{"id":"61e806d0-d144-4f3f-b285-889e60ee2111","arxiv_id":"2505.07682","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":8.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":0,"one_line_summary":"Under new 'rough radial structure' assumptions, ball averages on exponentially growing groups satisfy weak-type L(logL)^c maximal inequalities, and non-elementary hyperbolic groups satisfy the optimal weak-type (1,1) inequality.","lead":"The paper proves a sharp weak-type inequality for the Hardy-Littlewood maximal operator on ball averages in many groups with exponential growth. It shows that every non-elementary word-hyperbolic group satisfies the optimal weak-type (1,1) maximal inequality.","discovery_kind":"new_method","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Half-integer m case omitted in Prop. 20 leaves the complementary parity class i-j not congruent to r modulo 2 unbounded; Theorem 21's weak-type (1,1) is not yet established.","rationale":"Reader's CONDITIONAL verdict is appropriate. The paper's central claim for hyperbolic groups depends on converting an integer-parity coarse median estimate into a full correlation estimate. The text explicitly defers the half-integer case, and the omitted case corresponds to a parity class of pairs that the decomposition in Proposition 19 must sum over. This is a locatable, load-bearing gap; however, it is plausibly repairable, since the same delta-hyperbolic geometry should give the half-integer bound by choosing a vertex at distance floor(m) and absorbing the error into constants. I also checked the surrounding machinery: Coornaert's growth estimate (57) is standard, Theorem 10's Orlicz calculation is self-consistent, and Theorem 15's use of Harish-Chandra estimates is in line with known results. The concern is therefore not a refutation of the method but a precise missing argument at the endpoint c=0. The proposed test, completing the half-integer estimate and rerunning the parity-class summation, would settle it; until then, CONDITIONAL is the right status.","tokens_in":34610,"tokens_out":21474,"duration_ms":225362,"concrete_test":"Supply the omitted case: for m in Z+1/2, choose z on [1,x] with d(1,z)=floor(m) and repeat the delta-thin triangle argument, tracking all additive constants, and prove |{(x,y) in E_j x F_i : d(x,y)=r}| is at most C min{q^{r-m}|E_j|, q^m|F_i|} with C independent of r. Then rerun the summation in Proposition 19 allowing m in (1/2)Z and confirm the complementary parity class contributes only a constant factor to the bound in (56). If a factor q^{r/2} or r appears, recompute c in Theorem 18 and Theorem 21.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The load-bearing step for Theorem 21 is the proof of Proposition 20, where for i,j,r the parameter m=(j+r-i)/2 is fixed and the text states: 'We need a minor modification in the case that m is an half integer, which we omit.' This omission is not cosmetic. The spherical coarse median inequality (55) and the summation in Proposition 19 use only integer m: the lines i=j+r-2m cover exactly the pairs with i-j congruent to r modulo 2. Pairs in the complementary parity class require half-integer m, and for these no uniform bound of the form (58) is proved. Since Proposition 19 uses (55) to derive rapid decay of spherical correlations with b=0, and Theorem 18 then converts b=0 into weak type (1,1), the missing half-integer estimate is exactly what carries the endpoint result. If the half-integer class only satisfies (58) with a factor q^{r/2} or r, the exponent in Theorem 21 would be positive rather than 0. The gap is explicit in the manuscript, and the proof framework is otherwise coherent; this is a repair that should be supplied before the endpoint theorem is accepted.