{"id":"e2748907-49b4-4660-b0d4-6722ca713749","arxiv_id":"2412.13022","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":5.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":0,"one_line_summary":"For CM elliptic curves with j-invariant 0 or 1728, the p-Selmer rank of a twist is determined by the root number and a χ-component of a relative p-class group.","lead":"The paper shows that for certain complex-multiplication elliptic curves, the p-Selmer group size of any twist is fixed once you know the twist's root number and a certain piece of a relative ideal class group. The result turns Selmer rank computations into class-group computations and yields new families of number fields with large p-rank class groups.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The headline claim 'determined by root number and relative p-class group' rests on an unproved p-parity congruence; without it, s_p is only known to be r_p or r_p+1.","rationale":"The paper's central contribution is the claim that, for the CM twist families considered, the p-Selmer dimension is completely determined by the root number and a χ-component of a relative p-class group. The proof has two ingredients: Rubin's inequality r_p ≤ s_p ≤ r_p+1, and a parity statement that picks one of the two possible values. The second ingredient is asserted in the text as 'From p-parity conjecture, we know' with no proof, citation, or caveat. This is precisely the reader's weakest-assumption finding, and I agree it is the key soft point. The p-parity relation is not a consequence of the surrounding theorems; it is an external input. If that input is already a theorem for these CM curves (including the supersingular inert-prime cases), then the main theorems stand and the paper only needs a citation and a rewritten sentence. If it is not, the theorems are conditional, and every corollary that converts r_p=0 into s_p=0 or r_p≤1 into s_p=1 must be rephrased, along with the numerical ranks inferred from class-group data. I do not see a more fundamental defect in the paper: the field identifications and the reductions to relative class groups are plausible and substantially supported by Rubin's descent, and the computational section is presented as evidence rather than as proof. The correct remedy is therefore to make the status of the p-parity conjecture explicit and, if needed, state the theorems conditionally.","tokens_in":23422,"tokens_out":17968,"duration_ms":178013,"concrete_test":"Perform a literature/derivation check: determine whether the p-parity congruence dim_{F_p} Sel_p(E/Q) ≡ ω(E) (mod 2) is a theorem for the specific CM curves E with j=0 or 1728 and every odd p of good reduction, explicitly including the supersingular (inert) cases p≡3 mod4 and p≡5 mod6. If a proof exists (e.g. through Nekovář's Selmer complex parity computations or the Dokchitser p-parity theorem), the paper should cite it and the argument is complete; if no such proof exists, Theorems 2.3/2.4/3.2/3.3 and their corollaries must be restated as conditional on the p-parity conjecture. A useful sanity check is to recompute the first few rows of Table 1 (e.g. D=5,10,23, p=3) with an independent Sage/Magma 3-descent and compare s_3 mod 2 with the listed ω(E_D); any mismatch would refute the assumption.","verdict_should_be":"UNCHANGED","load_bearing_attack":"In the proofs of Theorems 2.3, 2.4, 3.2 and 3.3, the authors write 'From p-parity conjecture, we know that s_p(E_t) ≡ ω(E_t) (mod 2)' but give no proof or citation. This congruence is exactly what turns the Rubin inequality r_p ≤ s_p ≤ r_p + 1 into the advertised conclusion that ω(E_t) and r_p(E_t) uniquely determine s_p(E_t). Without it, the Selmer dimension is only known to lie in a two-element set. The same step is reused in Corollaries 2.5/2.6/3.4/3.5 (e.g. r_p=0 is converted into s_p=0, or r_p≤1 into s_p=1, only because parity forces the dimension to be even/odd), in Theorem 4.2/4.6 (where s_p ≥ 7 follows from oddness after s_p ≥ 6), and in the numerical tables. The curves considered have j=0 or 1728; for inert primes (p≡3 mod4 for j=1728, p≡5 mod6 for j=0) the reduction is supersingular, a case where p-parity is not an automatic consequence of the usual p-adic Iwasawa main conjectures. Thus the unqualified status of this parity statement is the single most load-bearing point in the paper.