{"id":"07383d63-7fff-465d-bb50-78fb74daea91","arxiv_id":"2505.22625","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":7.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":0,"one_line_summary":"The biquadratic Guo-Jacquet and linear arithmetic fundamental lemmas for GL(4) with the unit Hecke function are proved by reduction to the coquadratic GL(2) case.","lead":"A full proof is given of two local conjectures, the biquadratic Guo-Jacquet fundamental lemma and its arithmetic version, for GL(4) over local fields with the unit test function. The proof works by reducing the biquadratic GL(4) statement to the already known coquadratic GL(2) case, plus explicit lattice counting.","discovery_kind":"first_principles","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Section 5.3 rests on unproved parity and valuation claims for z_delta; the rigid-case orbital integral is misstated (Orb(1,alpha,s) = -q^s vs. Theorem 4.11), leaving the reduction to coquadratic GL2 insecure.","rationale":"The reader's weakest_assumption identifies Section 5.3's parity assertion for z_delta and the derived consequences as the critical juncture, and I agree. The parity claim itself is likely true by standard residue-field arguments, so the more pressing problem is that the case analysis is not actually carried out: the rigid-case orbital integral is quoted as -q^s, which conflicts with Theorem 4.11, and the implication from high valuation of z_delta to 'w is a uniformizer and 1 - z_delta is a unit' is asserted without proof. These are not stylistic quibbles; they are the bridge from the biquadratic GL4 statement to the known coquadratic GL2 AFL. The biquadratic FL half has a separate numerical slip in Theorem 4.10 ((4.4) says Orb(1,beta,0) = 2 for ramified L, while the displayed formula and Theorem 4.15 give 1), but that appears to be a localized typo rather than a structural gap. Because the Section 5.3 gap is localized to unstated valuation arguments and a false-looking equality whose derivative may still be correct, I do not recommend rejecting the paper; I recommend conditioning acceptance on a complete derivation of the case split and a corrected orbital-integral statement.","tokens_in":24158,"tokens_out":30030,"duration_ms":343352,"concrete_test":"Recompute the Section 5.3 case analysis in full. (1) Prove that v_D(z_delta) is odd by passing to O_D/(varpi_D) and using the induced action on the residue field of O_{K1}. (2) In the case v_D(z_delta) = 1, evaluate Orb(1, alpha, s) using Theorem 4.11 with the actual v_L(z_delta^2) forced by z_delta^2 = (w - varpi_3)(w - varpi_3^sigma), and verify that -(1/ln q) d/ds at s = 0 equals 1; check whether the displayed equality Orb = -q^s holds or only holds up to an omitted constant. (3) For v_D(z_delta) >= 3, derive the valuation of z_delta^2 in L from the same equation and prove that this forces v_L(w) = 1 and 1 - z_delta in O_D^times; if not, produce a concrete pair (K1, K2) -> O_D satisfying the hypotheses of Theorem 5.10 but violating one of these conclusions.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The arithmetic half of Theorem 1.4 is proved only in Section 5.3 (Theorem 5.10), and that proof splits on the valuation of z_delta. Three claims are load-bearing and underived. First, 'we must have z_delta in varpi^{2Z+1} O_D^times' is asserted without proof; a residue-field Frobenius argument would justify it, but the paper does not give it. Second, in the case v_D(z_delta) = 1, the proof asserts Int(delta) = 1 and then 'if z_delta^2 in varpi_L^2 O_L^times, then Orb(1, alpha, s) = -q^s'. But Theorem 4.11, with v_L(w) = 1, gives Orb(1, alpha, s) = sum_{i=0}^{v_L(z^2)} (-q^s)^i. For v_L(z^2) = 2 this is 1 - q^s + q^{2s}, not -q^s; for v_L(z^2) = 1 it is 1 - q^s. The derivative at s = 0 is unchanged, so the conclusion Int(delta) = 1 may survive, but the displayed identity is false as written and the valuation normalization in the hypothesis is not