Lean, informalized

Nielsen & Chuang · Chapter 12, Quantum information theory · p. 574

Exercise 12.18 - Majorization of real vectors through their largest subset sums, and the convexity of the set it cuts out

Vectors majorized by one vector form a convex set.

Hover over a blue or green term in the text, or a node in the graph, to find it in the other. Select it to read its declaration both in Lean and in words.

The Exercise

Fix a real vector y→\vec y. The exercise asks to show that the set of all vectors majorized by y→\vec y is convex.✦

Majorization

Let II be a finite set. A real vector x→\vec x on II assigns a real number xix_i to each index i∈Ii \in I.

The largest kk-sum Lk(x→)L_k(\vec x) of a real vector x→\vec x on II is the largest sum of kk of its entries, Lk(x→)=maxS⊆I,|S|=k∑i∈Sxi,L_k(\vec x) = \max_{S \subseteq I,\ |S| = k} \; \sum_{i \in S} x_i ,(1)✦ the maximum running over the subsets SS of II with kk elements; it is taken to be 00 when kk exceeds the number of elements of II, since then there is no such subset.

A real vector x→\vec x on II is majorized by a real vector y→\vec y on II, written x→≺y→\vec x \prec \vec y, when the two have the same total and no kk entries of x→\vec x sum to more than the largest kk-sum of y→\vec y: ∑i∈Ixi=∑i∈Iyiand∑i∈Sxi≤L|S|(y→) for every subset S⊆I.\sum_{i \in I} x_i = \sum_{i \in I} y_i \quad\text{and}\quad \sum_{i \in S} x_i \le L_{|S|}(\vec y) \ \text{ for every subset } S \subseteq I .(2)✦

The Claim

Theorem 1. Let y→\vec y be a real vector on a finite set II, let x→\vec x and x→′\vec x\,' be real vectors on II with x→≺y→\vec x \prec \vec y and x→′≺y→\vec x\,' \prec \vec y, and let 0≤p≤10 \le p \le 1. Then px→+(1−p)x→′≺y→.p\,\vec x + (1-p)\,\vec x\,' \prec \vec y .(3)✦

What Is Checked

The claim is stated over majorization (2) of real vectors on an arbitrary finite index set, and through it over the largest kk-sum (1). The one property of these constructions the library has is the claim itself, (3), for every vector y→\vec y and every finite set II.