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper develops a general framework for weak-type maximal inequalities for ball averages on discrete groups with exponential volume growth. It introduces three rough radial structure conditions: almost exact polynomial-exponential growth of spherical shells, rapid decay of spherical shell correlations, and a spherical coarse median inequality. The main conditional theorem (Theorem 10) shows that rapid decay of spherical shell correlations with polynomial parameter b, together with almost exact growth, implies a weak-type L(log L)^{2b} maximal inequality for ball averages; Theorem 18 gives a variant under the spherical coarse median inequality. Applications are given to lattices in connected semisimple Lie groups, RAAGs, Coxeter groups, braid groups, and products of hyperbolic groups. The headline result is Theorem 21: for every non-elementary hyperbolic group and every finite symmetric generating set, the Hardy-Littlewood operator for word-metric balls satisfies the weak-type (1,1) maximal inequality. The proof of the hyperbolic case passes through Proposition 20, which asserts the spherical coarse median inequality of rank 1 for every word metric on a non-elementary hyperbolic group.","tokens_in":34779,"tokens_out":7179,"duration_ms":73555,"significance":"If Theorem 21 is established with a complete proof, it is a significant advance: it extends the known weak-type (1,1) results for free groups and trees to all non-elementary hyperbolic groups, which is the optimal result in this setting. The conditional engine provided by Theorem 10 is elegant and is supported by a wide range of examples, and the paper is careful to delineate open problems and to credit earlier methods, especially those of Naor-Tao and the authors' earlier transfer and counting framework. The conditional theorems and the non-hyperbolic applications appear coherent. However, the explicit gap in Proposition 20 leaves the endpoint statement for hyperbolic groups unproved as it stands, and this is the central advertised achievement. The manuscript therefore needs a substantial repair before the main theorem can be accepted.","major_comments":[{"comment":"The proof fixes m=(j+r-i)/2 and states that the half-integer case is a minor modification that is omitted. This omission is load-bearing. In Proposition 19 the decomposition into the lines i=j+r-2m only covers pairs with i-j congruent to r modulo 2; pairs in the complementary parity class require half-integer m. A uniform bound of the form (58) for those pairs, with constants independent of r, is exactly what is needed to obtain rapid decay of spherical correlations with parameter b=0 in Proposition 19, and Theorem 21 then follows via Theorem 18 only in that case. The geometric construction in the proof chooses points z,w,v at distance m from e or x or y; if m is half an integer, such points do not exist in the 1-skeleton, so a rounding argument is needed and its effect on the constants must be quantified. If the half-integer class only satisfies (58) with an additional factor q^{r/2} or a power of r, the exponent in the hyperbolic theorem would become positive and the weak-type (1,1) conclusion would not follow. This gap must be repaired before Theorem 21 can be accepted.","section":"Section 7, Proposition 20 (proof leading to inequality (58))"},{"comment":"The spherical coarse median inequality (55) is stated with quantities S_{r-m} and S_m, which are only meaningful for integer radii. The definition should say explicitly that m ranges over integers, and then the half-integer case must be handled separately in Proposition 19 or by an extension of (55). As written, the statement of Proposition 19 presupposes the integer case and leaves the complementary parity class uncovered. This is not merely a notational issue: the proof of Proposition 20 explicitly acknowledges the missing case, and the endpoint theorem depends on it.","section":"Section 6, Definition 17 and Proposition 19"}],"minor_comments":[{"comment":"The phrase 'for alx 0∈X' should read 'for all x_0∈X'.","section":"Section 1.1, first paragraph"},{"comment":"The statement says 'Let E_j⊂S_j, F_j⊂S_j' but the proof uses F_i⊂S_i; the notation for the two sets should be made consistent.","section":"Section 7, Proposition 20 statement"},{"comment":"Inequality (45) is a key transfer estimate for the lattice subgroup applications, but it is only cited via a discussion following Lemma 7.3 of [BS93]. Please state the inequality and its hypotheses explicitly, or give a precise page-level reference, so that the reader can verify the normalization and the applicability to the spherical shells used here.","section":"Section 5.1, inequality (45)"}],"recommendation":"major_revision","confidential_remarks":"The paper has a strong and interesting conditional framework, and the non-hyperbolic applications appear to be in good shape modulo standard citations. The reason for major revision rather than acceptance is the explicit omission in Proposition 20: the half-integer case is exactly the parity class needed for the claimed weak-type (1,1) theorem for hyperbolic groups. This is a repair that could plausibly be supplied, but it is not a cosmetic detail."