REVIEW
Local Factors in the BSD Conjecture: A Unified Statistical View
Number Theory · Submitted 7 Oct 2026 · 10 references
Pith's verdict: A solid arithmetic-statistics paper whose master generating function is probably right, but whose p=3 local density correction needs an independent Tate-algorithm check before the numerical constants are trusted.
T0 · AI review
Why Pith says this
Highlight any text to comment on it.
Referee report
The paper
The paper claims
One explicit Euler product is claimed to govern all Tamagawa statistics of elliptic curves by naive height: densities π₀ ≈ 0.5054, π₁ ≈ 0.4220, π₂ ≈ 0.0684, strictly negative cross-prime covariances, and all moments finite.
The reading
What carries the argument
The load-bearing object is the finite set $\mathcal{K}$ of local reduction data, each element recording a Kodaira symbol together with split/nonsplit information and the component group, with the map $c(K)$ giving the local Tamagawa number in the same way that $c_p(E)$ is computed from a $p$-minimal model. The three trivial data $K_0=\{I_0,I_1^{\mathrm{sp}},I_1^{\mathrm{ns}}\}$ all satisfy $c(K)=1$; Lemma 2.4 shows the remaining local mass $\sum_{K\notin K_0}\delta_p(K)=1-\beta_p$ is $O(p^{-2})$, which is exactly what makes the infinite product converge absolutely. Three ingredients carry the argument from local to global: equidistribution of short Weierstrass models modulo every fixed modulus (Lemma 2.6), a uniform large-prime tail estimate for models with $\tau_p(E)\notin K_0$ (Lemma 2.7), and a finite-prime factorization of densities (Proposition 2.9) built on the local constancy of the reduction datum on the nonsingular locus. Coefficient extraction then uses the Vitali–Montel normal-family theorem and Cauchy's coefficient formula. The explicit densities $\delta_p(t)$ come from the predecessor paper's tables for Tate's algorithm, with two corrections: the $p\ge5$, $t=4$ numerator becomes $3p^2-2p+1$ (Remark 2.2, checked by a normalization identity), and the $p=3$ values for $t=1,2,3$ are revised because a Case 9(iii) family is reclassified from type III* to type IV* (Remark 2.3).
Extended reading notes
Core claim
The central claim is Theorem 1.1: for any finite set $\Lambda$ of non-trivial local reduction data and any finite set $\Pi$ of primes, the limiting joint generating function of the count of contributing primes, $\Omega(\mathrm{Tam}(E))$, the valuation vector, and the per-datum counts converges, on the closed unit domain, to the absolutely convergent Euler product $\prod_p(\beta_p+\sum_{K\notin K_0}\delta_p(K)\Phi_p(K))$, where $K_0=\{I_0,I_1^{\mathrm{sp}},I_1^{\mathrm{ns}}\}$ collects the three local data with $c(K)=1$ and $\delta_p(K)$ is the Haar-measure density of a given local reduction datum. Reading off coefficients gives the limiting probabilities $\pi_0\approx0.50540113$, $\pi_1\approx0.42200272$, and $\pi_2\approx0.06841047$ for zero, one, or two primes contributing nontrivially to $\mathrm{Tam}(E)$, and the identity $\pi_0=P_{\mathrm{Tam}}(1)$ identifies the trivial-Tamagawa density with the density of curves whose Tamagawa product is exactly one. The same product yields a covariance $\mathrm{Cov}(v_\ell(\mathrm{Tam}(E)),v_{\ell'}(\mathrm{Tam}(E)))$ that is strictly negative for every distinct pair $\ell,\ell'$ (about $-0.02489$ at $\ell=2,\ell'=3$) and decays like $O((\ell\ell')^{-2})$, a correlation of about $0.9801$ between $\omega_{\mathrm{Tam}}(E)$ and $\Omega(\mathrm{Tam}(E))$, and the Euler product $\prod_p\sum_{t\ge1}\delta_p(t)t^k$ for every $k$-th moment, giving super-polynomial decay of the tail $P_{\mathrm{Tam}}(m)$.
Load-bearing premise
Every numerical conclusion rests on the corrected explicit local densities $\delta_p(t)$ of Theorem 2.1 — in particular the $p\ge5$, $t=4$ numerator $3p^2-2p+1$ and the three revised $p=3$ values — so if either correction is wrong, the constants $\pi_0,\pi_1,\pi_2$, the covariance $-0.02489$, and the correlation $0.9801$ all fail, even though the formal Euler-product framework would survive.
Editorial extensions
If this is right
- The densities $\pi_0,\pi_1,\pi_2$ become sharp predictions: any enumeration of curves by naive height must converge to $0.50540113\ldots$, $0.42200272\ldots$, and $0.06841047\ldots$, with the remaining $0.42\%$ of curves having three or more contributing primes.
- The strict negativity of $\mathrm{Cov}(v_\ell,v_{\ell'})$ for every distinct pair of primes is a repulsion among local factors: conditioning on one prime dividing $\mathrm{Tam}(E)$ lowers the chance that another does, with the effect bounded by $C/(\ell\ell')^2$.
- All moments of the limiting Tamagawa distribution are finite and given by $\prod_p\sum_{t\ge1}\delta_p(t)t^k$, so the tail $P(\mathrm{Tam}(E)\ge M)$ decays faster than $M^{-k}$ for every fixed $k$; the first moment is $\mathbb{E}[\mathrm{Tam}(E)]\approx1.81861587$, correcting the value $1.8193\ldots$ quoted in the predecessor paper.
