Documentation

LeanPool.Superdiffusion.Algsuperdiff.Section5.Field.StreamField

StreamField #

Ported from scottnarmstrong/Superdiffusion at 7f20476385d4af418f110303085ad0740c9eba46. Module imports, support namespace and local option directives adapted for Lean Pool.

The full stream matrix #

ABK26's stream matrix is the formal sum k(x) = Σ_{n ∈ ℤ} j_n(x), which does not converge; the paper records that it is the differences k(x) - k(y) that converge, and that only ∇ · k enters the generator, so the field is defined modulo constants. This module builds the normalized representative

k(x) = Σ_{n ∈ ℤ} (j_n(x) - j_n(0)) ,

as an honest tsum over ℤ, on the carrier FullSample where both halves converge locally uniformly: the descending half from the value gauges of the cutoff carrier, the ascending half from the gradient gauges through the mean value inequality.

Main definitions #

Main results #

References #

1. Origin cubes of nonnegative scale exhaust the space #

2. The mean value bound on a cube #

The mean value inequality for one shell on an origin cube, in the ambient norms, with the dimensional cost d of the conversion between the ambient norm of the stored derivative and its exact Euclidean induced norm.

The entrywise form of the mean value bound.

3. Summability of the two halves on the carrier #

The ascending gradient series on the origin cube of scale ℓ is summable with sum at most the carrier's constant, uniformly in the starting index.

The ascending gradient series on □_ℓ sums to at most the sample constant.

theorem Algsuperdiff.Section5.Field.summable_localCubeDerivNorm_ascending {d : ℕ} {gamma : ℝ} (omega : FullSample d gamma) (ell : ℕ) (N : ℤ) :
Summable fun (r : ℕ) => Section3.Provider.Stream.localCubeDerivNorm (↑ell) (↑↑omega (N + ↑r))

The ascending gradient series on the origin cube of scale ℓ is summable from any starting index: shifting the starting index changes the family by finitely many terms.

The descending value series on any origin cube is summable, by the lower-tail condition already carried by the ambient carrier.

4. Summability of the normalized increments #

theorem Algsuperdiff.Section5.Field.summable_streamIncrement_ascending {d : ℕ} {gamma : ℝ} (omega : FullSample d gamma) (N : ℤ) (x : HCPolySupport.Vec d) (i k : Fin d) :
Summable fun (r : ℕ) => (↑(↑↑omega (N + ↑r))).1 x i k - (↑(↑↑omega (N + ↑r))).1 0 i k

The ascending half, from any starting index, is summable.

theorem Algsuperdiff.Section5.Field.summable_streamIncrement_descending {d : ℕ} (omega : Section3.Cutoff.CutoffSample d) (m : ℤ) (x : HCPolySupport.Vec d) (i k : Fin d) :
Summable fun (r : ℕ) => (↑(↑omega (m - ↑r))).1 x i k - (↑(↑omega (m - ↑r))).1 0 i k

The descending half is summable entrywise.

5. The field #

noncomputable def Algsuperdiff.Section5.Field.streamField {d : ℕ} {gamma : ℝ} (omega : FullSample d gamma) (x : HCPolySupport.Vec d) :

The normalized stream matrix k(x) = Σ_{n ∈ ℤ} (j_n(x) - j_n(0)). The sum is over all of ℤ and is genuinely convergent on the carrier; the normalization at the origin is the additive constant the source leaves free.

Equations
Instances For
    @[simp]
    theorem Algsuperdiff.Section5.Field.streamField_apply_entry {d : ℕ} {gamma : ℝ} (omega : FullSample d gamma) (x : HCPolySupport.Vec d) (i k : Fin d) :
    streamField omega x i k = ∑' (n : ℤ), ((↑(↑↑omega n)).1 x i k - (↑(↑↑omega n)).1 0 i k)
    @[simp]
    theorem Algsuperdiff.Section5.Field.streamField_zero {d : ℕ} {gamma : ℝ} (omega : FullSample d gamma) :
    streamField omega 0 = 0

    The normalization: the field vanishes at the origin.

    Antisymmetry: every shell is antisymmetric, hence so is the sum.

    6. The identification with the cutoffs #

    theorem Algsuperdiff.Section5.Field.streamField_eq_cutoff_add_upperTail {d : ℕ} {gamma : ℝ} (omega : FullSample d gamma) (m : ℤ) (x : HCPolySupport.Vec d) (i k : Fin d) :
    streamField omega x i k = (Section3.Cutoff.cutoff m ↑omega).toFun x i k - (Section3.Cutoff.cutoff m ↑omega).toFun 0 i k + ∑' (r : ℕ), ((↑(↑↑omega (m + 1 + ↑r))).1 x i k - (↑(↑↑omega (m + 1 + ↑r))).1 0 i k)

    The cutoff identification. For every scale m, the normalized field is the normalized m-th cutoff plus the tail of the ascending half: k(x) = (k_m(x) - k_m(0)) + Σ_{n > m} (j_n(x) - j_n(0)).

    theorem Algsuperdiff.Section5.Field.streamField_sub_cutoff_sub_eq {d : ℕ} {gamma : ℝ} (omega : FullSample d gamma) (m : ℤ) (x : HCPolySupport.Vec d) (i k : Fin d) :
    streamField omega x i k - (Section3.Cutoff.cutoff m ↑omega).toFun x i k + (Section3.Cutoff.cutoff m ↑omega).toFun 0 i k = ∑' (r : ℕ), ((↑(↑↑omega (m + 1 + ↑r))).1 x i k - (↑(↑↑omega (m + 1 + ↑r))).1 0 i k)

    The identification in the subtractive form of the source: the difference between the field and its m-th cutoff, both normalized at the origin, is the tail Σ_{n > m} (j_n(x) - j_n(0)).