reconciled with v_D(z_delta) = 1. Third, the remaining case asserts 'z_delta in varpi^2 O_D implies z_delta in varpi^3 O_D' and 'since z^2 = (w - varpi_3)(w - varpi_3^sigma), the element w must be a uniformizer of O_L, and thus 1 - z_delta in O_D^times'. This is exactly the hypothesis needed for the maximal-order reduction (Lemma 5.9) to the coquadratic GL2 AFL, but no derivation is supplied. If any regular semisimple pair has v_D(z_delta) = 1 with v_L(z_delta^2) = 2, or has v_D(z_delta) >= 3 without w a uniformizer, that orbit is covered by neither the rigid Int = 1 argument nor the reduction.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper claims to prove the biquadratic Guo–Jacquet Fundamental Lemma and the biquadratic linear Arithmetic Fundamental Lemma for GL(4) with the unit test function, over both p-adic fields and local fields of positive characteristic. The strategy is to use an explicit coproduct construction for pairs of quadratic embeddings, convert orbital integrals into lattice-counting problems, compute the analytic side for GL(4) by a case analysis on the valuation of w, and then reduce the arithmetic side by a maximal-order construction to the known coquadratic linear AFL for GL(2). The paper also states partial results for GL(2n) when the relevant order is integral.","tokens_in":24601,"tokens_out":14603,"duration_ms":153924,"significance":"If the main theorems were fully established, this would be a substantial advance in the program of arithmetic fundamental lemmas in the biquadratic setting, extending the Howard–Li results from GL(2) to GL(4) and providing the first higher-rank case with ramification. The combinatorial lattice computations in Section 4 are explicit and constitute a useful technical contribution, and the reduction strategy of Section 5.2, which relates biquadratic GL(4) to coquadratic GL(2), is conceptually appealing. However, the proof of the arithmetic AFL in Section 5.3 contains several unsupported valuation assertions and at least one displayed identity that is inconsistent with the paper's own orbital-integral formula, so the central claim is not established as written.","major_comments":[{"comment":"The assertion \"we must have z_delta in \\varpi^{2Z+1} O_D^\\times\" is load-bearing for the entire proof but is not proved. The parity of the valuation of the semilinear element z_delta is what makes the case split work, and the paper gives no derivation. The author should supply a lemma proving this parity statement, for example from the action of z_delta on the residue field of the unramified K_1 and the fact that D has invariant 1/2.","section":"§5.3, first paragraph"},{"comment":"The displayed identity \"if z_delta^2 \\in \\varpi_L^2 O_L^\\times, then Orb(1, alpha, s) = -q^s\" is inconsistent with Theorem 4.11. For v_L(w)=1, Theorem 4.11 gives Orb(1, alpha, s) = \\sum_{i=0}^{v_L(z^2)} (-q^s)^i. If v_L(z^2)=2, this equals 1 - q^s + q^{2s}, whose derivative at s=0 is +ln q, so the conjectural identity would give Int(delta) = -1, not the asserted Int(delta) = 1. If v_L(z^2)=1, the correct value is 1 - q^s, not -q^s. The proof must reconcile these formulas with the geometric rigid computation, or explain why the case v_L(z^2)=2 cannot occur for a matching pair.","section":"§5.3, rigid case"},{"comment":"The implications \"z_delta \\in \\varpi^2 O_D implies z_delta \\in \\varpi^3 O_D\", \"w must be a uniformizer of O_L\", and \"1 - z_delta \\in O_D^\\times\" are asserted without proof. These are exactly the hypotheses needed to apply Lemma 5.9 and reduce to the coquadratic GL(2) AFL, so they are load-bearing. The author should add a derivation from the relation z^2 = (w - \\varpi_3)(w - \\varpi_3^\\sigma) and the valuation parity of z_delta.","section":"§5.3, reduction case"},{"comment":"The claim that if \\delta_1(\\zeta_1) and z_\\delta generate the full ring O_D, then there is no non-trivial deformation and consequently Int(\\delta) = 