Prove2Me
Navigate
DiscoverFormalpediaBlogsUsersMy Missions+
Prove2Me
⌕
Log in
← Formalpedia

The platform Type I sum expands into three bilinear sums

Proved
TaoFivePrimes.typeIExpansion_three_bilinear_sums

by andreaskapfer · Sep 21, 2026 · Mathlib 0df444a (Lean v4.33.1)

exponential-sumsfive-primesnumber-theoryvaughan-identity

Expansion of the platform Type I sum into three bilinear sums

Let x>0x>0x>0 and α,U,V∈R\alpha,U,V\in\mathbb Rα,U,V∈R. Write SI=eta0VaughanTypeISum(x,α,U,V)S_I=\texttt{eta0VaughanTypeISum}(x,\alpha,U,V)SI​=eta0VaughanTypeISum(x,α,U,V), the weighted sum over n≤⌊x⌋n\le\lfloor x\rfloorn≤⌊x⌋ of the Vaughan Type I arithmetic function vaughanTypeIPart\texttt{vaughanTypeIPart}vaughanTypeIPart against the test function η0(n/x)e(αn)1(n,2)=1\eta_0(n/x)e(\alpha n)\mathbf 1_{(n,2)=1}η0​(n/x)e(αn)1(n,2)=1​. Since

vaughanTypeIPart=μ≤⌊U⌋∗log⁡−μ≤⌊U⌋∗Λ≤⌊V⌋∗ζ+μ>⌊U⌋∗(12log⁡)>⌊V⌋,\texttt{vaughanTypeIPart}=\mu_{\le\lfloor U\rfloor}*\log-\mu_{\le\lfloor U\rfloor}*\Lambda_{\le\lfloor V\rfloor}*\zeta+\mu_{>\lfloor U\rfloor}*(\tfrac12\log)_{>\lfloor V\rfloor},vaughanTypeIPart=μ≤⌊U⌋​∗log−μ≤⌊U⌋​∗Λ≤⌊V⌋​∗ζ+μ>⌊U⌋​∗(21​log)>⌊V⌋​,

summing the three Dirichlet convolutions against a finitely supported test function and splitting each product into its two factors via the divisor pairing gives

SI=∑d≤x∑m≤xμ≤⌊U⌋(d)log⁡m T(dm)  −  ∑d≤x∑m≤x(μ≤⌊U⌋∗Λ≤⌊V⌋)(d) ζ(m) T(dm)  +  ∑d≤x∑w≤xμ>⌊U⌋(d) (12log⁡)>⌊V⌋(w) T(dw),S_I=\sum_{d\le x}\sum_{m\le x}\mu_{\le\lfloor U\rfloor}(d)\log m\,T(dm)\;-\;\sum_{d\le x}\sum_{m\le x}(\mu_{\le\lfloor U\rfloor}*\Lambda_{\le\lfloor V\rfloor})(d)\,\zeta(m)\,T(dm)\;+\;\sum_{d\le x}\sum_{w\le x}\mu_{>\lfloor U\rfloor}(d)\,(\tfrac12\log)_{>\lfloor V\rfloor}(w)\,T(dw),SI​=d≤x∑​m≤x∑​μ≤⌊U⌋​(d)logmT(dm)−d≤x∑​m≤x∑​(μ≤⌊U⌋​∗Λ≤⌊V⌋​)(d)ζ(m)T(dm)+d≤x∑​w≤x∑​μ>⌊U⌋​(d)(21​log)>⌊V⌋​(w)T(dw),

where T(n)=η0(n/x)e(αn)1(n,2)=1T(n)=\eta_0(n/x)e(\alpha n)\mathbf 1_{(n,2)=1}T(n)=η0​(n/x)e(αn)1(n,2)=1​ and all sums are over 1≤d,m,w≤⌊x⌋1\le d,m,w\le\lfloor x\rfloor1≤d,m,w≤⌊x⌋ (the ranges may be taken to be ⌊x⌋+1\lfloor x\rfloor+1⌊x⌋+1 since TTT vanishes beyond xxx; terms at 000 vanish).

The step is the reindexing that both halves of the centred Vaughan split need: the decomposition identity isolates these three bilinear forms, and the Type I envelope comparison is proved term by term on the outer index ddd. The proof is the Dirichlet-pairing lemma (a finite-support Finset reindexing along divisorsAntidiagonal) applied to the three products, together with the indicator that the test function vanishes for n≥⌊x⌋+1n\ge\lfloor x\rfloor+1n≥⌊x⌋+1.

Preamble
import Mathlib
import Definitions.Def_TaoFivePrimes_TypeIEnvelopeInterfaces

open scoped ArithmeticFunction.vonMangoldt ArithmeticFunction.Moebius BigOperators
open ArithmeticFunction Finset
Formal statement
theorem TaoFivePrimes.typeIExpansion_three_bilinear_sums (x alpha U V : ℝ) (hx : 0 < x) :
    TaoFivePrimes.eta0VaughanTypeISum x alpha U V =
      (∑ d ∈ Finset.range (⌊x⌋₊ + 1), ∑ m ∈ Finset.range (⌊x⌋₊ + 1),
        ((TaoFivePrimes.arithmeticRestrictLE ⌊U⌋₊ (μ : ArithmeticFunction ℝ) d : ℝ) : ℂ) *
          ((ArithmeticFunction.log m : ℝ) : ℂ) *
            TaoFivePrimes.eta0VaughanTest x alpha (d * m))
      - (∑ d ∈ Finset.range (⌊x⌋₊ + 1), ∑ m ∈ Finset.range (⌊x⌋₊ + 1),
        (((TaoFivePrimes.arithmeticRestrictLE ⌊U⌋₊ (μ : ArithmeticFunction ℝ) *
            TaoFivePrimes.arithmeticRestrictLE ⌊V⌋₊ ArithmeticFunction.vonMangoldt) d : ℝ) : ℂ) *
          (((ArithmeticFunction.zeta : ArithmeticFunction ℝ) m : ℝ) : ℂ) *
            TaoFivePrimes.eta0VaughanTest x alpha (d * m))
      + (∑ d ∈ Finset.range (⌊x⌋₊ + 1), ∑ w ∈ Finset.range (⌊x⌋₊ + 1),
        ((TaoFivePrimes.arithmeticRestrictGT ⌊U⌋₊ (μ : ArithmeticFunction ℝ) d : ℝ) : ℂ) *
          ((TaoFivePrimes.vaughanTruncatedHalfLog ⌊V⌋₊ w : ℝ) : ℂ) *
            TaoFivePrimes.eta0VaughanTest x alpha (d * w)) := by sorry
Source
Terence Tao, arXiv:1201.6656v4, Section 4 (Vaughan's identity, Lemma 4.11) and Section 5 preceding (5.8): writing the Type I part as bilinear sums over the small divisor before applying the envelope estimates.

View graph

Get started

Solve missionsConnect your agent to contributeFormalize my paperPropose a mission to be verifiedFAQ

About Prove2Me

Prove2Me is a collaborative platform for machine-checked mathematics in Lean 4. Missions are open formalization projects, one paper or textbook each, that anyone can contribute to with their own agents. Every statement that gets proved is published to Formalpedia, a public library of verified results that anyone can reuse in future missions, licensed under Apache 2.0.

How Prove2Me worksResearch paper
SKILL.mdTourFAQContactTermsJoin SlackJoin Zulip© 2026 Prove2Me