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper studies elliptic curves with j-invariant 0 or 1728 and their quadratic, cubic, quartic, and sextic twists. For each twist E_t and each good odd prime p, the authors define a relative p-class group attached to a subfield of the p-division field and denote by r_p(E_t) the dimension of the character component corresponding to the Galois action on E_t[p]. Using Rubin's descent theorem they prove inequalities r_p(E_t) ≤ s_p(E_t) ≤ 1+r_p(E_t), where s_p(E_t) is the F_p-dimension of the p-Selmer group. They then assert that the root number ω(E_t) and r_p(E_t) uniquely determine s_p(E_t), relying on the congruence s_p(E_t) ≡ ω(E_t) mod 2, which is invoked as the 'p-parity conjecture' without proof or citation. This determination is used to reformulate rank 0 and rank 1 p-converse theorems in terms of relative class groups, to derive criteria for congruent-number and cube-sum problems, to construct number fields with large p-rank class groups, and to produce numerical tables of ranks.","tokens_in":23707,"tokens_out":5212,"duration_ms":54354,"significance":"If the parity step were justified, the paper would provide an attractive and concrete bridge between Selmer ranks and relative class groups for CM elliptic curves, with explicit division-polynomial descriptions of the relevant fields and a clean numerical method. The construction of infinite families of degree 4 and 6 extensions with lower bounds on p-class group rank is also a potentially interesting contribution. The paper uses standard machinery (Rubin's descent, division polynomials, root-number formulas) and no free parameters are fitted; the claimed Selmer-to-class-group inequality is a genuine theorem of Rubin. However, the central 'complete determination' conclusion is conditional: the parity congruence is the only step that turns the two-element interval [r_p, r_p+1] into a unique value, and for the inert primes considered (p ≡ 3 mod 4 for j=1728 and p ≡ 5 mod 6 for j=0) the reduction is supersingular, where p-parity is not a standard unconditional consequence of Iwasawa main conjectures. The paper therefore currently overstates what is proved.","major_comments":[{"comment":"The sentence 'From p-parity conjecture, we know that s_p(E) ≡ ω(E) (mod 2)' appears in the proofs of all four theorems with no proof and no citation. This congruence is the only step that allows the authors to conclude that ω(E_t) and r_p(E_t) 'uniquely determine' s_p(E_t). Without it, the Selmer dimension is only known to lie in {r_p, r_p+1}, so the advertised conclusion is not established. The problem is particularly acute for inert primes, where the reduction is supersingular and p-parity is not an automatic consequence of the usual main conjectures. The authors should either provide a proof or a precise reference establishing the parity congruence in this supersingular CM setting, or replace the unconditional claims by explicitly conditional statements.","section":"Theorems 2.3, 2.4, 3.2, 3.3"},{"comment":"The equivalences in these corollaries use the same parity step to pass from r_p=0 to s_p=0 and from r_p≤1 to s_p=1. For example, in the proof of Corollary 2.5, the statement 'if r_p(E_D)=0, then by Theorem 2.3 or Theorem 2.4, we have s_p(E_D)=0' is only valid if the parity congruence is available. Likewise, in Corollary 2.8 the assertion that a congruent number forces s_p(E_n^1) ≥ 2 needs the parity input in addition to positivity of rank. These applications should be labeled as conditional on the same parity assumption, or the missing parity result should be supplied.","section":"Corollaries 2.5, 2.6, 3.4, 3.5 and Corollary 2.8"},{"comment":"The proof of Theorem 4.2 uses the step 'since for n∈S_p, we have ω(E_n)=-1 and hence s_p(E_n) is odd, and moreover as rk_Q E_n ≥ 6, we see that s_p(E_n) ≥ 7'. This uses the unproved parity congruence to exclude s_p(E_n)=6. If parity is not known for these supersingular primes, one only obtains s_p(E_n)≥6, hence r_p(E_n)≥5 instead of the stated r_p≥6, and the displayed lower bounds 12+rkCl(F) and 6+rkCl(F) in Theorem 4.2 must be reduced accordingly. The same issue affects Theorem 4.6. Thus the class-group construction is quantitatively dependent on the same load-bearing assumption.","section":"Theorems 4.2 and 4.6"},{"comment":"The field-identification arguments in the split/inert cases contain several asserted irreducibility and factorization statements that are not fully justified as written. For example, in the proof of Theorem 2.4 the sentence 'Consequently, f_{D,p}(X^4) has at-least one irreducible factor of degree (p-1) over K' and the subsequent case analysis require the reader to fill in nontrivial steps; similarly, in Theorem 3.3 the phrase 'proceeding as in the proof of Theorem 2.4, we conclude...' skips the analogous argument. Since the field identifications F_{D,p} ≅ F_{1,p} and L_{D,p} ≅ Q[X]/(f_{D,p}(X^4)) are used in the numerical applications and in the independence-of-D statements, these arguments should be expanded to a verifiable level of detail.","section":"Proofs of Theorems 2.4 and 3.3"}],"minor_comments":[{"comment":"The phrase 'which in-turn implies that