section":"Definition 2.4; Lemma 4.15; Proposition 4.18; Corollary 5.5"},{"comment":"The derivation of the Pfaffian sign is not given. After marginalizing over ghost positions and ghost sign patterns, each perfect matching receives contributions from k! ghost-pair numberings and 2^k sign patterns. The text says that the 1/k! factor absorbs the numberings and that summing over matchings gives Pf(A), but it never shows the sign of an individual matching after the determinant sign and the formal-variable signs are combined. Since the entire combinatorial content of the Pfaffian is the alternating sign (see Example 5.4's counterterm A13A24), this step needs an explicit lemma or calculation. This is likely fixable, but as written Theorem 5.3 is not fully proved.","section":"Section 5.4, proof of Theorem 5.3"}],"minor_comments":[{"comment":"The formal-variable algebra is under-specified. The paper should state explicitly that the variables are commutative (or define the noncommutative ordering rules), define t_j^+ and t_j^- as independent formal monomials, and give the relation for products such as τ^-_J τ^+_I that occur in the Leibniz expansion when row order is not the index order.","section":"Section 3.1, Eq. (3.1)"},{"comment":"The sentence 'The matching {(1,3),(2,4)} is not physically realizable' is too strong. Under the discrete-time random walk model of Corollary 5.5, particles 1 and 3 can meet at a vertex while particle 2 is elsewhere; the physical realizability claim fails unless an additional convention is imposed. Please rephrase this as a statement about cancellation in the Pfaffian expansion rather than an a priori physical impossibility.","section":"Example 5.4"},{"comment":"The appendix reports an exact-arithmetic inconsistency but does not include the matrix or a reproducibility statement. Since the claim is used to support the statement that ghost anonymity is essential, please provide the system, the code, or at least the full set of equations checked.","section":"Appendix A"},{"comment":"The introduction announces a Pfaffian formula for complete annihilation, while Section 5 states and proves a Pfaffian formula for pairwise coalescence. The connection via the cancellative labeling should be stated more explicitly at the point where the complete-annihilation version is first introduced, so the reader knows the two formulations are the same identity.","section":"Section 1.3.3 vs Section 5.4"}],"recommendation":"major_revision","confidential_remarks":"The manuscript relies heavily on two companion papers ([Śni26a], [Śni26b]) that were not available to me. This matters because the consecutive-collision property P2 is both essential and, as stated in this manuscript, false for the paper's own discrete random walk graph. The author should be asked to make P2 a precise, self-contained axiom and verify it for the claimed process class, or to restrict the theorems accordingly. I would also ask the editor to verify the availability and status of the companion papers, since the present paper's proof is not self-contained without them."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Colleague,\n\nHere's what you should know. The ghost-pair determinant is a genuinely new construction: each annihilated pair continues as an ordered pair of anonymous ghosts, formal sign variables filter the Leibniz expansion, and the whole thing collapses to a Pfaffian for complete annihilation. The discrete proof of Theorem 3.1 is worked out in real detail — castings, attribution, rehearsal, sign-reversing involution — and the examples are helpful. It is also honest about what it cannot do, e.g., the computational evidence for ghost anonymity in Appendix A.\n\nBut there is a load-bearing gap. The consecutive-collision property P2 (Definition 2.4) is false for the discrete-time random walk graph used in Corollary 5.5. Take sources -1, 0, 1; paths from -1 and 1 meet at (time 1, position 0), while the path from 0 through (1,1) neither passes through that vertex nor intersects the other paths before time 1. So the first crossing among active paths need not involve adjacent particles, and the proof's key steps — Lemma 4.15 and Proposition 4.18 — do not go through. The paper says P2 holds for random walks and cites the companion paper [Śni26a], but the literal statement here fails and no verification appears in this manuscript. That means the advertised applications to random walks, and the Pfaffian corollary, are not covered by the proof as written.\n\nLesser issues: the Pfaffian proof (Theorem 5.3) is a sketch — marginalizing over ghost positions and the sign bookkeeping are asserted rather than derived. The continuous-time treatment is deferred entirely to the companion paper, which is fine only if that paper genuinely proves the needed axioms. The formal-variable algebra in (3.1) is a bit under-specified, but the intended meaning is clear from the examples; that is minor.\n\nThe paper deserves a serious referee. The ideas are new and the formulas are probably correct, but the proof needs repair or a restricted scope. I would send it to review with a clear request to fix P2 (or weaken it and adjust the involution) and to make the companion-paper dependencies explicit. A reader in interacting particle systems or LGV combinatorics should read it; just don't rely on the random-walk corollary yet.\n\nFor me: reading group maybe; would cite after the fix. Recommend: engage, require revision.","headline":"New ghost-pair determinant and Pfaffian for annihilation; the central proof does not cover its advertised random-walk case because the consecutive-collision axiom P2 is false for that graph.","tokens_in":20975,"tokens_out":11256,"would_cite":true,"duration_ms":93116,"reading_group":"maybe","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["05A15","05A19","15A15","60C05","60J65","82C22"],"pacs":[],"model":"deepseek-v4-flash","headline":"Ghost pairs make annihilating particles exactly solvable on a line: every outcome probability is a single determinant, and complete extinction is a Pfaffian.","keywords":["annihilating random walks","ghost particles","determinantal formula","Pfaffian","pairwise coalescence","planar spacetime graph","exact finite-time probabilities"],"falsifier":"On a small planar lattice, e.g., four walkers on Z with T = 3 steps, enumerate all annihilation evolutions directly and compare the right-hand side of Theorem 3.1 for every final state; a single mismatch refutes the formula. A sharper test targets the hidden axiom: construct a planar DAG satisfying the crossing property but violating the