{"id":"93d94ba4-dd09-4c30-9c7c-09e4da91ad6c","arxiv_id":"2505.01704","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":7.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":0,"one_line_summary":"Constructs the stochastic many-delta motions for the 2D N-body delta-Bose gas, strong Markov processes with lifetime solving an explicit singular SDE, and proves the no-triple-contacts property pathwise.","lead":"This mathematics paper constructs stochastic many-delta motions for the two-dimensional N-body delta-Bose gas: strong Markov processes realizing multiple two-body contact interactions, built by gluing local Girsanov transformations of one-delta motions. The no-triple-contacts property is proven at the path level, supplying the processes needed for the forthcoming Feynman-Kac formulas.","discovery_kind":"new_method","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Gap in proof of Theorem 3.4: the change-of-variables step omits nonnegative jump terms, so the Gronwall comparison and the NTC conclusion are not justified.","rationale":"The reader's conditional verdict rests on the external dependency on the unpublished companion preprint [7]. That is a real concern, but the present stress-test finds a more direct, internal gap: the proof of Theorem 3.4, which is the stated mechanism for no-triple-contacts, appears to contain an incorrect change-of-variables computation. The processes involved are cadlag, not continuous: ρ^(2) jumps upward exactly because the time-change γ has flat stretches, and A_ℓ jumps accordingly. For a convex φ_n, the correct formula contains a nonnegative jump correction. Equation (4.17) omits it and then uses the resulting inequality to feed Gronwall's lemma. Without the jump correction, the inequality is in the wrong direction and cannot yield Δ ≤ 0. This is not a matter of citing an unpublished result; it is an internal step of the proof. The rest of the construction—Girsanov transformations, local-time Itô formulas, concatenation—depends on the NTC statement, so the main theorem is not proved as written. The result may be repairable, and I am not claiming the theorem is false; but in its current form the central argument has a load-bearing gap. Therefore the appropriate disposition is REJECT (revise and resubmit after a correct proof of Theorem 3.4), rather than ACCEPT or a conditionality that merely asks for external verification. The good-faith reading of Section 4 shows deliberate care elsewhere, including the handling of cadlag paths in Lemma 4.1 and the statement about upward jumps in (4.13), which makes the omission in (4.17) especially consequential.","tokens_in":71777,"tokens_out":23726,"duration_ms":259321,"concrete_test":"Construct a path configuration in the setting of Theorem 3.4, e.g. n0 = 3, α = 0, with covariance σ(s) that forces the martingale sum Σ B^j to have zero quadratic variation on a small interval [u,v] while the individual B^j are nonconstant (for instance σ_{1,3} = σ_{2,3} = −1 and σ_{1,2} = 1 on that interval). Then γ jumps at ℓ = ⟨ρ,ρ⟩_u, ρ^(2) jumps upward, and Δ can have a negative jump with Δ_{s−} > 0. Compute both sides of the corrected change-of-variables formula at this jump: the omitted sum is strictly positive, so equation (4.17) is false. Then either provide a bound for this jump sum and rerun the Gronwall step, or conclude that Theorem 3.4 needs a new proof.","verdict_should_be":"REJECT","load_bearing_attack":"Section 4, Step 4, equation (4.17): the process Δℓ = ρ^(1),k_ℓ − ρ^(2)_ℓ is right-continuous with downward jumps, because ρ^(2)_ℓ = ρ_{γ(ℓ)} jumps upward on intervals where γ(ℓ−) < γ(ℓ) (Lemma 4.1 and (4.13)); the process A_ℓ in (4.9) has the same upward jumps. For the convex smoothing φ_n, the correct change-of-variables formula is φ_n(Δ_t) = φ_n(Δ_0) + ∫ φ'_n(Δ_{s−}) dΔ_s + Σ_s [φ_n(Δ_s) − φ_n(Δ_{s−}) − φ'_n(Δ_{s−}) ΔΔ_s], where the sum over jumps is nonnegative. Equation (4.17) drops this sum, states an equality, and then bounds the integral from above. Dropping a nonnegative jump term turns a lower bound into a claimed upper bound, so the displayed inequality, the subsequent Gronwall estimate, and the conclusion Δ ≤ 0 in (4.21) do not follow from the argument as written. Since Theorem 