{"id":"85e1fd94-f07b-4a49-9efc-a1820fa92776","arxiv_id":"2507.13199","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":7.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":0,"one_line_summary":"The degrees of points with rational j-invariant on X0(n) and X1(n) are classified for all n, unconditionally for infinitely occurring degrees and assuming Zywina's conjecture for finitely occurring ones.","lead":"This paper classifies the possible degrees of points with rational j-invariant on the modular curves X0(n) and X1(n), for every level n. It is an unconditional classification for degrees that occur infinitely often, and a classification conditional on a conjecture of Zywina for degrees that occur only finitely often.","discovery_kind":"new_method","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Unconditional Theorems 1.1–1.2 depend on an unverified LMFDB completeness claim; no snapshot or independent enumeration is given, so a missing conjugacy class could silently drop a degree.","rationale":"The reader identified the same load-bearing concern: the unconditional theorems depend on a database completeness assumption that is asserted, not proved or pinned. I agree that this is the most serious weakness. The mathematical architecture around H-closures appears coherent: Corollary 3.3 correctly reformulates degrees through orbit data, Theorem 4.4 supplies the required Hilbert-irreducibility input for genus 0 curves, and the conditional part is honestly marked as depending on Conjecture 5.5. The main risk is not in the group-theoretic framework but in the exhaustive-search step that turns an LMFDB query into the finite tables. Since the manuscript gives no versioned snapshot and no independent enumeration, a missing conjugacy class would not be detected by reading the paper. This does not move the reader's verdict from CONDITIONAL; it reinforces it. I would keep the verdict conditional pending a reproducible, versioned database query or an independent enumeration, and I would also request a commit hash for the GitHub code so the computations can be rerun.","tokens_in":57670,"tokens_out":10247,"duration_ms":118188,"concrete_test":"Pin a fixed LMFDB release (e.g., the 10 June 2025 snapshot with a git hash or DOI), then independently regenerate Tables 1 and 2 without relying on LMFDB completeness: enumerate all conjugacy classes of subgroups of GL2(Z/nZ) with full determinant for every n in the set S appearing in Section 4.2 (and, as a safety margin, all n <= 70) using a standalone group-theoretic tool such as Magma's Subgroups or GAP, compute the B0(n)- and B1(n)-closures, filter by the six conditions of Section 4.2, and compare the resulting lists exactly with Tables 1 and 2. If the independent enumeration produces any additional B0- or B1-closed class of level <= 70 with full determinant, genus <= 1, and X_G(Q) infinite, the unconditional theorems fail as stated.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The unconditional classification rests on Theorem 4.2, which is obtained by querying the LMFDB for all full-determinant conjugacy classes of open subgroups of GL2(\\hat Z) of level at most 70. Section 4.2 states this completeness as a fact ('at the time of writing, the LMFDB contains a list of all conjugacy classes ... of level at most 70') and also says that the rational-point data is incomplete but 'complete for all subgroups G which are required in our application.' No LMFDB version, snapshot, or independent certificate is supplied, and the URL is a live beta site. Because Theorems 1.1 and 1.2 are exhaustive classifications, a single omitted conjugacy class with full determinant, level in S, genus 0 or 1, and X_G(Q) infinite would change the D^∞ sets and therefore the final degree lists. This is a genuine correctness risk, not just a reproducibility nicety: the database completeness assertion is load-bearing and currently unverifiable from the manuscript. The conditional theorems are explicitly labeled as relying on Conjecture 5.5, so that assumption is not the weak point; the weak point is the apparently unconditional finite enumeration behind the unconditional theorems.