Wiki
Wiki

Research notes on every problem, and a library of the papers behind them. Built from the open erdos repository.

Updated


Source. Theorem 1.6, pp. 4--5, of Victor Y. Wang, Sums of cubes and the Ratios Conjectures, arXiv:2108.03398v2 (19 April 2023), the edition named on the source card. The hypotheses are Conjectures 1.2 (p. 3), 1.4 and 1.5 (p. 4) and 1.8 (p. 5).

Read depth. Claims checked: the statement, the definitions (1.2)--(1.5) and (1.11), and the four conjectures it assumes were read clause by clause on pp. 2--5; the proof on p. 57 was read for its structure. Nothing here is independently reviewed.

Statement

Fix a cubic form F∈Z[x1,…,x6]F\in\mathbb Z[x_1,\ldots,x_6] with nonzero discriminant, let VV be the hypersurface F=0F=0 in PQ5\mathbb P^5_{\mathbb Q}, and let Υ\Upsilon be the set of 33-dimensional subspaces L⊆Q6L\subseteq\mathbb Q^6 on which FF vanishes (p. 2). For w∈Cc∞(R6)w\in C_c^\infty(\mathbb R^6) and real X≥1X\ge1 put

NF,w(X)=∑x∈Z6: F(x)=0w(x/X),EF,w(X)=NF,w(X)−SF σ∞,F,w X3−∑L∈Υ ∑x∈L∩Z6w(x/X),N_{F,w}(X)=\sum_{\mathbf x\in\mathbb Z^6:\,F(\mathbf x)=0}w(\mathbf x/X),\qquad E_{F,w}(X)=N_{F,w}(X)-\mathfrak S_F\,\sigma_{\infty,F,w}\,X^3-\sum_{L\in\Upsilon}\ \sum_{\mathbf x\in L\cap\mathbb Z^6}w(\mathbf x/X),

with SF\mathfrak S_F the singular series and σ∞,F,w\sigma_{\infty,F,w} the real density of (1.4) (displays (1.2)--(1.4), pp. 2--3). The asymptotic (1.5) is lim⁡X→∞X−3EF,w(X)=0\lim_{X\to\infty}X^{-3}E_{F,w}(X)=0.

Theorem 1.6 (pp. 4--5). Let m=6m=6 and let FF be diagonal. Assume Conjectures 1.2, 1.4, 1.5 and 1.8. Then (1.5) holds for every w∈Cc∞(Rm)w\in C_c^\infty(\mathbb R^m) whose support satisfies

{x∈Rm:w(x)≠0}‾⊆{x∈Rm:x1⋯xm≠0}\overline{\{\mathbf x\in\mathbb R^m: w(\mathbf x)\ne0\}}\subseteq\{\mathbf x\in\mathbb R^m: x_1\cdots x_m\ne0\}

(display (1.11)). Hence the Hasse principle holds for VV. Moreover, if F=x13+⋯+x63F=x_1^3+\cdots+x_6^3, then 100% of the integers a≢±4(mod9)a\not\equiv\pm4\pmod 9 lie in F0(Z3)F_0(\mathbb Z^3), where F0(x,y,z)=x3+y3+z3F_0(x,y,z)=x^3+y^3+z^3.

Conjectures 1.2, 1.4 and 1.5 are stated on the Theorem 1.3 page. Conjecture 1.8 (RA1oo, p. 5), for even mm and assuming Conjecture 1.2, is a first-moment prediction: for M≥1M\ge1, a modulus n0∈[1,M]n_0\in[1,M] and a,b∈Zm∩[−M,M]m\mathbf a,\mathbf b\in\mathbb Z^m\cap[-M,M]^m, the sum of Φc,1(s)\Phi^{\mathbf c,1}(s) over c∈S1\mathbf c\in\mathcal S_1 in the dilated box Z⋅BM(b)Z\cdot\mathcal B_M(\mathbf b) of (1.12) with c≡a mod n0\mathbf c\equiv\mathbf a\bmod n_0 equals the sum over the same c\mathbf c of (1+oF,M;Z→∞(1))AF,1a,n0(s)(1+o_{F,M;Z\to\infty}(1))A^{\mathbf a,n_0}_{F,1}(s), at s=σ(Z)+its=\sigma(Z)+it with σ(Z)=1/2+1/log⁡Z\sigma(Z)=1/2+1/\log Z and t∈[−log⁡Z,log⁡Z]t\in[-\log Z,\log Z], for Z≥2Z\ge2 (display (1.14)); here AF,1a,n0(s)A^{\mathbf a,n_0}_{F,1}(s) is the Euler product of §6.3.1, absolutely convergent in Re⁡(s)>1/3\operatorname{Re}(s)>1/3. All four conjectures are unproved, so every conclusion is conditional.

The third conclusion concerns integer cubes of either sign. It says nothing about sums of three nonnegative cubes, and it gives no representation of any particular integer.

Proof pointer

§10.3, p. 57. Unconditionally, (1.3), (2.10) and Theorem 2.5 give EF,w(X)/X3=O(X2.75+ϵ)/X3+Σ♮(X,S1)E_{F,w}(X)/X^3=O(X^{2.75+\epsilon})/X^3+\Sigma^\natural(X,\mathcal S_1) (display (10.30)). Under (1.11), Theorem 10.7 (p. 56) gives Σ♮(X,S1)=oX→∞(X(6−m)/4)\Sigma^\natural(X,\mathcal S_1)=o_{X\to\infty}(X^{(6-m)/4}), its moment hypotheses supplied by Propositions 9.7 and 9.9; with m=6m=6 this is (1.5). The Hasse principle follows by choosing ww with σ∞,F,w>0\sigma_{\infty,F,w}>0. The density statement for F0(Z3)F_0(\mathbb Z^3) is deduced from (1.5) through Wang's earlier work (Theorem 1.1 of the paper cited as [Wan23c], or Theorem 2.1.8 of [Wan22]), not proved here.

Dependencies

Conjectures 1.2, 1.4, 1.5 and 1.8 as hypotheses; within the paper, Theorem 2.5, Propositions 9.7 and 9.9, and Theorem 10.7; outside it, the deduction cited above from [Wan23c] or [Wan22].

Bears on

  • Problem 940: no direct bearing. The 100% statement counts representations a=x3+y3+z3a=x^3+y^3+z^3 with x,y,z∈Zx,y,z\in\mathbb Z of either sign, while the problem's sums use positive rr-powerful numbers; the conditional relation at r=3r=3 comes from Theorem 1.3.