OneStepLaplaceComposer #
Ported from scottnarmstrong/Superdiffusion at 7f20476385d4af418f110303085ad0740c9eba46. Module imports, support namespace and local option directives adapted for Lean Pool.
The contraction factor and the site embedding of the one-step Laplace estimate #
The one-step estimate on a good cube produces, at one centre, a strict
contraction for the negative exponential of the exit time from the enlarged
cube. Its discount rate is (C T(3^n))⁻¹ and its factor is
oneStepLaplaceContraction at the two scale constants of the exit-time
comparison, and neither depends on the centre of the cell, on the direction of
the datum, or on the sample: the constant C of the exit-time comparison is
returned once and for all, before the model, the accuracy, the scale and the
centre are quantified.
The two indexings #
The cells are the closed scale-(n-1) cells z + □_{n-1} centred at the
embedded lattice sites 3^{n-1} z, and the enlargements are the open cubes
z + □_n. The one-step estimate speaks of the pair
(closure (y + □_{n-1}), y + □_n), so the two agree once the site embedding
rescaledLatticePoint n is recognized as rescaleSite (n - 1).
Main results #
oneStepLaplaceContraction_ne_top,oneStepLaplaceContraction_lt_one— the contraction factor is finite and strictly below one.rescaledLatticePoint_eq_rescaleSite,good_goodSiteEnum— the site embedding of the two indexings above, and the goodness of an enumerated site.
References #
- ABK26, the one-step Laplace estimate and the chaining of Section 5.3.
1. The contraction factor #
The contraction factor is finite.
The contraction factor is strictly below one whenever both scale constants are positive.
3. The site embedding #
The embedding of the site lattice used by the good cube events is the
embedding of the scale-(n-1) partition.
Every enumerated site of a centred lattice cube is good.