{"id":"e794a00f-0802-4f84-adde-2f549b0c4849","arxiv_id":"2504.19367","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":8.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":0,"one_line_summary":"For any hyperbolic triangle group, the reduced random walk converges almost surely to a boundary point; for PGL2(Z), its limit CDF is expressed explicitly via the new interrobang function.","lead":"The paper proves that Lam's reduced random walk in a hyperbolic triangle group almost surely converges to a point on the real boundary, and for the group PGL2(Z) it gives the exact probability formula for where that point lands using a newly introduced 'interrobang' function.","discovery_kind":"new_method","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Interrobang definition as printed gives ‽(1/2)=1/8, while CDF (23) requires ‽(1/2)=1/16; the printed recursion then fails monotonicity, so Theorem 8.1 is unsupported unless the denominator is corrected.","rationale":"The reader identified Lemma 4.1 as the weakest assumption and noted the Definition 7.4 rendering issue only in passing. My stress test finds a more direct obstruction to the central claim: the printed recursive definition of the interrobang function is inconsistent with Lemma 7.6, with Theorem 8.1's own verification, and with the boundary values required for (23) to be a continuous CDF. Exact rational computations show that the printed recursion fails strict monotonicity. This is load-bearing because Theorem 8.1 is the paper's headline result and it is proved by checking that the proposed f satisfies the stationary equation; if f is not a genuine CDF, that check cannot succeed. The issue appears likely to be a typographical omission of an exponent in the denominator, since the exponential form makes the boundary identities and the monotonicity arguments cohere. Because the intended mathematics is plausibly correct but the manuscript as written is internally inconsistent at the level of its central definition, the appropriate disposition remains conditional rather than accept or reject. I therefore keep the reader's CONDITIONAL verdict unchanged, while shifting the focus from Lemma 4.1 to the interrobang recursion inconsistency.","tokens_in":19667,"tokens_out":22411,"duration_ms":205952,"concrete_test":"Implement Definition 7.4 exactly as printed and compute the recursive values for 1/3 and 51/100; verify that ‽(51/100)=2019/51200<1/12, contradicting strict monotonicity. Then recompute the boundary agreement of formula (23) at x=−2, −1/2, and 2 using the three possible denominators 4⌊1/x⌋, 4^{⌊1/x⌋}, and 4⌊1/x⌋^2. The corrected version should make both sides agree at all three boundaries and should make the recursion monotone on [0,1]∩Q.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The central claim Theorem 8.1 depends entirely on the interrobang function being continuous and strictly increasing, and on the boundary values of the piecewise CDF being consistent. As displayed, Definition 7.4 (formula (15)) uses denominator 4⌊1/x⌋ in its second branch, which forces ‽(1/2)=1/8 and, via Lemma 7.6(i), gives ‽(x)=1/8(1−2‽(1/x−2)) on [1/3,1/2]. However, the verification in Theorem 8.1 uses 1/16 in exactly this region, and continuity of (23) at x=−2, x=−1/2, and x=2 forces ‽(1/2)=1/16, not 1/8. With the printed value 1/8, the recursion is not strictly increasing: exact computation gives ‽(1/3)=1/12 while ‽(51/100)=2019/51200<1/12, contradicting Proposition 7.8 and hence Theorem 7.13. This is an internal inconsistency in the central construction, not merely an omitted routine verification. The likely fix is that the denominator should be exponential, e.g. 4^{⌊1/x⌋} or 4⌊1/x⌋^2, which would restore the boundary identities and monotonicity; but as written the CDF formula cannot be correct.","agreement_with_reader":"partial"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper studies Lam's reduced random walk in hyperbolic triangle groups, realized geometrically as a random walk in the upper half-plane. The authors prove that the walk converges almost surely to a boundary point (Theorem 4.2), that the limiting distribution is the unique solution of a simple distributional equation (Theorems 4.4 and 5.1), and, in the special case of PGL_2(Z), that the cumulative distribution function of the limit is given by an explicit piecewise formula (Theorem 8.1) in terms of a newly introduced \"interrobang function.