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper introduces the notions of H-equivalence and H-closures for subgroups of GL2(Ẑ) and uses them to translate the problem of degrees of points with rational j-invariant on X0(n) and X1(n) into finite orbit computations for adelic Galois images. Theorems 1.1 and 1.2 give an unconditional classification of the degrees that occur infinitely often, via the sets D∞0(m) and D∞1(m). Theorems 1.3 and 1.4 give the analogous classification of all degrees of non-cuspidal, non-CM points with rational j-invariant, conditional on Zywina's Conjecture 5.5. Theorem 1.5 applies the same methods to isolate four possible j-invariants of isolated points on X1(n).","tokens_in":57940,"tokens_out":6424,"duration_ms":83764,"significance":"If the computational inputs are correct, this is a substantial and useful classification: the H-closure formalism cleanly separates the Galois-theoretic core from the modular-curve computations, and the orbit-to-degree translation in Corollaries 2.2 and 3.3 is elegant and sound. The Hilbert irreducibility argument for infinitely occurring closures (Theorem 4.4 and Corollary 4.6) is a nice contribution, and Section 6.5 contains a self-contained Chabauty computation for one exceptional genus-3 curve. The paper also ships Magma code, which is a strength. However, the unconditional theorems depend on a completeness assertion about the LMFDB that is not accompanied by a version, snapshot, or certificate, and the finite-part computations are summarized rather than independently verifiable from the manuscript alone.","major_comments":[{"comment":"The unconditional Theorems 1.1 and 1.2 rest on the assertion that the LMFDB contains all conjugacy classes of open subgroups of GL2(Ẑ) of level at most 70 with full determinant, and that the associated rational-point data are complete for the subgroups 'required in our application.' No LMFDB version, frozen snapshot, or independent certificate is supplied. Because Theorems 1.1 and 1.2 are exhaustive classifications, a single omitted conjugacy class of genus 0 or 1 with infinitely many rational points would alter the D∞0(m) or D∞1(m) sets and hence the final degree lists. The phrase 'complete for all subgroups G which are required in our application' is also circular as written, since the set of required subgroups is exactly what the enumeration is supposed to determine. Please supply a versioned database snapshot or an independent enumeration from the Cummins–Pauli data, together with scripts that verify completeness of the list in Tables 1 and 2.","section":"§4.2, Theorem 4.2"},{"comment":"The classification of finitely occurring closures depends on a Magma enumeration of 2651 B1-closed conjugacy classes that is presented only as Figure 1, and on rational-point computations for 160 modular curves summarized in Table 5 by method, without machine-readable output or exact versions of Magma and the LMFDB. This part is conditional on Conjecture 5.5, so it is not a threat to the unconditional theorems, but Tables 6–8 are load-bearing for the finite-degree theorems and for Theorem 1.5. Please include the enumeration data, the exact scripts with version information, or an independent verification of the exceptional j-invariants listed in Table 6.","section":"§5.2, Theorem 5.9 and Table 6"}],"minor_comments":[{"comment":"The title contains spacing artifacts: 'RA TIONAL j-INV ARIANT' should read 'RATIONAL j-INVARIANT'. Please correct throughout.","section":"Title and Abstract"},{"comment":"There is a typo: 'we showed that it the problem of determining' should read 'we showed that the problem of determining'.","section":"§4.3, first paragraph"},{"comment":"The text says 'By Theorem 3.11', but the relevant statement is Lemma 3.11; please correct the cross-reference.","section":"§4.2, bullet on B0/B1-closed subgroups"},{"comment":"In the sentence beginning 'Therefore, Theorem 7.1, the j-invariant...', a preposition is missing; it should read 'Therefore, by Theorem 7.1, the j-invariant...'.","section":"§7, Remark 7.3"},{"comment":"Several entries in Table 4 are written as products such as '9 · 12063' and '51 · 78843'; this notation is not explained and the values do not look like the surrounding j-invariants. Please clarify whether these are products or typesetting artifacts.","section":"Table 4"}],"recommendation":"major_revision","confidential_remarks":"The central theoretical framework is sound and the paper is likely correct in substance, but the advertised unconditional theorems currently rest on an unverifiable LMFDB completeness claim. If the author can supply a frozen database snapshot or an independent enumeration certifying Tables 1 and 2, I would view the paper as acceptable; without that, the unconditional classification is not yet reproducible."