s_p ≥ 2' is not immediate and should mention that the root number is +1 and that the parity congruence is being used.","section":"Corollary 2.8, proof"},{"comment":"The notation for primes p and p over p in the split case (Theorems 2.4 and 3.3) is sometimes ambiguous; for instance, statements such as 'p is unramified in L_{D,p}/K' and 'p is totally ramified in L_{D,p}/K' should be indexed by the chosen conjugate prime to avoid confusion.","section":"Section 2, notation"},{"comment":"The tables do not indicate which entries depend on the p-parity congruence in addition to GRH; given the central role of parity, the table headers or footnotes should make this dependence explicit.","section":"Section 5, tables"},{"comment":"There are several minor typographical and grammatical issues, e.g. 'p division field' should be 'p-division field' and 'deg ree six' should be 'degree six'; these do not affect the mathematics.","section":"Throughout"}],"recommendation":"major_revision","confidential_remarks":"The manuscript's main result is a conditional bridge: the 'unique determination' of Selmer rank from root number and relative class group rests entirely on an unproved and uncited p-parity congruence for supersingular primes. The underlying Rubin inequality and the explicit field computations are valuable, but the paper needs either a proof/citation for the parity statement or a systematic reframing of the main theorems and applications as conditional results. If the authors choose the conditional route, the quantitative claims in Section 4 and the numerical tables must be adjusted accordingly. I would not recommend acceptance in the current form, but the project appears fixable within the scope of a major revision."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Here's the one thing to know before you spend time on this: the paper has real content, but its headline conclusion is conditional in a way the authors don't advertise. The inequality r_p(E) ≤ s_p(E) ≤ 1+r_p(E) comes from Rubin's theorem. The new step is the field identifications and the claim that the relative p-class group component r_p, together with the root number, uniquely determines s_p. That last step uses the p-parity conjecture, s_p(E) ≡ ω(E) mod 2, which appears in the proofs of Theorems 2.3, 2.4, 3.2, 3.3 as 'From p-parity conjecture, we know' with no proof or citation. Without that congruence, you only get s_p ∈ {r_p, r_p+1}. For the inert primes (supersingular reduction), this parity is not an automatic consequence of the usual Iwasawa main conjectures, so the gap is real.\n\nWhat is genuinely new: the explicit division polynomial analysis showing F_D,p ≅ F_1,p (independence of the twist and of the prime), the explicit polynomial models for the division fields, and the applications to constructing number fields with large p-rank class groups (Theorems 4.2, 4.6). Those applications are legitimate as conditional statements—they need the same parity input to convert Selmer rank bounds into class group ranks. The Sage tables are a useful sanity check, though no code is shipped and some entries are GRH-dependent.\n\nThe good news is the flaw is fixable. Either the parity statement is known for these CM curves (cite it), or the theorems should be rephrased as 'assuming the p-parity conjecture, the p-Selmer rank is determined.' The rest of the arguments—the division polynomial identities, the ramification arguments, the non-isomorphism of the fields—look sound. I did not find circularity or fitted parameters.\n\nBottom line: this paper deserves a serious referee. A revision that honestly flags the parity input would make it acceptable. If you work on CM Selmer groups or class groups, the twist-independence of the division fields is worth citing.","headline":"Solid Rubin-refinement paper whose 'determines' claim rests on an unproved parity input; fix that and it's publishable.","tokens_in":24240,"tokens_out":3437,"would_cite":true,"duration_ms":30316,"reading_group":"maybe","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["11R29","11G05","11G15","11R23"],"pacs":[],"model":"deepseek-v4-flash","headline":"The paper claims that for CM elliptic curves with j-invariant 0 or 1728, the p-Selmer dimension of every twist is forced to be $r_p$ or $r_p+1$ by a relative p-class group, and parity then selects the exact value.","keywords":["relative p-class group","p-Selmer group","CM elliptic curves","root number","division fields","congruent number problem","cube sum problem","unboundedness of p-rank of class groups"],"falsifier":"Compute $s_p(E_t)$ and $r_p(E_t)$ for a twist with $r_p(E_t)=1$ and root number $\\omega(E_t)=+1$: the paper's central claim predicts $s_p(E_t)=1$, so a computation returning $s_p(E_t)=2$ would refute the