Projected majorant moments and truncation at one
DefinitionGreenTaoPort_Wikipedia_GreenTao_Transference_Densificationarithmetic-progressionsgreen-tao-port-8822number-theory
On a finite type , the projected-moment hypothesis for a real majorant and error records , pointwise nonnegativity of , and
The pointwise truncation of a real function is , and its excess above one is . Truncation alone does not impose a lower bound on an arbitrary input; with a nonnegative input it gives a weight in .
Definition code
import Definitions.Def_GreenTaoPort_Wikipedia_SzemeredisTheorem_Finite_Mean
import Lean
import Mathlib.Algebra.BigOperators.Expect
import Mathlib.Data.Fintype.Card
import Mathlib.Tactic
set_option autoImplicit true
set_option relaxedAutoImplicit false
set_option maxSynthPendingDepth 3
/-!
# Finite analytic lemmas for densification
This file isolates the elementary projected-majorant estimates used in one
step of Conlon--Fox--Zhao densification. It does not state a relative
counting theorem.
If a nonnegative finite majorant `ν` has first and second normalized moments
within `η` of one, then
```
mean (fun x => |ν x - 1|) ≤ √(3 * η).
```
For `0 ≤ g ≤ ν`, truncating `g` pointwise at one produces a `[0,1]`-valued
function. Its `L¹` loss, and its loss in any pairing against a unit-bounded
factor, are bounded first by the mean excess `max (ν - 1) 0`, and hence by
the same square-root moment bound. The final section records pointwise and
averaged product-perturbation inequalities used to package the remaining
bounded factors in a densification step.
-/
namespace Wikipedia.SzemeredisTheorem
open scoped BigOperators
/-! ## Projected-majorant moments -/
/-- The finite moment hypotheses used after projecting all majorant factors
other than the edge currently being densified.
The common error parameter is recorded as nonnegative so downstream
square-root statements do not have to recover this fact from an absolute
value inequality. -/
structure HasProjectedMajorantMoments
{Ω : Type*} [Fintype Ω]
(ν : Ω → ℝ) (η : ℝ) : Prop where
error_nonneg : 0 ≤ η
nonneg : ∀ x, 0 ≤ ν x
firstMoment_close : |mean ν - 1| ≤ η
secondMoment_close :
|mean (fun x => ν x ^ 2) - 1| ≤ η
namespace HasProjectedMajorantMoments
end HasProjectedMajorantMoments
/-! ## Truncation at one -/
/-- Pointwise truncation used to turn a nonnegative function into a dense
`[0,1]` model. -/
def truncateAtOne {Ω : Type*} (g : Ω → ℝ) : Ω → ℝ :=
fun x => min (g x) 1
/-- Pointwise mass above one. -/
def excessAboveOne {Ω : Type*} (f : Ω → ℝ) : Ω → ℝ :=
fun x => max (f x - 1) 0
namespace HasProjectedMajorantMoments
end HasProjectedMajorantMoments
/-! ## Pairing and product perturbations -/
namespace HasProjectedMajorantMoments
end HasProjectedMajorantMoments
namespace HasProjectedMajorantMoments
end HasProjectedMajorantMoments
end Wikipedia.SzemeredisTheorem
/- Materialize compiler-generated declarations promised by this bundle. -/
run_meta do
for requiredName in #[
`Wikipedia.SzemeredisTheorem.excessAboveOne.eq_1,
`Wikipedia.SzemeredisTheorem.truncateAtOne.eq_1
] do
let actualName ← Lean.realizeGlobalConstNoOverloadCore requiredName
unless actualName == requiredName do
throwError "Generated companion resolved to an unexpected name"
Source