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Dear colleague,\n\nQuick take: this is a genuinely strong paper, but the starring result—weak-type (1,1) for every word metric on a nonexlementary hyperbolic group—is not fully proved as written. The gap is explicit and locatable, and it needs to be fixed before the theorem is accepted.\n\nThe real news is the general machine. The rough-radial-structure assumptions (almost exact polynomial-exponential growth, rapid decay of spherical shell correlations, and the spherical coarse median inequality) give a clean route to weak-type L(logL)^c bounds for ball averages. The reduction theorems (10 and 18) are the core, and they look correct. The applications are broad: lattices in semisimple groups, RAAGs, Coxeter groups, braid groups, and products of hyperbolic groups. The transfer argument for lattices (Theorem 14) is well constructed and carefully assembled from existing results. The paper is also honest about what it does not prove, including the optimality questions it leaves open.\n\nNow the soft spot. In Proposition 20, the proof of the spherical coarse median inequality for hyperbolic groups fixes m=(j+r−i)/2 and says, roughly, \"if m is a half-integer, pretend it is an integer; the modification is minor, omitted.\" This is not cosmetic. Proposition 19 uses the inequality to sum over the lines i=j+r−2m; those lines only cover pairs with i+j−r even. Word metrics on non-bipartite Cayley graphs (e.g., a nonexlementary hyperbolic group with torsion, or even Z with generators {±2,±3}) admit pairs with i+j−r odd, so the complementary parity class is real. As written, those pairs are not bounded by the spherical coarse median inequality, so the rapid-decay-of-correlations step, and hence Theorem 21, is incomplete. The same issue appears in Lemma 34 for products, where the proof also passes over half-integer m without comment.\n\nThat said, I doubt the gap is fatal. The natural fix—using midpoints of edges and rounding m, or applying the argument to the nearest integer shell—should produce the same bound with only a constant factor (something like q^{1/2}), which would preserve the endpoint. But \"should\" is not a proof, and the authors have explicitly left this to the reader. The referee should insist on seeing it.\n\nThe rest of the claims hold up well on inspection. The conditional theorems are not circular; the new assumptions are stated independently and verified via existing external results. No invented parameters, no data issues.\n\nWho is this for? Anyone working on maximal inequalities in geometric group theory or harmonic analysis on groups. It deserves a serious referee, not a desk reject. Recommendation: send to peer review, with the requirement that the half-integer case in Prop. 20 (and by extension Lemma 34) be supplied. If that pans out, this is a significant paper.\n\nTalk soon,\n[Your name]","headline":"The framework and most applications are solid, but the hyperbolic endpoint relies on an omitted half-integer parity case that must be supplied before the paper's headline result is fully proven.","tokens_in":35353,"tokens_out":12957,"would_cite":true,"duration_ms":120710,"reading_group":"yes","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["43A05","20F65","43A80","22E40","22F30"],"pacs":[],"model":"deepseek-v4-flash","headline":"Every non-elementary hyperbolic group has a weak-type (1,1) Hardy–Littlewood maximal inequality for every word metric.","keywords":["Hardy-Littlewood maximal operator","weak type (1,1)","exponential volume growth","hyperbolic groups","spherical coarse median inequality","rapid decay of spherical shell correlations","ball averages"],"falsifier":"Verify the counting bound in Proposition 20, $|\\{(x,y)\\in E_j\\times F_i: d(x,y)=r\\}|\\le C\\min\\{q^{r-m}|E_j|,\\,q^m|F_i|\\}$, for all half-integer values of $m=(j+r-i)/2$; a single Cayley graph of a hyperbolic group where this inequality fails for every constant $C$ would invalidate the proof of Theorem 21. A direct numerical test would compute the distribution of $M(\\delta_e)$ on a large ball in such a Cayley graph and compare $|\\{M\\delta_e\\ge \\eta\\}|$ with $C/\\eta$; a violation of the uniform constant would disprove the weak-type $(1,1)$ claim.","tokens_in":34358,"feed_emoji":"📐","tokens_out":9720,"duration_ms":82669,"temperature":0.7,"pith_summary":"The paper establishes sufficient conditions under which the Hardy–Littlewood maximal operator for ball averages on a discrete group with exponential volume growth satisfies a weak-type maximal inequality. Its main theorem (Theorem 10) shows that almost-exact polynomial-exponential growth of spherical shells together with rapid decay of spherical shell correlations with parameter $b$ forces a weak-type $L(\\log L)^{2b}$ bound for the ball averages. The headline application (Theorem 21) is that for every non-elementary hyperbolic group and every symmetric finite generating set, the word-metric ball averages satisfy the optimal weak-type $(1,1)$ inequality on $\\ell^1(\\Gamma)$. The same machinery produces $L(\\log L)^c$ maximal inequalities for lattices in semisimple Lie groups, right-angled Artin groups, Coxeter groups, and braid groups, and it frames the open question of whether the optimal exponent $c=0$ holds in all these cases.","feed_headline":"Weak-type (1,1) proved for all hyperbolic group word metrics","feed_subtitle":"For any symmetric generating set, the ball-averaging maximal operator on a non-elementary hyperbolic group reaches the optimal endpoint.","key_machinery":"The load-bearing object is the spherical shell average $\\sigma_r$, the uniform average on the annulus $\\{r\\le G(\\gamma)<r+L\\}$, together with the rapid-decay-of-shell-correlations estimate $|\\{(u,v)\\in A\\times B: r\\le d_G(u,v)<r+L\\}|\\le C r^b\\sqrt{|A|\\,|B|\\,|SS_r|}$. Proposition 8 proves a distributional inequality for $\\sigma_r$: the measure of the set where $\\sigma_r f$ exceeds $\\eta$ is controlled by a weighted sum of level sets of $f$, with weights involving $\\sqrt{2^n/|SS_r|}$. Summing these bounds over $r$ and using the two-sided growth law $|SS_r|\\asymp r^d q^r$ yields Theorem 10. For word metrics, the spherical coarse median inequality—a counting bound for pairs on spheres at a fixed distance—is the tool that verifies the correlation decay; for hyperbolic groups, the $\\delta$-thin triangle property supplies it with parameter $b=0$.","core_discovery":"On the paper's own terms, the central claim is that the weak-type $(1,1)$ maximal inequality for ball averages, long known for the free group with free generators and for symmetric spaces, holds for every non-elementary hyperbolic group with every word metric. The route is a general transfer: Theorem 10 converts rapid decay of spherical shell correlations plus almost-exact polynomial-exponential growth into a weak-type $L(\\log L)^{2b}$ maximal inequality for balls; Theorem 18 shows that a spherical coarse median inequality yields the required correlation decay with $b=d_2+\\frac12 d$; and Proposition 20 verifies that inequality with $d_2=0$ for every word metric on a hyperbolic group, while the cited growth estimate [Co93] gives $d=0$. The conclusion is $L(\\log L)^0=L^1$, i.e., weak type $(1,1)$.","pith_inferences":["If the omitted half-integer case in Proposition 20 can be handled uniformly, the same weak-type $(1,1)$ argument would apply to arbitrary hyperbolic left-invariant integer-valued metrics, provided their spherical growth is almost-exact exponential.","The exponent $2b$ in Theorem 10 may be far from sharp; a testable conjecture, raised as an open problem in the paper, is that all groups treated here satisfy the optimal $c=0$ endpoint.","The spherical coarse median inequality could serve as a sufficient condition for other classes with rational growth and rapid decay, for instance graph products or relatively hyperbolic groups with suitable generating sets, where both ingredients are already known.","A concrete numerical check on a small RAAG or hyperbolic Cayley graph, using finitely supported $f$ and comparing the distribution of $Mf$ with $C/\\eta$, could indicate whether the maximal inequality is sharp or whether $c>0$ is genuinely needed."],"forward_implications":["Every non-elementary hyperbolic group, with any symmetric finite generating set, has a Hardy–Littlewood maximal operator of weak type $(1,1)$ on $\\ell^1(\\Gamma)$, so