\" The paper also develops analytic and arithmetic properties of this function, including strict monotonicity, uniform continuity, rational-to-dyadic and quadratic-irrational-to-rational mapping properties, and several conjectures. The general convergence and uniqueness arguments are clean and well presented. However, the central construction of the interrobang function contains a concrete internal inconsistency that affects the validity of Theorem 8.1 as stated.","tokens_in":19952,"tokens_out":9906,"duration_ms":92909,"significance":"If the interrobang construction is corrected, the paper would make a substantial contribution by providing an explicit, parameter-free description of the limit distribution of a natural random walk on PGL_2(Z), in the spirit of Letac and Piccioni's work with Minkowski's question-mark function. The coupling proof of uniqueness in Theorem 5.1 is elegant, and the plan to encode the distribution through a function with arithmetic properties is original and likely of independent interest. The paper also honestly identifies several open conjectures, which is a strength. However, the current inconsistency in the definition of the interrobang function is load-bearing: it directly undermines Theorem 8.1, the paper's central explicit result.","major_comments":[{"comment":"There is an internal inconsistency in the definition of the interrobang function. As printed, the first branch of (15) gives ‽(1/2) = 1/(4·2) = 1/8, and Lemma 7.5 states ‽(1/n) = 1/(4n). However, the piecewise formula (23) in Theorem 8.1 is continuous at x = -2 only if ‽(1/2) = 1/16, and the worked verification for x in [-3,-2] uses Lemma 7.6 with coefficient 1/16 rather than the printed 1/8. Moreover, the printed recursion violates monotonicity: ‽(1/3) = 1/12, while an exact computation from (15) gives ‽(51/100) = 2019/51200 < 1/12, contradicting Proposition 7.8 and hence Theorem 7.13. The likely repair is to make the denominator in the first branch exponential, for example 4 raised to the power ⌊1/x⌋, which would make the boundary identities and the 1/16 coefficient in the proof of Theorem 8.1 consistent. As written, however, Theorem 8.1 is not supported by the preceding definitions, and every lemma in Section 7 that relies on the printed denominator needs to be rechecked after the correction.","section":"Definition 7.4, Lemma 7.5, Lemma 7.6, and Theorem 8.1"},{"comment":"Lemma 4.1, the assertion that the reduced random walk never crosses a given line more than once, is stated informally and then used essentially in the proof of Theorem 4.2 to conclude that any two subsequential limits lie in the closure of a single region in a finite line arrangement. No proof of Lemma 4.1 is supplied, despite the fact that it is the key geometric input for the existence of the boundary limit zeta. Since Theorem 4.2 is the basis for the distributional equation (5) and for everything that follows, a rigorous proof of Lemma 4.1 (or a precise reference) should be included.","section":"Lemma 4.1 and Theorem 4.2"}],"minor_comments":[{"comment":"The plot of the cumulative distribution function F_zeta would benefit from labeled axes and tick marks; currently the vertical scale is not readable.","section":"Figure 6"},{"comment":"The numerical values ‽^{-1}(1/8) = 0.61242994... and ‽^{-1}(1/4) = 0.61834758... are reported without a description of the algorithm used to compute them; a brief reproducibility note would be helpful.","section":"Conjectures 7.16 and 7.17"},{"comment":"The notation 4⌊1/x⌋ is ambiguous as typeset; the authors should clearly indicate whether it is a product or a superscript, and the notation should be consistent with the corrected definition used in Section 7.","section":"Equation (15)"}],"recommendation":"major_revision","confidential_remarks":"The central issue appears to be a fixable slip in the constants defining the interrobang function, but the current printed version is internally inconsistent and cannot be accepted as is. I recommend that the editors require the authors to correct the denominator in Definition 7.4, update all dependent lemmas, and supply the complete casework for the verification of the fixed-point equation in Theorem 8.1. The general results in Sections 4 and 5 are sound and worth publishing once the explicit PGL_2(Z) part is repaired."