Documentation

LeanPool.Superdiffusion.Algsuperdiff.Section4.Provider.ExcessDecay.ForceTransport

ForceTransport #

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

The A4 transport of the fractional (Gagliardo) force seminorm #

The frozen §4.3 anchor writes its force leg as

  [g]_{H^s(W)} = normalizedGagliardoESeminormOn W s g ,
      W = ((z + □_{n+2}) ∩ □_m) ,

while every proved §4.3 estimate lives in the translated frame at the origin cube □_{n+2}. The anchor's z is an arbitrary real point of □_m, so that lemma does not apply.

This module proves the real-translation statement directly. The route is the one the lattice proof uses, carried out on an arbitrary set: the map x ↦ x + z is measure preserving for the restricted Lebesgue measure between A and z + A (CoarseGraining's measurePreserving_addRight_restrict_translateSet), and volume (z + A) = volume A, so it is measure preserving for both slots of normalizedGagliardoMeasureOn — the volume-normalized first slot and the plain restricted second slot. The Gagliardo kernel is translation covariant because dist (x + z) (y + z) = dist x y, so the two eLpNorms agree. No constant and no side condition appear: the identity is exact, and holds for an arbitrary (possibly non-measurable) f, because it is proved through the lintegral form of eLpNorm and a measurable equivalence.

Main results #

References #

1. eLpNorm under a measure-preserving measurable equivalence #

2. The one-slot change of variables #

The same in the frozen statement's (fun y => z + y) '' A spelling.

3. The two-slot change of variables #

The Gagliardo product measure of a translate is the pushforward under the pair translation. Both slots move: the normalized first slot because the normalizing volume is translation invariant, the plain second slot because restricted Lebesgue measure is.

4. The real-translation covariance of the fractional seminorm #

[f]_{H^s(z+A)} = [f(· + z)]_{H^s(A)} for an arbitrary real translation vector z and an arbitrary set A.

This is the statement CoarseGraining does not have: its cubeGagliardoESeminorm lattice translation only moves a cube by an integer multiple of its scale factor. The identity is exact — no constant, no s-restriction, no regularity of f.

The same in the frozen statement's (fun y => z + y) '' A spelling: the anchor's force leg, read at the origin cube.

5. Evenness #

6. The force datum transports #

The L² half of the force datum transports. g ∈ L^p(z+A) in the anchor's normalized measure is g(· + z) ∈ L^p(A).

The fractional half of the force datum transports. The Gagliardo kernel of g is L² for the anchor's window measure exactly when the kernel of the transported force is L² at the origin frame.

7. The open cube #

On the open realization of a triadic cube the volume-normalized measure is CoarseGraining's normalizedCubeMeasure (the boundary is Lebesgue null).

The anchor's fractional seminorm on an open cube is CoarseGraining's cube seminorm. The bridge into the whole cubeGagliardo* A.

The anchor's force leg, transported. The [g]_{H^s} slot of the frozen statement, on the bare translated parent window, equals CoarseGraining's cube seminorm of the transported force at the origin cube.