3.4 is the tool that supplies the no-triple-contacts property used in the concatenation construction, Theorem 3.1 is not fully established by the present proof.","agreement_with_reader":"disagree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper constructs, for every integer N ≥ 3, families of probability measures on path space that make the coordinate process a time-homogeneous strong Markov process with an absorbing state ∂, solving the singular SDE (3.9) up to a terminal time T∂ and satisfying the no-triple-contacts (NTC) property. The construction has two ingredients: a first class of processes, defined for contact-free initial data as a weighted average of the one-δ laws from the companion paper [7] (Definition 3.10, Proposition 3.11), and a second class, defined by a local Girsanov transformation and handling initial conditions with one contact (Proposition 3.15). The two classes are concatenated using NTC, which is deduced from a general no-simultaneous-contacts theorem (Theorem 3.4). Section 5 supplies the required Itô-formula-type semimartingale decompositions for the logarithms of the Radon–Nikodým derivative processes.","tokens_in":71899,"tokens_out":12023,"duration_ms":127852,"significance":"If the main theorem is correct, this is a substantial advance: it provides path-level stochastic motions for the two-dimensional many-body delta-Bose gas, with explicit SDEs involving Macdonald functions, a pathwise proof of the no-triple-contacts phenomenon, and a clean Markovian framework for the Feynman–Kac-type formulas announced for [8]. The paper is also commendably candid about its open points, such as the explicit admission in Remark 3.3(1◦) that it is not known whether P(T∂ = ∞) = 1. However, two issues prevent me from accepting the paper in its present form: the proof of the key NTC theorem contains an unhandled jump term in the comparison argument, and the construction is heavily dependent on the unpublished companion preprint [7] for many load-bearing fine properties. The first issue is local and appears repairable; the second is a verification concern that should be addressed explicitly.","major_comments":[{"comment":"The displayed equality in (4.17) is not valid for the càdlàg process Δ_ℓ^(k) = ρ_ℓ^(1),k − ρ_ℓ^(2). By Lemma 4.1 and (4.13), ρ^(2) and A have upward jumps whenever γ(ℓ−) < γ(ℓ), so Δ has downward jumps. For the convex regularizations φ_n, the correct Stieltjes change-of-variable formula contains the nonnegative jump sum Σ_s [φ_n(Δ_s) − φ_n(Δ_{s−}) − φ'_n(Δ_{s−}) ΔΔ_s], which is omitted in (4.17). Dropping a nonnegative jump term turns the asserted equality into a one-sided bound whose direction is not the one needed for the subsequent Gronwall argument. As written, the estimate (4.21), and hence the conclusion Δ ≤ 0, does not follow. Since Theorem 3.4 is the source of the NTC property used in Proposition 3.6, Lemma 3.19, Proposition 3.15(7◦), and the concatenation in Section 3.4, Theorem 3.1 is not fully established by the proof as written. The gap appears local and likely fixable — the monotonicity of the positive part at downward jumps should make the comparison go through once the jump part is treated separately — but the repair must be written out in full.","section":"Section 4, Step 4, Eq. (4.17)"},{"comment":"The paper imports a large number of results from the unpublished companion preprint [7] without reproving them: the SDEs (3.18)–(3.20), transition densities [7, Theorem 2.1], integrability bounds [7, Propositions 4.2 and 4.5], the local-time growth [7, (3.2)], and the change-of-measure relation [7, (2.7)] used in Lemma 3.13. These inputs are load-bearing: the weighted-average definition (3.46), every Girsanov transformation, and the limiting arguments in Section 5 are applied to processes whose fine properties come from [7]. Because [7] is cited as “Chen, Y.