1, is stated without a deformation-theoretic justification. Generation of the endomorphism ring does not by itself make the length of the intersection of the two CM cycles obvious; this step needs an argument, e.g., via the Lubin–Tate deformation functor.","section":"§5.3, rigid case"},{"comment":"In the ramified case, the displayed formula Orb(1, \\beta, s) = (1 - q^s + q^{2s}) + (1 - q^s)^2(q + \\cdots + q^r) gives Orb(1, \\beta, 0) = 1, but the theorem states Orb(1, \\beta, 0) = 2. The geometric computation in Theorem 4.15 gives 1, so the \"In particular\" line should be corrected to 1. This is a local correction, but as printed the value compared with the geometric side is wrong.","section":"Theorem 4.10"}],"minor_comments":[{"comment":"The notation \"OF[w] = OF[w]\" appears where the intended condition seems to be integrality of the order OF[w], presumably meaning OF[w] = O_L or a similar explicit condition. Please clarify the notation.","section":"Theorem 1.6 and §5.2"},{"comment":"The hypothesis is written as \"OF[w_\\beta] = O_L\", but in this lemma the pair is \\delta : (K_1, K_2) \\to D, so the symbol should almost certainly be w_\\delta (or the condition should be stated for the pair being reduced). Please correct the notation.","section":"Lemma 5.9"},{"comment":"The expression \"1 + c_L \\cdot (-q)^s + (-q^s)^2\" is easy to misread; the intended term appears to be c_L \\cdot (-q^s), not c_L \\cdot (-q)^s. Please use a consistent notation such as (-q^s)^i throughout.","section":"§4.1, proof of Theorem 4.10"},{"comment":"There are minor typographical issues, for example \"this reduction allow us\" in the abstract and the identical-looking conditions in Theorem 1.6, which should be corrected before publication.","section":"Abstract and Introduction"}],"recommendation":"major_revision","confidential_remarks":"The main concern is that Section 5.3, which proves the arithmetic half of Theorem 1.4, is not internally consistent as written. The parity claim for z_delta is unproved, and the rigid case contains an orbital-integral identity that appears to contradict Theorem 4.11; if a regular semisimple matching pair with v_D(z_delta)=1 and v_L(z_delta^2)=2 exists, the conjectural sign would fail. The author needs to resolve this case explicitly, not merely assert it away. This is substantial additional work, but it is the kind of gap that could in principle be fixed by a careful valuation analysis, so I recommend major revision rather than outright rejection."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Qirui Li proves both the biquadratic Guo–Jacquet fundamental lemma and the biquadratic linear AFL for GL4 with the unit test function, by reducing to the coquadratic GL2 case. That is a genuine new result if it holds: the biquadratic case was open beyond GL2, and the reduction mechanism in Section 5.2 is a real structural idea, not just a new computation. The combinatorial lattice-counting in Section 4 is explicit and largely checkable, and Theorems 1.5 and 1.6 are useful extras. This is not a framework paper; it does the work.\n\nNow the soft spots, in proportion. The proof of the arithmetic half, Theorem 5.10, rests on three underexplained claims. First, the parity assertion z_delta in \\varpi^{2Z+1} O_D^\\times is stated without proof; it may follow from a Frobenius/residue argument, but the paper does not give it. Second, in the first case the text asserts Orb(1, alpha, s) = -q^s when z_delta^2 in \\varpi_L^2 O_L^\\times. Theorem 4.11 gives 1 - q^s for v_L(z^2)=1 and 1 - q^s + q^{2s} for v_L(z^2)=2; the derivative at 0 of the former gives Int = 1, but the latter gives Int = -1 after the -1/ln q factor. So either the valuation normalization forces v_L(z^2)=1 in this case, which needs proof, or the conclusion can fail. Third, the step 'z_delta in \\varpi^2 O_D implies z_delta in \\varpi^3 O_D' and 'w must be a uniformizer of O_L' is asserted without derivation, and it is exactly what makes Lemma 5.9 applicable. If any regular semisimple orbit falls outside these cases, the reduction to the coquadratic GL2 AFL does not cover it. These are load-bearing, but they look fixable rather than fatal; the Section 4 computations are coherent and the reduction strategy is sound. There is also a small numerical slip: Theorem 4.10 prints Orb(1,beta,0)=2 in the ramified case while its own formula gives 1, and Theorem 4.15 proves the geometric side equals 1. That is clearly a typo, but it should be corrected.