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Colleague — quick take on 2504.19367. The paper has a genuinely novel idea: a new 'interrobang' function that plays the role for the reduced PGL2(Z) random walk that Minkowski's question mark plays for the unreduced PSL2(Z) walk, and the general convergence/uniqueness framework is solid. But the central definition as printed is wrong. The stress-test note is accurate. Definition 7.4 gives ‽(1/2)=1/8 (first branch, denominator 4·2), while the CDF (23) forces ‽(1/2)=1/16 for continuity at x=-2. Worse, the printed recursion violates monotonicity: ‽(1/3)=1/12 but ‽(51/100)=2019/51200 < 1/12, contradicting Proposition 7.8 and hence Theorem 7.13. So the proof of Theorem 8.1 is unsupported as written.\n\nWhat is good: The general theorems — almost sure boundary convergence for reduced random walks in any hyperbolic triangle group (Theorem 4.2), the distributional equation (Theorem 4.4), and uniqueness of the stationary measure (Theorem 5.1) — appear clean and correct. The coupling proof of uniqueness via the Poincaré disk is elegant and uses the contraction lemmas properly. The arithmetical claims about the interrobang function — rationals to dyadic rationals, quadratic irrationals to rationals — are attractive, and the linear-algebra argument via the Levy–Desplanques theorem is a nice touch. The citation pattern is fine: Lam's reduced random walk, the oriented swap process, and the Minkowski question-mark literature are all there.\n\nThe soft spot is the definition. It is not just a missing routine verification; the printed recursion gives values that conflict with the CDF verification in Theorem 8.1. The likely fix is to replace the denominator 4⌊1/x⌋ with 4^{⌊1/x⌋} (or something equivalent), which restores the boundary values and monotonicity. That is fixable, but it has to be fixed before the main result can be trusted. Also, Lemma 4.1 (never crossing a line twice) is asserted informally; the proof of Theorem 4.2 depends on it, and a more explicit proof would help.\n\nBottom line: This paper deserves a serious referee, not a desk rejection. The general framework is likely correct, and the interrobang construction is interesting enough to justify the effort. But as printed, the central CDF theorem is not proven. I would send it to a referee with a clear request to check the definition and monotonicity of ‽. I would not cite it in its current form.","headline":"New interrobang construction for the reduced PGL2(Z) walk is clever and the general framework holds up, but the definition of ‽ as printed contradicts the CDF and fails monotonicity, leaving Theorem 8.1 unproven without a fix.","tokens_in":20475,"tokens_out":7146,"would_cite":false,"duration_ms":58059,"reading_group":"maybe","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["60B15","20F55","60J10"],"pacs":[],"model":"deepseek-v4-flash","headline":"The reduced random walk on a hyperbolic triangle group converges almost surely to a boundary point; in the (2,3,∞) case its distribution is given exactly by the interrobang function.","keywords":["reduced random walk","hyperbolic triangle groups","interrobang function","question-mark function","Demazure product","boundary limit distribution","Coxeter groups","PGL2(Z)"],"falsifier":"Produce one trajectory of the reduced random walk (one long sequence of uniform choices $i_1,i_2,\\ldots$) in a hyperbolic triangle group in which some arrangement line is crossed more than once; Lemma 4.1 asserts this never happens, and the convergence proof of Theorem 4.2 depends on it. For the explicit formula, compare long-run simulated probabilities $\\Pr[\\zeta\\le q]$ for rational $q$ in the $\\mathrm{PGL}_2(\\mathbb{Z})$ walk with the right-hand side of (23); a persistent discrepancy would refute Theorem 8.1.","tokens_in":19448,"feed_emoji":"🎲","tokens_out":17968,"duration_ms":164895,"temperature":0.7,"pith_summary":"The paper studies the reduced random walk on a hyperbolic triangle group, where at each step the walker chooses one of three neighboring alcoves uniformly but stays put if the move would recross a line it has already crossed. The central claim is that this walk almost surely converges to a point $\\zeta$ on the extended real line, and that the law of $\\zeta$ is the unique probability measure satisfying a self-similar equation built from the three one-way reflections. For the group $\\mathrm{PGL}_2(\\mathbb{Z})$ (the $(2,3,\\infty)$ triangle group), the paper writes the cumulative distribution function $F_\\zeta(x)=\\Pr[\\zeta\\le x]$ explicitly, in seven pieces, in terms of a new recursively defined