-T. (2024+)” and is not part of the submitted manuscript, the main theorem is conditional on an external document that cannot be checked from the present text. The authors should either include the needed statements and proofs, or make [7] available in a citable and ref ereed form before final acceptance.","section":"Throughout, especially Sections 3.1–3.3 and 5"}],"minor_comments":[{"comment":"Remark 3.16(2◦) refers to “Proposition 3.11 (5◦)”, but Proposition 3.11 has only properties (1◦)–(4◦); the intended reference is clearly Proposition 3.15(5◦).","section":"Remark 3.16(2◦)"},{"comment":"The text says that {L_t^i} is the local time of the C-valued process {Z_t^i} at level 0; the normalization (3.37) makes clear that it is the local time of |Z_t^i| at level 0, and this should be stated explicitly.","section":"Proposition 3.7(2◦)"},{"comment":"The reference measure P^{(0)}_{z0} is used from Lemma 3.13 onward but is never defined in the paper; a definition or a precise statement of its role as the law of the relevant Brownian motion should be added.","section":"Lemma 3.13 and following"},{"comment":"The proof uses Hölder conjugates (p_0, q_0) with 1 < q_0 < 1 + 1/√2 without explaining the origin of this range; a brief justification would improve readability.","section":"Proof of Proposition 3.15(5◦)"}],"recommendation":"major_revision","confidential_remarks":"The main mathematical problem is the unhandled jump term in the comparison proof of Theorem 3.4; I believe the issue is local and fixable, so I recommend major revision rather than rejection. The reliance on the unpublished companion [7] is also a policy question worth raising with the authors: the paper would be much easier to evaluate if the necessary results from [7] were stated with proofs or if [7] were posted in a complete, citable form."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"This is the paper that gives the 2D N-body delta-Bose gas program its path-level objects: for every N≥3 and every weight family w with at least two positive entries, Theorem 3.1 constructs stochastic many-δ motions — strong Markov processes with an absorbing state that solve the singular SDE (3.9) with Macdonald-function drifts up to a terminal time. The construction is explicit, not fitted: for no-contact initial conditions the law is a weighted average of the one-δ laws (Definition 3.10), and for the contact classes the paper uses explicit Girsanov transformations (Corollary 3.8). Theorem 3.4, the no-simultaneous-contacts comparison principle, genuinely extends Ichiba–Karatzas to many particles, and Section 5's Itô formulas for log-sums of Macdonald functions are heavy, honest analysis. The paper is also candid about what it cannot prove: finiteness of the terminal time is open, and lifetime finiteness is proved only for w-homogeneous couplings.\n\nThe main soft spot is external. Everything rests on the unpublished companion preprint [7]: transition densities, local-time normalizations, integrability bounds. The paper does not re-derive those, and an error there would propagate into the SDE (3.9) and the NTC proof. The author says so openly, but referees need [7] in hand. The Feynman–Kac link to the actual delta-Bose gas is also deferred to [8], so the significance is partly promissory.\n\nOn the stress-test claim about (4.17): I do not think it lands as a fatal gap. The change-of-variables step does omit a jump term, but the omitted term has the opposite sign from what the note claims. The Stieltjes integral in (4.17) evaluates φ'_n at the right-continuous values Δ_{ℓ'}, and Δ jumps downward because ρ^(2) jumps upward. For the convex φ_n, that makes the jump correction nonpositive, so the equality should be replaced by ≤, and the displayed bound on φ_n(Δ) from above is exactly the direction the proof needs. The Gronwall step and the conclusion Δ≤0 in (4.21) survive once that correction is made. The author should still be asked to write out the jump terms, but the fatal-gap reading does not hold up on close reading.