- One master product answers all four questions of the introduction: its specializations recover the Dirichlet series of the Tamagawa densities, the joint law of $(\omega_{\mathrm{Tam}}(E),\mathrm{Tam}(E))$, the distribution of $\Omega(\mathrm{Tam}(E))$, the fixed $\ell$-adic valuations, and the joint counts of individual reduction data.
- Because $\pi_0=P_{\mathrm{Tam}}(1)\approx0.50540113$, more than half of all curves in this height ordering have Tamagawa product exactly one, so the typical curve contributes no nontrivial component-group factor to the algebraic side of the BSD formula.
Reading between the lines
- I infer a conservation picture at the level of local factors: the negative covariance means the component groups at distinct primes cannot all be large at once, and the same machinery should yield analogous anti-correlations for other additive statistics built from local reduction data, such as statistics of the conductor $N_E$, for which $\omega_{\mathrm{Tam}}(E)\le\omega(N_E)$ always holds.
- I note that the $p=3$ correction is invisible to the two checks a reader would most naturally run — the normalization $\sum_t\delta_3(t)=1$ and the first moment $\sum_t t\delta_3(t)$ both survive the shift unchanged — so independent verification must target $\delta_3(1)$ itself or second-order statistics such as the covariance, rather than the mean.
- I read the reported machine-checked formalization of the theorems as a separate layer from the numerical evaluations: the entries of Table 1 at cutoff $p\le10^8$ rest on the correctness of the truncated sums and the stated tail bounds, which a formal proof of the main theorem does not by itself certify.
- The master product can be differentiated further, so exact higher-order cumulants of the valuations, or joint densities for three or more primes, follow from the same local estimates; nothing in Lemma 2.5 restricts the method to second moments.
Original abstract
For elliptic curves $E/\mathbb{Q}$ in short Weierstrass form \[ E=E(a_4,a_6): y^2=x^3+a_4x+a_6, \] we derive a multivariable Euler product generating function which encodes the Tamagawa product $Tam(E)=\prod_p c_p(E)$. Using this generating function, we compute limiting distributions, exact covariances, and moment and tail bounds for four important statistics on the Tamagawa number; for example we show that more than half of all curves in this height ordering have trivial Tamagawa product, about $42.2\%$ have exactly one prime with nontrivial local Tamagawa number, and only about $6.8\%$ have exactly two such primes. The product is obtained by specializing an Euler product indexed by local reduction data that we derive from Tate's algorithm. The results in this paper were autoformalized in Lean by AxiomProver.
The compiled record
The paper's own sentences, arranged, kept beside the review and separate from it. Each label is drawn from a closed vocabulary; the number beside it is the probability the classifier gave that label. Numbers in the tables are the authors', checked by code. No sentence in this record was written by a model.
The abstract, sentence by sentence
For elliptic curves $E/\mathbb{Q}$ in short Weierstrass form \[ E=E(a_4,a_6): y^2=x^3+a_4x+a_6, \] we derive a multivariable Euler product generating function which encodes the Tamagawa product $Tam(E)=\prod_p c_p(E)$. Using this generating function, we compute limiting distributions, exact covariances, and moment and tail bounds for four important statistics on the Tamagawa number; for example we show that more than half of all curves in this height ordering have trivial Tamagawa product, about $42.2\%$ have exactly one prime with nontrivial local Tamagawa number, and only about $6.8\%$ have exactly two such primes. The product is obtained by specializing an Euler product indexed by local reduction data that we derive from Tate's algorithm. The results in this paper were autoformalized in Lean by AxiomProver.
background contribution method result interpretation main claim
| # | sentence | role | the authors' own work | states the main conclusion | strength |
|---|---|---|---|---|---|
| s0 | For elliptic curves $E/\mathbb{Q}$ in short Weierstrass form \[ E=E(a_4,a_6): y^2=x^3+a_4x+a_6, \] we derive a multivariable Euler product generating function which encodes the Tamagawa product $Tam(E)=\prod_p c_p(E)$. | contribution 0.91 | 0.95 | 0.11 | asserted 1.00 |
| s1 | Using this generating function, we compute limiting distributions, exact covariances, and moment and tail bounds for four important statistics on the Tamagawa number; for example we show that more than half of all curves in this height ordering have trivial Tamagawa product, about $42.2\%$ have exactly one prime with nontrivial local Tamagawa number, and only about $6.8\%$ have exactly two such primes. | result 0.98 | 0.94 | 0.47 | asserted 1.00 |
| s2 | The product is obtained by specializing an Euler product indexed by local reduction data that we derive from Tate's algorithm. | method 0.99 | 0.83 | 0.06 | asserted 1.00 |
| s3 | The results in this paper were autoformalized in Lean by AxiomProver. | method 0.53 | 0.39 | 0.02 | asserted 1.00 |
What was examined, and what was not
- arXiv listing abstract v1
- full text
- figures and tables
- references and citing sentences
- relations to other papers
- P abstract
- S sections
- T body sentences
- C citations
- R cross-paper relations
References
Discussion
Disagree with the review, or know something it missed? Leave a comment, highlight any passage above to comment on it, or challenge this review formally.