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Kenji Terao's paper gives the first full classification of the degrees of rational-j points on X0(n) and X1(n), split into infinite (unconditional) and finite (conditional on Zywina's conjecture). The main new tool is the H-closure framework, which cleanly translates the orbit structure of Galois images into degree lists. The reduction in Corollaries 2.2 and 3.3 is solid, and the application to isolated j-invariants (Theorem 1.5) is a genuine bonus.\n\nThe infinite-degree theorems (1.1 and 1.2) are the core contribution. The logic from H-closures to the finite list of candidates is coherent, and the orbit computations are straightforward to redo. The conditional theorems are clearly labeled and the reliance on Zywina's conjecture is explicit, so no issue there.\n\nThe soft spot is exactly where the stress-test note points: Theorem 4.2, which feeds the unconditional theorems, is obtained by querying the LMFDB for all full-determinant conjugacy classes of level at most 70. The paper asserts that the database contains all such classes 'at the time of writing' but gives no snapshot, no version, and no independent enumeration. The rational-point data is also incomplete, with only a claim that it is complete for the subgroups needed. Since Theorems 1.1 and 1.2 are exhaustive classifications, a single missing conjugacy class with full determinant, level in S, and infinite rational points would change the D∞ sets. This is a correctness risk, not just a reproducibility nicety. The code is on GitHub but not pinned to a commit.\n\nThe author is aware of the frailty—Remark 4.3 notes the surprising fact that all the relevant genus-1 curves have rank 0, which smells like a selection effect but is not itself a flaw.\n\nOverall: the mathematical architecture is sound and the conditional results are trustworthy conditional on the conjecture. The unconditional results need a verifiable input: an LMFDB snapshot or an independent enumeration of the relevant subgroups. This is fixable, and the paper deserves a serious referee. I'd send it to review with a request to add a snapshot or an independent verification of the completeness claim.","headline":"New H-closure framework, but the unconditional classification depends on an unpinned LMFDB completeness claim that needs a snapshot or independent enumeration.","tokens_in":58418,"tokens_out":1786,"would_cite":true,"duration_ms":20720,"reading_group":"maybe","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["11G18","11F80","14G05","11G05"],"pacs":[],"model":"deepseek-v4-flash","headline":"This paper classifies all degrees of closed points with rational $j$-invariant on the modular curves $X_0(n)$ and $X_1(n)$.","keywords":["modular curves","rational j-invariant","Galois representations","H-closures","degrees of points","isolated points","X0(n)","X1(n)"],"falsifier":"Exhibit a conjugacy class $G$ of open subgroups of $\\operatorname{GL}_2(\\widehat{\\mathbb{Z}})$ of level at most 70 with full determinant, genus at most 1, with $G$ equal to its own $B_0(n)$- or $B_1(n)$-closure for some $n$, and with $X_G(\\mathbb{Q})$ infinite, that is not listed in Table 1 or Table 2; equivalently, exhibit a rational non-CM $j$ and an $n$ whose $B_0(n)$- or $B_1(n)$-closure is not among the classes listed in Tables 1/7 or 2/8. A single such example would disprove the corresponding theorem.","tokens_in":2241,"feed_emoji":"🧮","tokens_out":5370,"duration_ms":125176,"temperature":0.7,"pith_summary":"The paper determines, for every $n$, the set of degrees $d$ for which $X_0(n)$ or $X_1(n)$ has closed points of degree $d$ with rational $j$-invariant. For degrees occurring for infinitely many such points, the classification is unconditional and is expressed as a union over divisors $m$ of $n$ of sets $d \\cdot \\deg(X(n)\\to X(m))$ with $d$ in explicit finite lists. For degrees occurring only finitely often, the paper gives an analogous classification of all degrees of non-cuspidal, non-CM points, conditional on a broad conjecture stated in the paper. The engine is a new group-theoretic invariant, the $H$-closure of a Galois image, which exactly encodes the orbit sizes that determine these degrees. If correct, this