The Setting
For , the probabilistic construction for dimension reduction asks for a random vector in a normalized-gradient subspace with prescribed scalar marginals. The scalar variable is denoted by , and the gradient space by , where is a finite coordinate set.
The characteristic function of a probability measure on is the function defined by (1): (1) For a real exponent , the stable kernel is the function defined by (2) for all real , and the stable scale constant is defined by (3). (2)✦ (3)✦
For finite index sets and and a real linear subspace , a coordinatewise -normalized family in is a family of vectors in satisfying (4): (4) Write for the product probability measure on with each factor equal to . For a map and a vector , write for its th coordinate.
The Claim
Theorem 1. Let . There is a probability measure on with characteristic function (5), and the stable scale constant satisfies (6): (5)✦ (6)✦
Let and , and let be a probability measure on with for every . For finite sets , a real linear subspace , and a coordinatewise -normalized family in , there exists a map satisfying (7): (7)✦
Let and , and let be a probability measure on such that for every real . If and , there is a constant for which (8) holds: (8)✦ For every , the upper tail and capped second moment satisfy (9) and (10): (9)✦ (10)✦
In the paper, the result reads: the symmetric intensity construction gives a -valued random vector whose coordinates have characteristic function for every real and satisfy for , with constants .
What Is Checked
The scalar claims are stated over probability measures on , and the joint claim is stated over finite families in a real linear subspace with the normalization (4).