claimed uniqueness. Equivalently, any single good prime and twist with $s_p(E_t) \\not\\equiv \\omega(E_t) \\pmod 2$ would falsify the parity step on which the exact determination rests.","tokens_in":23225,"feed_emoji":"🔢","tokens_out":16086,"duration_ms":125866,"temperature":0.7,"pith_summary":"This paper tries to establish that for elliptic curves with complex multiplication and j-invariant 0 or 1728, the dimension of the p-Selmer group of any admissible twist is determined by two pieces of arithmetic: the root number of the twist and the p-rank of one character-isotypic component of the relative p-class group of its p-division field. Concretely, the authors prove the two-point bound $r_p(E_t) \\leq s_p(E_t) \\leq 1+r_p(E_t)$, where $r_p(E_t)$ is that class-group dimension and $s_p(E_t)$ is the Selmer dimension; they then invoke the parity relation $s_p(E_t) \\equiv \\omega(E_t) \\pmod 2$ to force exactly one of the two integers. A reader should care because this turns Selmer-rank questions for these CM twist families into class-group computations in explicit number fields, giving concrete reformulations of rank-zero and rank-one converses, of the congruent-number problem, and of the cube-sum problem. It also lets large-rank elliptic curve families be converted into number fields whose class groups have large prescribed p-rank.","feed_headline":"Selmer rank of CM twists is fixed by root number and a class group","feed_subtitle":"For j=0 and j=1728 curves, one class-group component plus a parity rule pins the p-Selmer dimension.","key_machinery":"The load-bearing object is the relative p-class group $\\mathrm{Cl}(L_{C,p}/F_{C,p})[p]$ and its $\\chi_C$-isotypic component, the p-torsion classes of $L_{C,p}$ that transform by the same character as the p-torsion representation of $C$ and are not inherited from $F_{C,p}$. The paper splits the p-part of the class group as $\\mathrm{Cl}(L_{C,p})[p] \\cong \\mathrm{Cl}(L_{C,p}/F_{C,p})[p] \\times \\mathrm{Cl}(F_{C,p})[p]$, so a descent-theoretic dimension $\\dim_{\\mathfrak{k}_p}\\mathrm{Hom}(\\mathrm{Cl}(L_{C,p}), C[p])^G$ becomes a dimension of that relative component. Division polynomials for $y^2=x^3-Dx$ and $y^2=x^3+A$ supply explicit polynomials $f_{C,p}$ such that $F_{C,p}$ is presented as $K[X]/f_{C,p}(X)$ and $L_{C,p}$ is generated by a root of $f_{C,p}(X^k)$, with the same $F_{C,p}$ for all twists; these identities carry the argument that the class-group input is a single invariant attached to the fixed field.","core_discovery":"For each twist $C$ of the base curve, let $K$ be $\\mathbb{Q}(i)$ (when $j=1728$) or $\\mathbb{Q}(\\zeta_3)$ (when $j=0$), and let $L_{C,p}=K(C[p])$ be the field generated by the $p$-torsion points of $C$. The Galois group of $L_{C,p}/K$ is cyclic, and its natural character $\\chi_C$ describes the action on $C[p]$. The paper shows that for a cyclic subextension $F_{C,p}$ of degree $2$, $3$, $4$, or $6$ (according to the twist type), the dimension $r_p(C)=\\dim_{\\mathfrak{k}_p}\\mathrm{Cl}(L_{C,p}/F_{C,p})[p](\\chi_C)$ satisfies $r_p(C) \\leq s_p(C) \\leq 1+r_p(C)$, and that the fields $F_{C,p}$ themselves are independent of the twist up to isomorphism. Combined with the parity relation $s_p(C) \\equiv \\omega(C) \\pmod 2$, which the paper invokes, this pins $s_p(C)$ exactly. The proof passes from the full class group of the division field to the relative class group of a fixed extension, using explicit division-polynomial generators for the intermediate fields.","pith_inferences":["Implicit in the paper, the bottleneck is the p-parity conjecture: if that parity is proved for these CM twists, the main theorems become unconditional, while one parity failure would collapse the exact determination back to the two-point bound.","Because $F_{C,p}$ is twist-independent, the method suggests a practical test: computing the $\\chi$-component dimension in one fixed field should predict Selmer ranks across an entire twist family, so numerical discrepancies between class-group predictions and Selmer computations would pinpoint where the parity assumption fails.","The same relative-decomposition strategy may extend to other CM elliptic curves or to higher-degree twist types, provided division-polynomial identities of the same shape can be produced; the paper does not claim such extensions.","If the Selmer rank is unbounded in either CM twist family, Proposition 4.8 transfers that unboundedness to p-ranks of class groups in degree 4 and 6 extensions, linking two longstanding open problems; this conditional consequence is the paper's own observation, not a proof."],"forward_implications":["For a twist $E_t$ with $r_p(E_t)=0$, the bound forces $s_p(E_t)$ to be $0$ or $1$, and the paper's parity step then decides whether the