The Setting
Entropy between two scales concerns how a real distribution is resolved as the observation scale changes. Let be a measure on ; positive numbers denote observation scales, and denotes a shift of the grid. The library uses natural logarithms in this entropy calculus, so division by gives entropy in bits.
The Definition
Definition 1. For a measure on a measurable space , a map and a label , the real fiber mass is defined by (1): (1)✦ The Shannon entropy of the label map under is the real sum in (2), with the convention : (2)✦
For real , the grid cell index is the integer-valued map on defined by (3): (3)✦ The shifted grid entropy of a measure on , for real , is the entropy of the grid cell index , as in (4): (4)✦ Averaging the shifted grid entropy of a measure on over a unit interval of shifts gives its averaged grid entropy at a real scale , as in (5): (5)✦
The entropy window of a measure on between real scales is the difference of its averaged grid entropies in (6): (6)✦ The difference between and the entropy window from to is the one-step deficit of a measure on at a real scale , as in (7): (7)✦
A measure on is compactly carried if it gives zero mass to the complement of some bounded interval, as specified in (8): (8)✦
In the paper, the result reads: for a bounded real variable and positive , writing for Shannon entropy in bits, , , and , with differential entropy for a density , and an independent uniform variable on .
What Is Checked
The entropy in (2) is defined for a measure and an arbitrary label map, while the grid constructions are defined for measures on and real scales and shifts, with compact carriage given by (8).