\n\nThis paper is for people working on relative trace formulas, arithmetic fundamental lemmas, and Lubin–Tate spaces. It deserves peer review: a serious referee can check the local computations and push on Section 5.3. My recommendation: send it to review, but ask the author to supply the missing valuation/parity arguments and fix the orbital integral identity before acceptance.","headline":"A serious and mostly explicit proof of the biquadratic GL4 FL/AFL, with a novel reduction to the coquadratic GL2 case, but the arithmetic half currently rests on unproved valuation claims that need to be fixed before acceptance.","tokens_in":25181,"tokens_out":3602,"would_cite":true,"duration_ms":36902,"reading_group":"maybe","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["11F70","11G18","11S37"],"pacs":[],"model":"deepseek-v4-flash","headline":"This paper proves the biquadratic Guo–Jacquet Fundamental Lemma and the biquadratic linear Arithmetic Fundamental Lemma for GL(4), for the unit test function.","keywords":["biquadratic fundamental lemma","arithmetic fundamental lemma","Guo–Jacquet fundamental lemma","orbital integrals","Lubin–Tate spaces","quadratic embeddings","GL(4)","unit test function"],"falsifier":"Find a regular semisimple pair of quadratic embeddings $(K_1,K_2)\\to D$, with $K_1/F$ unramified and $K_2/F$ ramified, whose $z_\\delta$ has even valuation; Section 5.3's case split would miss that orbit. A cheaper check is to enumerate the lattices in (4.1) for a small residue field, say $q=2$ with conductor $r=1$, and compare with the closed formula $\\mathrm{Orb}(\\mathbf{1},\\beta,s)=4-6\\cdot 2^s+4\\cdot 4^s$; any mismatch would disprove the explicit orbital-integral statement.","tokens_in":23921,"feed_emoji":"🧮","tokens_out":16453,"duration_ms":156204,"temperature":0.7,"pith_summary":"This paper proves two related identities at once: the biquadratic Guo–Jacquet Fundamental Lemma and the biquadratic linear Arithmetic Fundamental Lemma, both for the characteristic function of $\\mathrm{GL}_4(\\mathcal{O}_F)$ over a non-Archimedean local field. The first identity compares orbital integrals attached to two different pairs of quadratic embeddings; the second compares a derivative of one such orbital integral with an intersection number on a Lubin–Tate deformation space. The author reduces the $\\mathrm{GL}(4)$ statement to the already known coquadratic $\\mathrm{GL}(2)$ statement by studying the universal algebra generated by a pair of quadratic embeddings and splitting orbits according to the valuation of a canonical semilinear element $z$. If correct, this is the first proof of the biquadratic conjectures beyond the $\\mathrm{GL}(2)$ case, and it supplies explicit formulas for the orbital integrals in all regular semisimple orbits.","feed_headline":"Biquadratic fundamental lemma proved for GL4","feed_subtitle":"Both the Guo–Jacquet and linear arithmetic identities hold for GL(4), by reduction to the known GL(2) case.","key_machinery":"The central object is the coproduct $B_{K_1,K_2}=K_1\\amalg K_2$ in the category of $F$-algebras; under conditions (2.1)–(2.2) it is a quaternion algebra over $F[w]$ with canonical generators $w$ (which commutes with both embeddings) and $z$ (simultaneously $\\sigma_1$- and $\\sigma_2$-semilinear), subject to $(w-\\varpi_3)(w-\\varpi_3^{\\sigma_3})=z^2$. A pair of quadratic embeddings into a matrix algebra or a division algebra