yields a complete computable answer to a natural arithmetic-geometry question and also settles the conjectural list of rational $j$-invariants of isolated points on $X_1(n)$.","feed_headline":"All degrees of rational-j points on X0(n) and X1(n) found","feed_subtitle":"Unconditional for infinitely many points; conditional for the rest; isolated j-invariants settled.","key_machinery":"The central object is the $H$-closure of a subgroup $G$ of $\\operatorname{GL}_2(\\widehat{\\mathbb{Z}})$: the unique maximal overgroup of $G$ whose right action on the coset space $H\\backslash \\operatorname{GL}_2(\\widehat{\\mathbb{Z}})$ produces the same orbit decomposition as the action of $G$ itself. A related finer notion, the $H$-closure, records the full orbit decomposition rather than only the orbit of the identity coset. The size of such an orbit is exactly the degree of a closed point on the modular curve $X_H$ with a given rational $j$-invariant, so computing $B_0(n)$- and $B_1(n)$-closures of adelic Galois images attached to elliptic curves over $\\mathbb{Q}$ computes the degrees of rational-$j$ points and fibers on $X_0(n)$ and $X_1(n)$. The paper computes these closures for the relevant Galois images: unconditionally for closures occurring infinitely often, and conditionally on Conjecture 5.5 for those occurring only finitely often, with the rational-point computation on 160 auxiliary modular curves as the final step.","core_discovery":"The paper's central claim is that the degrees of points with rational $j$-invariant on $X_0(n)$ and $X_1(n)$ are exactly the union over divisors $m$ of $n$ of the multiples $d \\cdot \\deg(X_0(n)\\to X_0(m))$ and $d \\cdot \\deg(X_1(n)\\to X_1(m))$, with $d$ ranging over explicit finite sets: unconditionally for degrees occurring infinitely often, and conditionally on Conjecture 5.5 for degrees occurring finitely often, excluding CM $j$-invariants. The classification is finer than a statement about individual point degrees: it gives the full multiset of degrees in each rational fiber of the $j$-map, and the point-degree theorems are deduced from this fiber-level computation. The same machinery also yields that, assuming Conjecture 5.5, the only rational $j$-invariants of non-cuspidal, non-CM isolated closed points on any $X_1(n)$ are the four numbers listed in Theorem 1.5.","pith_inferences":["The $H$-closure framework is not tied to the specific subgroups $B_0(n)$ and $B_1(n)$; the same orbit-size dictionary should work for any open subgroup $H$ of $\\operatorname{GL}_2(\\widehat{\\mathbb{Z}})$, giving a general method for degree classifications on arbitrary modular curves.","The observation that all genus-1 candidates in Tables 1 and 2 have analytic rank zero suggests a hidden structural constraint relating $B_0(n)$- and $B_1(n)$-closed subgroups to modular ranks, which could simplify the finite classification.","A direct testable extension is to re-run the enumeration with more complete database coverage beyond level 70; any newly appearing conjugacy class would force a revision of the tables in the paper.","The conditional finite classification could be converted into an unconditional statement for each fixed $n$ by proving the underlying conjecture for only the finitely many $j$-invariants that actually appear, a potentially more tractable task than the full conjecture."],"forward_implications":["For every $n$, the full set of degrees of rational-$j$ points on $X_0(n)$ and $X_1(n)$ can be computed by a finite algorithm without case-by-case analysis of elliptic curves.","The infinite-degree classification is unconditional, so the explicit sets $D^\\infty_0(m)$ and $D^\\infty_1(m)$ give exact answers for infinitely occurring degrees for all $n$.","If Conjecture 5.5 is proved, the finite-degree lists $D_0(m)$ and $D_1(m)$ become unconditional, completing the classification for non-CM points.","The fiber-level computation refines the point-degree classification by recording how the degrees split above each rational $j$-invariant, not merely which degrees occur.","The isolated-point result reduces the possible rational $j$-invariants of non-cuspidal, non-CM isolated points on $X_1(n)$ to four explicit numbers, matching the conjecture stated in the paper's reference [3]."],"supporting_citations":[{"why":"Supplies the database of conjugacy classes, genera, and rational-point data used in the Section 4.2 