function called the interrobang function. The interrobang function extends to a continuous, strictly increasing function on $[0,1]$, sends rational numbers to explicitly computable dyadic rationals and quadratic irrationals to explicitly computable rationals, and is conjectured to be differentiable almost everywhere with derivative zero. A reader should care because this yields an exact, computable description of the boundary limit of a natural non-reversible walk and introduces a new function in the same family as the classical question-mark function.","feed_headline":"Never-crossing random walk converges to the hyperbolic boundary","feed_subtitle":"In the (2,3,∞) triangle group the limit's distribution is written exactly using the new interrobang function.","key_machinery":"The geometric mechanism is the one-way reflection $\\tau_i$: it reflects across the side $L_i$ when the current point lies on the starting side of $L_i$, and otherwise does nothing. Composing these maps realizes the reduced random walk, because the underlying Demazure product $\\star$ on the Coxeter group (defined by $s_i\\star w=w$ when $i$ is a left descent of $w$, and $s_i w$ otherwise) records exactly which alcoves can be reached without crossing any line twice. The probability machinery is the distributional equation $\\mu=\\frac{1}{3}((\\tau_1)_*\\mu+(\\tau_2)_*\\mu+(\\tau_3)_*\\mu)$; uniqueness is proved by coupling two solutions on the boundary circle and showing their expected distance decays geometrically via a contraction estimate for the $\\tau_i$. The explicit answer for $\\mathrm{PGL}_2(\\mathbb{Z})$ is carried by the interrobang function, a recursive function on rationals that extends to a continuous strictly increasing function on $[0,1]$; its arithmetic properties make the evaluations of $F_\\zeta$ at rational and quadratic-irrational inputs exactly computable.","core_discovery":"On the paper's own terms: with simple reflections $s_1,s_2,s_3$ acting on the upper half-plane, define the sequence $z_m$ by composing one-way reflections $\\tau_i$ applied to a starting point. Theorem 4.2 states that $z_m$ converges almost surely to a point $\\zeta\\in\\overline{\\mathbb{R}}$ on the extended real line, and Theorem 5.1 states that its distribution $\\mu$ is the unique probability measure satisfying $\\mu=\\frac{1}{3}((\\tau_1)_*\\mu+(\\tau_2)_*\\mu+(\\tau_3)_*\\mu)$. For the triangle group generated by $s_1(z)=-1-z$, $s_2(z)=1/z$, $s_3(z)=-z$, which is isomorphic to $\\mathrm{PGL}_2(\\mathbb{Z})$, Theorem 8.1 gives the cumulative distribution function $F_\\zeta$ by the explicit piecewise formula (23) in terms of the interrobang function; in particular, $F_\\zeta$ is continuous and strictly increasing, it sends rationals to explicitly computable dyadic rationals, and it sends quadratic irrationals to explicitly computable rationals.","pith_inferences":["Editorial inference: the coupling-and-contraction method used for uniqueness should transfer to other hyperbolic triangle groups such as $(3,3,\\infty)$ or $(2,4,\\infty)$, giving explicit stationary laws once the corresponding recursive functions are worked out.","Editorial inference: if Conjecture 7.14 holds, the interrobang function (and with it $F_\\zeta$) is a strictly increasing singular function, so the boundary limit would be supported on a Lebesgue-null set despite having no atoms.","Editorial inference: a concrete test of Conjectures 7.16 and 7.17 would be to compute the continued-fraction expansion of the inverse interrobang value at $1/8$; an eventually periodic expansion would disprove transcendence, while a certificate from a transcendence criterion could prove it."],"forward_implications":["Every hyperbolic triangle group carries a canonical probability measure on the extended real line: the limit law of the reduced random walk, uniquely determined by the one-way reflections.","In $\\mathrm{PGL}_2(\\mathbb{Z})$, the cumulative distribution function of the limit is continuous and strictly increasing, so the limit distribution has no atoms and every individual boundary point has probability zero.","Because $F_\\zeta$ sends rationals to explicitly computable dyadic rationals and quadratic irrationals to explicitly computable rationals, probabilities such as $\\Pr[\\zeta\\le q]$ can be written down exactly rather than approximated by simulation.","The convergence is sure, not merely almost sure, since the walk is almost surely not eventually