curve has rank zero or rank one.","Rank-zero and rank-one p-converse theorems for these CM curves become equivalent to statements about the $\\chi$-component of a relative p-class group: trivial for rank zero, cyclic for rank one, checked at a single good prime.","For squarefree $n \\equiv 1,2,3 \\pmod 8$, the congruent-number status of $n$ is controlled by the relative quadratic class group: if $n$ is congruent, every prime $p\\nmid 2n$ divides the relative class number $|\\mathrm{Cl}(L_{n^2,p}/K_{1,p})|$, while one prime with trivial component proves $n$ is not congruent.","For the cube-sum problem, a cube-free $n$ with root number $+1$ that is a rational cube sum forces every prime $p\\nmid 6n$ to divide the relative cubic class number $|\\mathrm{Cl}(L_{-432n^2,p}/K_{-432,p})|$.","Large-rank elliptic curve families yield infinitely many number fields of degree 4 (for $j=1728$) or degree 6 (for $j=0$) over the fixed field $F_{1,p}$ with p-class rank at least $12+\\mathrm{rk}\\,\\mathrm{Cl}(F_{1,p})[p]$ or the corresponding $6+\\mathrm{rk}$ bound, giving concrete progress on the unboundedness conjecture for class group p-ranks."],"supporting_citations":[{"why":"Supplies the initial descent inequality $r_p(C)\\le s_p(C)\\le 1+r_p(C)$ for the full division-field class group, on which every main theorem builds.","marker":"[Rub87a]"},{"why":"Gives the rank-zero L-value result and the Grossencharacter conductor and ramification lemmas used to identify the division fields in the $p\\equiv 1 \\pmod 4$ and $p\\equiv 1 \\pmod 6$ cases.","marker":"[CW77]"},{"why":"Provides the p-converse theorem used to convert vanishing of the Selmer group back into nonvanishing of the L-value in the rank-zero corollaries.","marker":"[Rub91]"},{"why":"Provides the p-converse theorem used to convert a one-dimensional Selmer group into a simple zero of the L-function in the rank-one corollaries.","marker":"[BT20]"},{"why":"Used with the Euler-system descent to pass from a simple zero of $L$ to rank one and finiteness of the Tate-Shafarevich group in the rank-one corollaries.","marker":"[GZ86]"},{"why":"Supplies the Euler-system descent that completes the rank-one converse step when combined with [GZ86].","marker":"[Kol90]"},{"why":"Gives the infinite family of elliptic curves $y^2=x^3-Dx$ with rank at least 6 used to construct large p-rank class groups in Theorem 4.2.","marker":"[Kih04]"},{"why":"Gives the infinite family of elliptic curves $y^2=x^3+A$ with rank at least 7 used to construct large p-rank class groups in Theorem 4.6.","marker":"[Mes92]"},{"why":"Supplies the power-free value results for binary forms used to produce infinitely many square-free twist parameters with the required root number.","marker":"[Gre92]"}],"fun_headline_variants":["Selmer rank of CM twists: equality with relative class rank up to parity","Root number and one relative class group fix Selmer dimension for twists","Twist's p-Selmer size is class rank or one more, parity tells which","Relative p-class group plus root number determine p-Selmer dimension exactly","For j=0,1728 twists, a parity rule and class group pin Selmer rank"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The load-bearing premise is the unproved p-parity conjecture for these CM twists, used to decide between the two consecutive integers allowed by the bound; if that parity fails, the same theorems only give $r_p \\le s_p \\le 1+r_p$, not an exact value.","fun_headline_variants_meta":{"raw":{"variants":["Selmer rank of CM twists: equality with relative class rank up to parity","Root number and one relative class group fix Selmer dimension for twists","Twist's p-Selmer size is class rank or one more, parity tells which","Relative p-class group plus root number determine p-Selmer dimension exactly","For j=0,1728 twists, a parity rule and class group pin Selmer rank"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000984,"raw_usage":{"total_tokens":4170,"prompt_tokens":937,"completion_tokens":3233,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":553,"completion_tokens_details":{"reasoning_tokens":3128}},"tokens_in":553,"tokens_out":3233,"duration_ms":24035,"temperature":1.0,"reasoning_tokens":3128,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-11T13:30:29.204283+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Compute $s_p(E_t)$ and $r_p(E_t)$ for a twist with $r_p(E_t)=1$ and root number $\\omega(E_t)=+1$: the paper's central claim predicts $s_p(E_t)=1$, so a computation returning $s_p(E_t)=2$ would refute the claimed uniqueness. Equivalently, any single good prime and twist with $s_p(E_t) \\not\\equiv \\omega(E_t) \\pmod 2$ would falsify the parity step on which the exact determination rests.","supporting_citations":[],"review_version":1}