A choice-free maximum of a finite natural family
DefinitionGreenTaoPort_Wikipedia_SzemeredisTheorem_Hypergraph_TowerDominatingGrowtharithmetic-progressionscombinatoricsgreen-tao-port-8822hypergraph-regularity
For a family of natural numbers indexed by , the finite maximum is defined recursively. The empty family has value zero. At successor size, the value is the maximum of the previous family's value and the final coordinate.
This computes a bound without choosing an index at which it is attained. It provides the finite maximum used to combine rankwise complexity estimates in the later growth-function construction.
Definition code
import Definitions.Def_GreenTaoPort_Wikipedia_SzemeredisTheorem_Hypergraph_GrowthFunctionRegularity
import Lean
import Mathlib.Algebra.BigOperators.Expect
import Mathlib.Algebra.BigOperators.Fin
import Mathlib.Algebra.BigOperators.Ring.Finset
import Mathlib.Algebra.Order.BigOperators.Expect
import Mathlib.Algebra.Order.BigOperators.GroupWithZero.Finset
import Mathlib.Algebra.Order.Chebyshev
import Mathlib.Analysis.LocallyConvex.Separation
import Mathlib.Data.Fin.Tuple.Basic
import Mathlib.Data.Fin.Tuple.Finset
import Mathlib.Data.Finset.Max
import Mathlib.Data.Fintype.Card
import Mathlib.Data.Int.Lemmas
import Mathlib.Data.ZMod.Basic
import Mathlib.Data.ZMod.Units
import Mathlib.Order.Partition.Finpartition
import Mathlib.RingTheory.Int.Basic
import Mathlib.Tactic
import Mathlib.Topology.ContinuousMap.Weierstrass
set_option autoImplicit true
set_option relaxedAutoImplicit false
set_option maxSynthPendingDepth 3
/-!
# Finite-stage tower-dominating growth functions
The triangular regularity schedule at rank `j` has one-step map
```
T_{F,j}(M) =
(2^(j+1))^(orderedRemovalRegularityBudget k j (1 / F(M))) * M.
```
It is circular to ask, without further proof, for one growth function to
dominate the tower map formed from its own reciprocal tolerance. The finite
regularity argument does not need such a fixed point. Instead, this file
constructs a sequence
```
F₀ = F, F_{s+1} = majorant of F_s and every T_{F_s,j}, j < r.
```
The construction first takes bounded maxima in the complexity variable and
then a finite maximum over the relevant ranks. Consequently every `F_s` is
monotone and strictly above the diagonal, while `F_{s+1}` dominates every
one-step map computed with `F_s`. This gives the exact stage-shifted
inequality
```
M_{F_s,j}(n+1) ≤ F_{s+1}(M_{F_s,j}(n)).
```
The last section applies this inequality to every selected rank of
`GrowthFunctionOrderedComplexRegularityCertificate` and embeds each selected
coarse/fine pair in an honest one-step `DescendingGrowthHierarchy`.
-/
namespace Wikipedia.SzemeredisTheorem
/-! ## The one-step map and bounded envelopes -/
/-! ## One majorant stage -/
/-! ## Iterated finite-stage closure -/
/-! ## Finite maxima for simultaneous hierarchy bounds -/
/-- Maximum of a finite family of natural numbers. The recursive
`Fin.lastCases` presentation is convenient for later all-layer bounds and
does not introduce a choice of an optimizing index. -/
def finiteMaximum : (n : ℕ) → (Fin n → ℕ) → ℕ
| 0, _ => 0
| n + 1, value =>
max
(finiteMaximum n (fun i => value i.castSucc))
(value (Fin.last n))
/-! ## Bridge to selected all-rank certificates -/
namespace GrowthFunctionOrderedComplexRegularityCertificate
/-! ### One hierarchy containing all independently selected layers -/
end GrowthFunctionOrderedComplexRegularityCertificate
end Wikipedia.SzemeredisTheorem
/- Materialize compiler-generated declarations promised by this bundle. -/
run_meta do
for requiredName in #[
`Wikipedia.SzemeredisTheorem.finiteMaximum.eq_1,
`Wikipedia.SzemeredisTheorem.finiteMaximum.eq_2,
`Wikipedia.SzemeredisTheorem.finiteMaximum.eq_def,
`Wikipedia.SzemeredisTheorem.finiteMaximum.match_1
] do
let actualName ← Lean.realizeGlobalConstNoOverloadCore requiredName
unless actualName == requiredName do
throwError "Generated companion resolved to an unexpected name"
Source