The Setting
Take a real scale and a point . Weighting a variance- Gaussian density by produces an odd window at scale along with its ratio to a rescaled Gaussian, while integrating a signed kernel over squared scales against produces a smooth kernel. The problem is to identify the triple convolution of the odd window with the signed kernel and to evaluate the smooth kernel away from via a smooth kernel profile.
Gaussian Density
For real variance , the variance- Gaussian density is the real-valued function given for by (1): (1)✦
The Claim
Definition 1. For a real scale , the odd window is the function defined for by (2): (2)✦ At scale , the odd window ratio is the real-valued function specified at by (3): (3)✦ Given , the signed kernel is the function on determined by (4) for : (4)✦ If and , then the odd window and the signed kernel satisfy the triple-convolution identity (5): (5)✦ Over arbitrary squared scale endpoints , the smooth kernel is the function given for by (6): (6)✦ At linear scale endpoints , the smooth kernel is the function on given for by . The heat kernel constant is the real scalar defined by (7): (7)✦ Using , the smooth kernel profile is the real-valued function given for by (8): (8)✦ Whenever satisfy and , the smooth kernel evaluates in terms of as (9): (9)✦
In the paper, the result reads: for and , , , and , where stands for .
What Is Checked
The claim is stated over real scale parameters and squared scale endpoints on the real line, defining from the Gaussian density the odd window , the odd window ratio , the signed kernel , the smooth kernel , the heat kernel constant , and the smooth kernel profile , with and arbitrary real shifts for the triple-window integral and with for the profile identity.