constant and the compactness argument produces a boundary limit for each such trajectory."],"supporting_citations":[{"why":"introduces the reduced random walk on Coxeter groups and its affine boundary behavior, the object this paper reinterprets in the hyperbolic plane.","marker":"[22]"},{"why":"studies a random walk on PSL2(Z) whose limit distribution is the question-mark function, the direct template for the interrobang result.","marker":"[24]"},{"why":"supplies the hyperbolic tiling, triangle-group presentation, and fundamental-domain facts underlying the geometric setup.","marker":"[25]"},{"why":"ensures the one-way reflection maps are well defined independently of the reduced word used to compose them.","marker":"[28]"},{"why":"provides the classical arithmetic and analytic properties of the question-mark function that the interrobang function imitates and that Theorem 7.21 relies on.","marker":"[29]"},{"why":"gives the diagonal-dominance invertibility criterion used to prove that the interrobang function sends quadratic irrationals to rationals.","marker":"[20]"}],"fun_headline_variants":["Hyperbolic random walk's limit law is the interrobang function","Exact CDF for hyperbolic walk limit via new interrobang","Reduced walk on hyperbolic plane converges to boundary a.s.","PGL2(Z) walk limit has explicit distribution: interrobang"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The load-bearing geometric premise is Lemma 4.1: the walk never crosses a given line of the hyperbolic arrangement more than once; if a line could be crossed twice, the walk could oscillate between its two sides and the boundary limit $\\zeta$ would not be forced to exist.","fun_headline_variants_meta":{"raw":{"variants":["Hyperbolic random walk's limit law is the interrobang function","Exact CDF for hyperbolic walk limit via new interrobang","Reduced walk on hyperbolic plane converges to boundary a.s.","PGL2(Z) walk limit has explicit distribution: interrobang"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000349,"raw_usage":{"total_tokens":1901,"prompt_tokens":931,"completion_tokens":970,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":547,"completion_tokens_details":{"reasoning_tokens":896}},"tokens_in":547,"tokens_out":970,"duration_ms":10077,"temperature":1.0,"reasoning_tokens":896,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-16T05:56:26.174360+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Produce one trajectory of the reduced random walk (one long sequence of uniform choices $i_1,i_2,\\ldots$) in a hyperbolic triangle group in which some arrangement line is crossed more than once; Lemma 4.1 asserts this never happens, and the convergence proof of Theorem 4.2 depends on it. For the explicit formula, compare long-run simulated probabilities $\\Pr[\\zeta\\le q]$ for rational $q$ in the $\\mathrm{PGL}_2(\\mathbb{Z})$ walk with the right-hand side of (23); a persistent discrepancy would refute Theorem 8.1.","supporting_citations":[{"cited_title":"The shape of a random affine Weyl group element and random core partitions","cited_arxiv_id":null,"evidence_quote":"introduces the reduced random walk on Coxeter groups and its affine boundary behavior, the object this paper reinterprets in the hyperbolic plane."},{"cited_title":"Random walks in the hyperbolic plane and the Minkowski question mark function","cited_arxiv_id":null,"evidence_quote":"studies a random walk on PSL2(Z) whose limit distribution is the question-mark function, the direct template for the interrobang result."},{"cited_title":"Noneuclidean tesselations and their groups , volume Vol","cited_arxiv_id":null,"evidence_quote":"supplies the hyperbolic tiling, triangle-group presentation, and fundamental-domain facts underlying the geometric setup."},{"cited_title":"G´ en´ erateurs et relations des groupes de Weyl g´ en´ eralis´ es.C","cited_arxiv_id":null,"evidence_quote":"ensures the one-way reflection maps are well defined independently of the reduced word used to compose them."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"provides the classical arithmetic and analytic properties of the question-mark function that the interrobang function imitates and that Theorem 7.21 relies on."},{"cited_title":"Horn and Charles R","cited_arxiv_id":null,"evidence_quote":"gives the diagonal-dominance invertibility criterion used to prove that the interrobang function sends quadratic irrationals to rationals."}],"review_version":1}