enumeration.","marker":"[20]"},{"why":"Provides the conjectural classification of agreeable closures of Galois images (Conjecture 5.5) on which the finite part rests.","marker":"[36]"},{"why":"Gives the classification of genus at most 24 congruence subgroups used to restrict the possible levels.","marker":"[6]"},{"why":"Supplies the $\\ell$-adic Galois image classifications used for the prime-power level curves in Section 6.","marker":"[26]"},{"why":"Gives the 2-adic Galois image classification used to handle level $2^n$ curves.","marker":"[27]"},{"why":"Supplies the conjectural list of isolated $j$-invariants and the algorithm used to filter the candidates in Theorem 7.2.","marker":"[3]"},{"why":"Supplies the framework of isolated points on modular curves and the single-source theorem that the $H$-equivalence argument generalizes.","marker":"[31]"}],"fun_headline_variants":["Degrees of rational-j points on X0(n) and X1(n) fully classified","Rational-j point degrees on X0(n), X1(n): complete classification","Exact degree sets of rational-j points on modular curves X0(n), X1(n)","Degrees of rational-j points on X0(n), X1(n): unconditional and conditional","Solving degree classification for rational-j points on X0(n) and X1(n)"],"cache_read_input_tokens":60672,"weakest_assumption_plain":"The unconditional classification assumes that the online database of conjugacy classes of open subgroups of $\\operatorname{GL}_2(\\widehat{\\mathbb{Z}})$ of level at most 70 with full determinant, together with its genus and rational-point data, is complete; the finite classification additionally assumes Conjecture 5.5.","fun_headline_variants_meta":{"raw":{"variants":["Degrees of rational-j points on X0(n) and X1(n) fully classified","Rational-j point degrees on X0(n), X1(n): complete classification","Exact degree sets of rational-j points on modular curves X0(n), X1(n)","Degrees of rational-j points on X0(n), X1(n): unconditional and conditional","Solving degree classification for rational-j points on X0(n) and X1(n)"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000666,"raw_usage":{"total_tokens":3023,"prompt_tokens":915,"completion_tokens":2108,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":531,"completion_tokens_details":{"reasoning_tokens":1997}},"tokens_in":531,"tokens_out":2108,"duration_ms":18475,"temperature":1.0,"reasoning_tokens":1997,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-06T16:29:42.938603+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Exhibit a conjugacy class $G$ of open subgroups of $\\operatorname{GL}_2(\\widehat{\\mathbb{Z}})$ of level at most 70 with full determinant, genus at most 1, with $G$ equal to its own $B_0(n)$- or $B_1(n)$-closure for some $n$, and with $X_G(\\mathbb{Q})$ infinite, that is not listed in Table 1 or Table 2; equivalently, exhibit a rational non-CM $j$ and an $n$ whose $B_0(n)$- or $B_1(n)$-closure is not among the classes listed in Tables 1/7 or 2/8. A single such example would disprove the corresponding theorem.","supporting_citations":[{"cited_title":"The L-functions and modular forms database","cited_arxiv_id":null,"evidence_quote":"Supplies the database of conjugacy classes, genera, and rational-point data used in the Section 4.2 enumeration."},{"cited_title":"Congruence subgroups of PSL(2 , Z) of genus less than or equal to 24","cited_arxiv_id":null,"evidence_quote":"Gives the classification of genus at most 24 congruence subgroups used to restrict the possible levels."},{"cited_title":"ℓ-adic im- ages of Galois for elliptic curves overQ (and an appendix with John Voight)","cited_arxiv_id":null,"evidence_quote":"Supplies the $\\ell$-adic Galois image classifications used for the prime-power level curves in Section 6."},{"cited_title":"Elliptic curves over Q and 2-adic images of Galois","cited_arxiv_id":null,"evidence_quote":"Gives the 2-adic Galois image classification used to handle level $2^n$ curves."},{"cited_title":"Towards a classification of isolated j-invariants","cited_arxiv_id":null,"evidence_quote":"Supplies the conjectural list of isolated $j$-invariants and the algorithm used to filter the candidates in Theorem 7.2."},{"cited_title":"Isolated points on modular curves","cited_arxiv_id":null,"evidence_quote":"Supplies the framework of isolated points on modular curves and the single-source theorem that the $H$-equivalence argument generalizes."}],"review_version":1}