\n\nWho this is for: people working on singular SDEs, Bessel-type comparison theorems, or the 2D delta-Bose gas. The paper deserves a serious referee, with the understanding that acceptance should be conditioned on [7] being public and on the Section 4 presentation being fixed.","headline":"The many-δ construction is a real step for the 2D delta-Bose gas program; the soft spots are the unpublished companion [7] and a repairable jump-term slip in Section 4, not the gap the stress test claims.","tokens_in":72593,"tokens_out":23344,"would_cite":true,"duration_ms":213664,"reading_group":"maybe","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["60J55","60J65","60H30","81S40"],"pacs":[],"model":"deepseek-v4-flash","headline":"For every N≥3, there are strong Markov processes of N planar particles that realize pairwise delta contact interactions at the path level and almost surely never bring three particles together before the terminal time.","keywords":["many-body delta-Bose gas","stochastic many-δ motions","no-triple-contacts","no-simultaneous-contacts","Girsanov transformation","singular stochastic differential equation","Macdonald functions","strong Markov processes"],"falsifier":"Take N=3 with all three pair weights $w_j>0$ and initial positions with no coincidences. The theorem asserts that the first time any pair distance is zero and the first time a second pair distance is zero are almost surely distinct, i.e. $P(\\exists t<T_\\partial \\text{ and distinct } j_1,j_2,j_3 \\text{ with } Z^{j_1}_t=Z^{j_2}_t=Z^{j_3}_t)=0$. A concrete disproof would be to show, for some eligible initial configuration and weights, that this event has positive probability—for example, by constructing a positive-probability set of Brownian paths under (3.9) on which two different pair distances hit zero at the same time.","tokens_in":71419,"feed_emoji":"⚛️","tokens_out":13845,"duration_ms":130922,"temperature":0.7,"pith_summary":"This paper constructs stochastic many-δ motions: random motions of N point particles in the plane that realize, at the path level, the contact interactions of the two-dimensional many-body delta-Bose gas. For every N≥3, every set of coupling constants, and every weight system with at least two active pairs, the main theorem produces probability measures on particle paths up to a terminal time under which the particles obey an explicit singular stochastic differential equation whose drift is written in terms of the Macdonald functions $K_0$ and $K_1$. The construction also proves a pathwise no-triple-contacts property: before the terminal time, three particles almost surely never occupy the same point simultaneously. The proof proceeds by changing measure locally via Girsanov transformations of the stochastic one-δ motions from the companion preprint [7], concatenating the resulting processes at contact-creation times, and closing the argument with a theorem on the impossibility of simultaneous contacts. If the theorem is right, these are the path-level stochastic counterparts of multiple two-body δ potentials in two dimensions, and they provide the process ingredient for the Feynman–Kac-type formulas developed in the next paper of the series.","feed_headline":"No three particles meet at once in this gas model","feed_subtitle":"Path-level stochastic motions now exist for every N≥3, with contacts always happening pairwise.","key_machinery":"The load-bearing object is the stochastic many-δ motion, defined as a strong Markov process satisfying the singular SDE (3.9)–(3.10) whose drift coefficients are written in terms of the Macdonald functions $K_0$ and $K_1$ and the weights $w$; before the first contact the drift is the logarithmic gradient of $\\Psi(z)=\\sum_j w_j K_0(\\sqrt{\\beta_j}\\,|z_{j'}-z_j|)$. The construction uses three tools. First, local Girsanov transformations of the one-δ motions from [7], with Radon–Nikodym derivative processes whose logarithms are expanded by an approximate Itô formula into finite-variation terms involving the local time of the one-δ process at zero, a continuous local martingale, and its quadratic variation (Proposition 3.7). Second, weighted-average probability measures (Definition 3.10): before any contact, the law is a weighted average of the laws of stochastic one-δ motions, with weights determined by $w$; this is a multi-particle analogue of the exponential change of measure used to construct $\\mathrm{BES}(0,\\beta^\\downarrow)$. Third, the no-simultaneous-contacts theorem (Theorem 3.4): a comparison argument showing that if finitely many nonnegative semimartingales with Bessel-type drifts and pairwise correlations bounded by $1$ are summed, the sum almost surely stays positive unless a single component hits zero; this yields no-triple-contacts. The pieces are assembled by the standard