The Setting
The entropy of a law on the line depends on the length and the origin of the intervals used to observe it. Averaged grid entropy measures this distribution of mass after averaging over the origin of the grid.
A measure on is compactly carried if it assigns zero mass to the complement of a bounded closed interval, as in (1): (1)✦
For a measure on a measurable space and a map , the fiber mass is the real mass of ✦; the partition entropy is the sum in (2), with : (2)
For a measure on and real parameters , the shifted grid cell index assigns a point to its cell at scale and offset ; the shifted grid entropy is the partition entropy of this index, as defined in (3): (3)
For a measure on and a real scale , the averaged grid entropy is the mean of over one unit interval of offsets.✦ The entropy window of a measure on at real scales is the difference of its averaged entropies at the two scales, as in (4): (4)
These entropies use natural logarithms. A hat denotes division by , giving entropy in bits. For a random variable, entropy notation refers to its law. For real parameters , write and ; denotes the image measure under .
The Claim
Theorem 1. Let be a compactly carried probability measure on and let . For and , affine covariance is (5): (5)✦
For a positive integer , the integer-ratio window satisfies (6): (6)✦
Let be a finite family of compactly carried probability measures, and let with . For their mixture , finite-mixture concavity is (7): (7)✦
For real translations and the same weights, the translated mixture satisfies (8): (8)✦
For every , the logarithmic scale comparison is (9): (9)✦
When , the averaged and fixed-grid comparisons are (10) and (11), with arbitrary real offsets : (10)✦ (11)✦
For every Borel measure on and real , grid translation is (12): (12)✦
For every real offset , the grid upper deviation is (13): (13)✦
In the paper, the result reads: for every compactly supported real random variable and , with independent of and uniform on and denoting differential entropy in bits, and for ; for with a positive integer, is concave in the law, and for independent bounded ; for , ; and and .
What Is Checked
The claims are stated over the grid and averaged entropies of compactly carried probability measures, their affine images and finite mixtures, with the grid translation identity applying to arbitrary Borel measures and real scale parameters.