is exactly a morphism out of this coproduct, so matching orbits can be read from the images of $w$ and $z$. Orbital integrals are counted combinatorially as sums over lattices stable under $(\\mathcal{O}_{K_1},w,z)$, with a transferring factor $\\Omega(\\Lambda,s)=(-q^s)^{\\log_q[\\Lambda_-:z\\Lambda_+]}$ on the analytic side; the maximal-order reduction of Section 5.2 rewrites configurations with highly divisible $z$ as coquadratic configurations over $L=F[w]$, which is what allows the descent to the known $\\mathrm{GL}_2$ statement.","core_discovery":"The paper's central claim, Theorem 1.4, is that for every regular semisimple matching orbit,\n$$\\mathrm{Orb}(\\mathbf{1},\\$\\alpha$) = \\mathrm{Orb}(\\mathbf{1},\\$\\beta$,0)$$\nand\n$$\\mathrm{Int}(\\delta) = -\\frac{1}{\\ln q}\\,\\frac{d}{ds}\\Big|_{s=0}\\mathrm{Orb}(\\mathbf{1},\\$\\beta$,s),$$\nwhere $(\\alpha,\\beta,\\delta)$ are matching pairs of quadratic embeddings into $\\mathrm{GL}_4(F)$ and into the quaternion division algebra $D$, and the test function is the characteristic function of $\\mathrm{GL}_4(\\mathcal{O}_F)$. The proof separates orbits by the size of the semilinear element $z_\\delta$: when $z_\\delta$ is a unit times the uniformizer of the quaternion order, the intersection is rigid and the orbital integral is computed directly; when $z_\\delta$ is more highly divisible, $w$ is a uniformizer of $L=F[w]$ and the biquadratic configuration descends to a coquadratic pair over $L$, where the known coquadratic linear AFL for $\\mathrm{GL}_2$ applies. The same argument works over p-adic fields and over local fields of positive characteristic, and it covers all regular semisimple orbits.","pith_inferences":["The odd-valuation assertion about $z_\\delta$ in the quaternion order is likely a special case of a general structural fact about semilinear involutions in division algebras; promoted to a lemma, it would remove the only unproved case split and could extend the argument to other inner forms.","The lattice-counting formulas suggest a finite-field computer check: evaluating (4.1) for small $q$ and $r$ and comparing with Theorem 4.10 would verify the local identity independently, and the same enumeration could be run for $h=3$ as a numerical test of the higher biquadratic conjectures.","Because the reduction of Theorem 1.6 is conditional on integrality of $\\mathcal{O}_F[w]$, the real obstacle for higher rank is classifying orbits for which $\\mathcal{O}_F[w]$ is not integrally closed; those are also the orbits where the present argument cannot descend and where any generalization would need a new idea."],"forward_implications":["The biquadratic Guo–Jacquet and linear AFL conjectures are settled in their first open case, $h=2$, for the unit test function, over both p-adic fields and local fields of positive characteristic.","For every regular semisimple orbit with $L/F$ unramified the geometric orbital integral is $\\mathrm{Orb}(\\mathbf{1},\\beta)=2$, and with $L/F$ ramified it is $1$; the conductor $r$ of $\\mathcal{O}_F[w]$ affects only the higher coefficients of the $s$-series.","Whenever $\\mathcal{O}_F[w]$ is integral and $1-z$ is a unit, the biquadratic FL and linear AFL for $\\mathrm{GL}(2n)$ follow from the coquadratic Guo–Jacquet FL and the coquadratic linear AFL for the unit test function (Theorem 1.6), so any future coquadratic result transfers automatically.","The explicit formulas of Theorem 4.10 determine the derivative values that give the arithmetic intersection numbers, so the biquadratic linear AFL becomes a finite identity in $q$ and the conductor $r$ rather than an open conjecture."],"supporting_citations":[{"why":"Proves the coquadratic Guo–Jacquet fundamental lemma for the characteristic function of GL(2h,O_F), the identity that the biquadratic statement generalizes.","marker":"[3]"},{"why":"Defines the biquadratic FL and linear AFL, gives the Lubin–Tate intersection-number formula, and proves