concatenation theorem for strong Markov processes.","core_discovery":"The central claim is the existence, for all N≥3, of the claimed stochastic many-δ motions. For eligible initial configurations—those in which at most one active pair of particles is initially in contact—the paper defines a family P of probability measures on paths in $C^N \\cup \\{\\partial\\}$ under which the coordinate process is a time-homogeneous strong Markov process with an absorbing state $\\partial$. Up to the terminal time $T_\\partial$, the process solves the singular SDE (3.9)–(3.10), whose drift before contact is the logarithmic gradient of $\\Psi(z)=\\sum_j w_j K_0(\\sqrt{\\beta_j}\\,|z_{j'}-z_j|)$, equivalently expressed through the Macdonald functions $K_0$ and $K_1$, and whose driving noises are independent planar Brownian motions obtained by Girsanov transformation. The no-triple-contacts property holds pathwise: the probability that three distinct particles occupy the same point at some time $t<T_\\partial$ is zero. The proof builds two classes of strong Markov processes with lifetime—one defined by a weighted average of the laws of the stochastic one-δ motions before any contact, the other by local Girsanov transformations that close at the first contact time of a new pair—and concatenates them at contact-creation times.","pith_inferences":["If the weighted-average measures of Definition 3.10 can be shown to be true martingales rather than merely supermartingales up to the first contact, the phrase 'conditioned to attain contacts' would correspond to literal conditioning; the paper only establishes supermartingale and local martingale status, so this is an open strengthening.","The abstract no-simultaneous-contacts theorem may transfer to other singular interacting particle systems whose radial parts have Bessel-type drifts; checking its μ and correlation conditions is the entry point.","The paper leaves open whether the terminal time $T_\\partial$ is almost surely infinite; if it can be finite with positive probability in inhomogeneous couplings, the absorbing state $\\partial$ would be genuinely reached, and one would need to understand the boundary behaviour of the process there.","One practical consequence of pairwise-only contacts is that a simulation could alternate between free planar Brownian motion and two-body delta-contact reweighting, avoiding the simultaneous singularity of the N-body drift; this is not stated in the paper."],"forward_implications":["For any N≥3 and any eligible starting configuration, the SDE (3.9)–(3.10) admits a strong Markov solution up to the terminal time, giving a diffusion-process representation of multiple two-body δ interactions in two dimensions.","At each contact-creation time exactly one new pair begins to interact (no-triple-contacts), so the same two-step construction can be restarted from the new configuration, making the concatenation scheme well-defined.","When all active couplings coincide (w-homogeneous β), every contact-creation time is almost surely finite, so the inductive concatenation never stalls.","For {0,1}-valued weights, the stochastic many-δ motion is a probabilistic counterpart of the formal Hamiltonian with δ-potentials on the selected pairs, and it reduces to stochastic one-δ motions by change of measure.","The log-Radon–Nikodym semimartingale decompositions (Proposition 3.7) are the intended main tool for the Feynman–Kac-type formulas in the next paper [8]."],"supporting_citations":[{"why":"Supplies the stochastic one-δ motions under $P^{\\beta_i\\downarrow,i}$ with their SDEs, transition densities, integrability bounds, and local-time growth, on which every Girsanov transformation and local-time formula in this paper is built.","marker":"[7]"},{"why":"Provides the BES(0,β↓) construction and the exponential change-of-measure idea that the weighted-average definition (Definition 3.10) generalizes, along with the normalization of the Markovian local time used in the paper.","marker":"[10]"},{"why":"The comparison theorem for collisions of Brownian particles is extended in Theorem 3.4 to many semimartingales and used to prove no-simultaneous-contacts and no-triple-contacts.","marker":"[14]"},{"why":"Develops the two-body