The Setting
Entropy between two scales records the distribution of mass across an interval of resolutions. For a fair two-point convolution, the distance between the two translations supplies an intermediate scale.
A measure on is compactly supported if there is a number such that .✦ For a real random variable , write for its probability law.
Grid Entropy and Translations
For a measure on a measurable space , a map , and a label , the fiber mass is the real mass of the fiber over , as in (1).✦ The partition entropy of a map with respect to a measure on is the sum of the entropy contributions of its fiber masses , as in (2). (1)✦ (2)✦ All entropies here use natural logarithms; division by expresses them in bits.
The shifted grid cell index at scale and offset assigns the integer in (3). The shifted grid entropy of a measure on is its partition entropy under the shifted grid cell index , and the averaged grid entropy is its Lebesgue average over , as in (3). (3)
The entropy window of a measure on is the difference of its averaged grid entropies at and .✦ The dyadic entropy deficit of a measure on is minus its entropy window from to , as in (4).✦ (4)
The translate of a measure on is its image under .✦ For a measure on , the fair two-point convolution and the equal mixture of translates are the measures given in (5). (5)
The Claim
Theorem 1. Let be a compactly supported probability measure on . For scales , the fair two-point convolution satisfies (6)✦ For and scales , the equal mixture of translates satisfies (7)✦
In the paper, the result reads: for a bounded random variable , an independent fair bit , and , , with the same inequality for convolution by any fair two-point law whose points have distance .
What Is Checked
The inequalities (6) and (7) are stated for compactly supported probability measures on , their averaged shifted-grid entropies, and strict separation of the two endpoint scales from the translation distance.