the operator is bounded from $\\ell^1$ to weak $\\ell^1$—the best possible endpoint.","For any lattice in a connected semisimple Lie group with finite center, the ball averages defined by the Riemannian distance restricted to the lattice satisfy a weak-type $L(\\log L)^c$ inequality, with $c$ an explicit function of the root system.","Right-angled Artin groups, exponential-growth Coxeter groups, braid groups, and extra-large type Artin groups satisfy $L(\\log L)^c$ maximal inequalities for their standard or stated generating sets.","An $\\ell^1$-product of two non-elementary hyperbolic groups satisfies a weak-type $L(\\log L)^3$ inequality when the growth parameters coincide and $L(\\log L)^4$ when they differ.","The Hardy–Littlewood problem for ball averages on these groups is reduced to two checkable conditions: almost-exact polynomial-exponential growth of spherical shells and rapid decay of spherical shell correlations."],"supporting_citations":[{"why":"Supplies the radial shell-averaging distributional inequality and the coarse-geometric counting method that Proposition 8 and Proposition 19 generalize to arbitrary groups.","marker":"[NT09]"},{"why":"Provides the two-sided exponential growth estimate for spheres in word-hyperbolic groups that gives almost-exact growth with d=0 in Theorem 21.","marker":"[Co93]"},{"why":"Supplies the polynomial bound on median points in CAT(0) cube complexes used in Lemma 24 and establishes rapid decay for Coxeter groups, underpinning Theorem 27 and part of Theorem 28.","marker":"[CR05]"},{"why":"Characterizes rapid decay on connected Lie groups and provides the operator-norm bound for bi-K-invariant averages used in Theorem 15 to prove rapid decay of Riemannian spherical shells.","marker":"[CPSC]"},{"why":"Gives the effective lattice point counting in Riemannian balls used in Theorem 15 to transfer almost-exact growth from the symmetric space to the lattice.","marker":"[DRS93]"},{"why":"Shows the growth series of a right-angled Artin group with standard generators is rational, which supplies almost-exact growth in Theorem 27.","marker":"[LMW]"},{"why":"Establishes rationality of Coxeter group growth series, supplying almost-exact growth in Theorem 28.","marker":"[S]"},{"why":"Establishes rapid decay for Artin groups of extra-large type, used in Theorem 28.","marker":"[CHR2]"}],"fun_headline_variants":["Hyperbolic groups: weak (1,1) for every word metric","Ball averages on hyperbolic groups reach optimal weak type","Maximal inequality: weak (1,1) for all hyperbolic word lengths","All hyperbolic word metrics get weak-type (1,1) bound"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The load-bearing premise in the hyperbolic-group proof is the spherical coarse median inequality of rank $1$ for every word metric; the proof sets $m=(j+r-i)/2$ and omits the half-integer case, saying only that a minor modification is needed, so the entire Theorem 21 depends on that case being handled with constants independent of $r$.","fun_headline_variants_meta":{"raw":{"variants":["Hyperbolic groups: weak (1,1) for every word metric","Ball averages on hyperbolic groups reach optimal weak type","Maximal inequality: weak (1,1) for all hyperbolic word lengths","All hyperbolic word metrics get weak-type (1,1) bound"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000305,"raw_usage":{"total_tokens":1774,"prompt_tokens":996,"completion_tokens":778,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":612,"completion_tokens_details":{"reasoning_tokens":704}},"tokens_in":612,"tokens_out":778,"duration_ms":7184,"temperature":1.0,"reasoning_tokens":704,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-15T22:10:46.567128+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Verify the counting bound in Proposition 20, $|\\{(x,y)\\in E_j\\times F_i: d(x,y)=r\\}|\\le C\\min\\{q^{r-m}|E_j|,\\,q^m|F_i|\\}$, for all half-integer values of $m=(j+r-i)/2$; a single Cayley graph of a hyperbolic group where this inequality fails for every constant $C$ would invalidate the proof of Theorem 21. A direct numerical test would compute the distribution of $M(\\delta_e)$ on a large ball in such a Cayley graph and compare $|\\{M\\delta_e\\ge \\eta\\}|$ with $C/\\eta$; a violation of the uniform constant would disprove the weak-type $(1,1)$ claim.","supporting_citations":[],"review_version":1}