Critical logarithmic weight on the four-torus #
The binary convolution of two inverse-square kernels has a logarithmic singularity. This file packages the elementary pointwise inequalities needed to average that logarithm at a moving mollifier scale.
The nonnegative critical weight produced by the sharp binary
inverse-square convolution. Mathlib uses Real.log 0 = 0; this harmless
value agrees with all a.e. applications below.
Instances For
The half-power kernel |z|⁻¹, expressed intrinsically through the
paper's inverse-square model.
Equations
- Anderson4D.invSqKerHalf z = Anderson4D.invSqKer z ^ (1 / 2)
Instances For
The inverse norm is controlled by twice the intrinsic half-power
kernel. The factor two is exactly the comparison
torusDistSq z ≤ 4 ‖z‖².
Weighted Young inequality at the exponents used in the local logarithmic convolution.
If a torus point stays at least a fixed multiple of ε away from
the logarithmic singularity, its critical weight costs only one copy of
|log ε|. The constant depends explicitly on that fixed multiple.
The critical logarithm has the expected r⁴(1+|log r|) mass on
a four-dimensional torus ball, uniformly in its centre.