The Setting
Let be the subset of the Hilbert space built in the next section, and let be an infinite-dimensional real normed space. The question is how faithfully a finite subset of can be copied into , and whether the copy can be put near a prescribed point of without losing control of its doubling constant.
The Constructed Set
The Hilbert space is the space of real sequences with , normed by .✦ The base plane is the set of pairs of real numbers with the sup distance .✦ The base vector of is the vector of with entry at index , entry at index and all other entries zero, and also stands for it.✦
Levels are indexed by . The scale of level is .✦ Fix a bijection , the pairing bijection. The number of colors of level is and the width factor of level is , where the inverse of the pairing bijection sends to a pair and to the pair . The horizontal strips of level are and the vertical strips are , for ; the strip of index has color ✦, and is the union of the open strips of color of both kinds.✦ For a level and a color , the unit vector is the coordinate vector of at index ; these vectors are orthogonal to each other and to the base plane, and is the level vector of level and color .
An admissible displacement at a point of the base plane is a finitely supported sequence in which each is either or a level vector of a color with .✦ The constructed set is the subset of of all vectors (1)✦ with in the base plane and an admissible displacement at ; it carries the distance of .
Doubling Subsets
For a natural number , a subset of a metric space with distance is -doubling if, for every point of and every radius , there is a set of at most points of such that (2)✦
The Claim
Theorem 1. Let be an infinite-dimensional real normed space.
For every natural number there are Auerbach vectors in : for all real numbers and every index , (3)✦
For every finite subset of there is a linear map such that, for all , (4)✦
Suppose that is -doubling. For every finite subset of , every vector of and every , there are a map whose image is -doubling and a number such that, for all , (5)✦
Let be a natural number, an -doubling subset of a metric space , a metric space, a set, and and two maps such that, for some and all , (6)✦ Then the image is -doubling.
In the paper, the result reads: each finite witness , which plays the part of , is placed in by Dvoretzky’s theorem with distortion less than , so that each image is doubling with constant at most .
What Is Checked
The four parts of the claim are stated over an infinite-dimensional real normed space and, in part 4, over arbitrary metric spaces; over the Hilbert space and its subset of points (1), built from the base plane, the scales , the colors , the width factors , the strips and the unit vectors ; and over the doubling condition (2).