consecutive-collision property and check whether the first-crossing involution still cancels failed castings; if a pairing of non-adjacent particles survives, the determinant includes spurious terms.","tokens_in":20004,"feed_emoji":"👻","tokens_out":4416,"duration_ms":40664,"temperature":0.7,"pith_summary":"Annihilating particles on a line have resisted exact probability calculations because every collision removes two walkers, leaving more starting points than final positions. This paper establishes that if destroyed particles keep walking as invisible ghost pairs, the number of trajectories stays fixed and every prescribed outcome—how many collisions, where survivors end, where ghosts end—has probability given by one determinant. For complete extinction, the determinant collapses to a Pfaffian built solely from pairwise annihilation probabilities, explaining the Pfaffian structure earlier work found through differential equations. The same formula yields an exact Pfaffian for the event that prescribed consecutive pairs coalesce. If correct, it gives finite-time, configuration-level probabilities uniformly for lattice walks, birth-death chains, and diffusions including Brownian motion.","feed_headline":"Ghost pairs turn annihilation into a determinant","feed_subtitle":"Exact probabilities for who survives and where they end, with a Pfaffian when everyone dies.","key_machinery":"The central object is the ghost pair: when two particles annihilate, both trajectories continue as anonymous invisible walkers joined as a numbered pair, so the entity count never drops below n. This restores the square-matrix structure needed for determinant methods. The proof operates on a planar directed acyclic graph with a crossing property and a consecutive-collision property, using a sign-reversing involution that swaps path segments at the first spurious crossing; the consecutive-collision property guarantees the crossing particles are adjacent, so the swap preserves the coefficient-extraction constraints.","core_discovery":"On a planar spacetime graph with n walkers, fix k annihilations, s = n−2k survivors at prescribed positions, and k ghost-pair endpoints. The paper proves that the probability of this exact outcome equals 1/k! times the coefficient of formal variables t_j^{±1} in det(M), where M's survivor columns are transition weights and its ghost columns carry variables whose product rule records which initial particle is higher-indexed. The proof's sign-reversing involution cancels all nonphysical assignments pairwise, leaving exactly the physical performances. When s = 0, marginalizing over ghost positions collapses the determinant into Pf(A), the Pfaffian of pairwise annihilation weights. The same mech","pith_inferences":["If the consecutive-collision property fails but the crossing property holds, the involution may still pair first crossings; testing the determinant formula on non-nearest-neighbor walks would show whether the adjacency hypothesis is truly necessary or just a proof convenience.","The appendix's n = 3 inconsistency suggests a structural barrier: no Karlin–McGregor-style expression can specify which particles annihilated, so exact identity of annihilating pairs is not a determinantal observable in this framework—a warning for any attempt to refine ghost labels.","Because marginalizing ghost positions turns the determinant into a Pfaffian only when survivors are absent, partial annihilation formulas have mixed determinant/Pfaffian structure; similar partial-collapse identities may hold in Pfaffian point process theory.","The method's applicability to arbitrary space-time-varying transition weights suggests exact formulas for disordered or inhomogeneous systems where no generator-based approach is available."],"forward_implications":["Exact finite-time probabilities for any prescribed annihilation outcome on any process satisfying the two planarity axioms: lattice walks, birth-death chains, and Brownian motion.","Complete extinction is determined by pairwise annihilation probabilities alone, so many-particle extinction becomes computable from two-particle data.","The Pfaffian coalescence formula follows from an annihilation theorem, giving exact probabilities for prescribed pairwise mergers and connecting coalescence and annihilation at the configuration level.","The determinant-to-Pfaffian collapse explains Pfaffian point-process structure for annihilating systems by a single combinatorial argument, without differential equations.","For Ising–Glauber domain walls and A + A → ∅ reactions, outcome-level probabilities—not just densities or correlations—are now available."],"fun_headline_variants":["Ghost particles unlock exact annihilation odds","Determinant formula for particle annihilation","Pfaffian emerges when all particles die","Ghosts turned collisions into math"],"cache_read_input_tokens":2304,"weakest_assumption_plain":"The consecutive-collision property—whenever paths from two non-adjacent starting positions meet, every intermediate path must cross one of them before the meeting—is load-bearing for the sign-reversing involution; the paper cites its verification for lattice walks, birth-death chains, and Brownian motion to a companion paper rather than proving it here.","fun_headline_variants_meta":{"raw":{"variants":["Ghost particles unlock exact annihilation odds","Determinant formula for particle annihilation","Pfaffian emerges when all particles die","Ghosts turned collisions into math"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000157,"raw_usage":{"total_tokens":1061,"prompt_tokens":748,"completion_tokens":313,"prompt_tokens_details":{"cached_tokens":256},"prompt_cache_hit_tokens":256,"prompt_cache_miss_tokens":492,"completion_tokens_details":{"reasoning_tokens":263}},"tokens_in":492,"tokens_out":313,"duration_ms":3499,"temperature":1.0,"reasoning_tokens":263,"cache_read_input_tokens":256,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-02T23:34:13.303842+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"On a small planar lattice, e.g., four walkers on Z with T = 3 steps, enumerate all annihilation evolutions directly and compare the right-hand side of Theorem 3.1 for every final state; a single mismatch refutes the formula. A sharper test targets the hidden axiom: construct a planar DAG satisfying the crossing property but violating the consecutive-collision property and check whether the first-crossing involution still cancels failed castings; if a pairing of non-adjacent particles survives, the determinant includes spurious terms.","supporting_citations":[],"review_version":1}