the GL2 cases that the present proof extends.","marker":"[4]"},{"why":"Introduces the linear AFL, computes intersection numbers, and supplies the coquadratic linear AFL for GL2 used in the final descent.","marker":"[7]"},{"why":"Provides the computational linear AFL for GL4 and the lattice-counting viewpoint that Section 3 adapts to the biquadratic setting.","marker":"[8]"},{"why":"Gives the reduction from hyperbolic to elliptic orbits and the spherical Hecke algebra structure used in the reduction formula.","marker":"[10]"}],"fun_headline_variants":["Biquadratic linear AFL proved for GL(4)","Biquadratic AFL for GL(4) via reduction to GL(2)","GL(4) biquadratic FL and linear AFL now proven","Proof of biquadratic Guo-Jacquet and linear AFL for GL(4)","Biquadratic FL and AFL for GL(4) established"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The proof rests on a structural claim stated in Section 5.3 without full derivation: every regular semisimple pair of quadratic embeddings into the quaternion division algebra $D$ has $z_\\delta$ of odd valuation, so that $z_\\delta$ is either a uniformizer times a unit or is divisible by $\\varpi^2$; if an orbit violated that parity statement, or if the implication that $z_\\delta\\in\\varpi^2\\mathcal{O}_D$ forces $w$ to be a uniformizer and $1-z_\\delta$ a unit failed, that orbit would not be covered.","fun_headline_variants_meta":{"raw":{"variants":["Biquadratic linear AFL proved for GL(4)","Biquadratic AFL for GL(4) via reduction to GL(2)","GL(4) biquadratic FL and linear AFL now proven","Proof of biquadratic Guo-Jacquet and linear AFL for GL(4)","Biquadratic FL and AFL for GL(4) established"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000228,"raw_usage":{"total_tokens":1489,"prompt_tokens":972,"completion_tokens":517,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":588,"completion_tokens_details":{"reasoning_tokens":421}},"tokens_in":588,"tokens_out":517,"duration_ms":4766,"temperature":1.0,"reasoning_tokens":421,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-07T13:03:45.080682+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Find a regular semisimple pair of quadratic embeddings $(K_1,K_2)\\to D$, with $K_1/F$ unramified and $K_2/F$ ramified, whose $z_\\delta$ has even valuation; Section 5.3's case split would miss that orbit. A cheaper check is to enumerate the lattices in (4.1) for a small residue field, say $q=2$ with conductor $r=1$, and compare with the closed formula $\\mathrm{Orb}(\\mathbf{1},\\beta,s)=4-6\\cdot 2^s+4\\cdot 4^s$; any mismatch would disprove the explicit orbital-integral statement.","supporting_citations":[{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Proves the coquadratic Guo–Jacquet fundamental lemma for the characteristic function of GL(2h,O_F), the identity that the biquadratic statement generalizes."},{"cited_title":"Howard and Q","cited_arxiv_id":null,"evidence_quote":"Defines the biquadratic FL and linear AFL, gives the Lubin–Tate intersection-number formula, and proves the GL2 cases that the present proof extends."},{"cited_title":"Li, An intersection formula for CM cycles on Lubin–Tate spaces.Duke Mathematical Journal 1.1 (2022): 1-89","cited_arxiv_id":null,"evidence_quote":"Introduces the linear AFL, computes intersection numbers, and supplies the coquadratic linear AFL for GL2 used in the final descent."},{"cited_title":"Li,A Computational Proof of the Linear Arithmetic Fundamental Lemma of GL4, Canadian Journal of Mathematics 74.2 (2022): 381-427","cited_arxiv_id":null,"evidence_quote":"Provides the computational linear AFL for GL4 and the lattice-counting viewpoint that Section 3 adapts to the biquadratic setting."},{"cited_title":"On the Linear AFL: The Non-Basic Case","cited_arxiv_id":"2208.10144","evidence_quote":"Gives the reduction from hyperbolic to elliptic orbits and the spherical Hecke algebra structure used in the reduction formula."}],"review_version":1}