case and ground-state transformations with $K_0$ and $K_1$, whose derivative formulas and local-time approximations are adapted for the log-sum Itô formulas in Section 5.","marker":"[6]"},{"why":"Earlier functional-integral solutions exhibit the no-triple-contacts phenomenon at the expectation level, which this paper converts into a pathwise statement.","marker":"[5]"},{"why":"Standard reference for Girsanov's theorem, the Dambis–Dubins–Schwarz theorem, stochastic calculus tools, and the exponential-martingale criteria used throughout the construction.","marker":"[23]"},{"why":"General concatenation theorem for strong Markov processes used to assemble the two classes of processes with lifetime into the final process on $C^N \\cup \\{\\partial\\}$.","marker":"[24]"},{"why":"Provides the smooth regularization technique used in the comparison argument in Section 4 to turn the difference inequality into a Grönwall inequality.","marker":"[28]"}],"fun_headline_variants":["No three particles meet: new stochastic many-δ motions for all N","Pairwise contacts only: pathwise proof for N-body delta gas","For every N, contacts stay pairwise in this 2D delta-gas model","No triple contacts: strong Markov many-δ motions constructed","Stochastic many-δ motions: no triples, all N≥3"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The construction imports the stochastic one-δ motions and their fine analytic properties—SDEs, transition densities, exponential moment bounds, and local-time growth—from the unpublished companion preprint [7], and every Girsanov transformation and local-time formula in this paper is applied to those processes, so an error in [7] would propagate into the main theorem.","fun_headline_variants_meta":{"raw":{"variants":["No three particles meet: new stochastic many-δ motions for all N","Pairwise contacts only: pathwise proof for N-body delta gas","For every N, contacts stay pairwise in this 2D delta-gas model","No triple contacts: strong Markov many-δ motions constructed","Stochastic many-δ motions: no triples, all N≥3"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.001142,"raw_usage":{"total_tokens":4763,"prompt_tokens":992,"completion_tokens":3771,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":608,"completion_tokens_details":{"reasoning_tokens":3677}},"tokens_in":608,"tokens_out":3771,"duration_ms":23258,"temperature":1.0,"reasoning_tokens":3677,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-16T04:13:01.397620+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Take N=3 with all three pair weights $w_j>0$ and initial positions with no coincidences. The theorem asserts that the first time any pair distance is zero and the first time a second pair distance is zero are almost surely distinct, i.e. $P(\\exists t<T_\\partial \\text{ and distinct } j_1,j_2,j_3 \\text{ with } Z^{j_1}_t=Z^{j_2}_t=Z^{j_3}_t)=0$. A concrete disproof would be to show, for some eligible initial configuration and weights, that this event has positive probability—for example, by constructing a positive-probability set of Brownian paths under (3.9) on which two different pair distances hit zero at the same time.","supporting_citations":[{"cited_title":"and Dol´ eans-Dade, C.A","cited_arxiv_id":null,"evidence_quote":"Supplies the stochastic one-δ motions under $P^{\\beta_i\\downarrow,i}$ with their SDEs, transition densities, integrability bounds, and local-time growth, on which every Girsanov transformation and local-time formula in this paper is built."},{"cited_title":", Høegh-Krohn, R","cited_arxiv_id":null,"evidence_quote":"Develops the two-body case and ground-state transformations with $K_0$ and $K_1$, whose derivative formulas and local-time approximations are adapted for the log-sum Itô formulas in Section 5."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Standard reference for Girsanov's theorem, the Dambis–Dubins–Schwarz theorem, stochastic calculus tools, and the exponential-martingale criteria used throughout the construction."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Provides the smooth regularization technique used in the comparison argument in Section 4 to turn the difference inequality into a Grönwall inequality."}],"review_version":1}