8.3 Convex combinations (凸组合)

本文为《Linear algebra and its applications》的读书笔记

Convex combinations

convex hull: 凸包

The convex hull of a single point ${\mathbf{v}}_1$ is just the set { v 1 } \{\boldsymbol v_1\} {v1​}, the same as the affine hull. In other cases, the convex hull is properly contained in the affine hull. Recall that the affine hull of distinct points ${\mathbf{v}}_1$ and ${\mathbf{v}}_2$ is the line

Because the weights in a convex combination are nonnegative, the points in c o n v { v 1 , v 2 } conv\{\boldsymbol v_1,\boldsymbol v_2\} conv{v1​,v2​} is the line segment between ${\mathbf{v}}_1$ and ${\mathbf{v}}_2$, hereafter denoted by $\overline{{\mathbf{v}}_1{\mathbf{v}}_2}$.

If a set $S$ is affinely independent and if $\mathbf{p}\in affS$, then $\mathbf{p}\in convS$ if and only if the barycentric coordinates of $\mathbf{p}$ are nonnegative.

Example 1 shows a special situation in which $S$ is much more than just affinely independent.

EXAMPLE 1
Let

and S = { v 1 , v 2 , v 3 } S=\{\boldsymbol v_1,\boldsymbol v_2,\boldsymbol v_3\} S={v1​,v2​,v3​}. Note that $S$ is an orthogonal set. Determine whether ${\mathbf{p}}_1$ is in $SpanS$, $affS$, and $convS$.
SOLUTION
Let $W$ be the subspace spanned by $S$.

The orthogonal projection of ${\mathbf{p}}_1$ onto $W$ is ${\mathbf{p}}_1$ itself: This shows that ${\mathbf{p}}_1$ is in $SpanS$. Also, since the coefficients sum to 1, ${\mathbf{p}}_1$ is in $affS$. In fact, ${\mathbf{p}}_1$ is in $convS$, because the coefficients are also nonnegative.

---

Recall that a set $S$ is affine if it contains all lines determined by pairs of points in $S$. When attention is restricted to convex combinations, the appropriate condition involves line segments rather than lines.

Intuitively, a set $S$ is convex if every two points in the set can “see” each other without the line of sight leaving the set. Figure 1 illustrates this idea.

PROOF
The argument is similar to the proof of Theorem 2.

---

The intersection of any collection of subspaces is itself a subspace. A similar result holds for affine sets and convex sets.

PROOF
Let $T$ denote the intersection of all the convex sets containing $S$. Since $convS$ is a convex set containing $S$, it follows that $T\subset convS$. On the other hand, let $C$ be any convex set containing $S$. Then $C$ contains every convex combination of points of $C$, and hence also contains every convex combination of points of the subset $S$. That is, $convS\subset C$. Since this is true for every convex set $C$ containing $S$, it is also true for the intersection of them all. That is, $convS\subset T$ .

Theorem 9 shows that $convS$ is in a natural sense the “smallest” convex set containing $S$. For example, consider a set $S$ that lies inside some large rectangle in ${\mathbb{R}}^2$, and imagine stretching a rubber band around the outside of $S$. As the rubber band contracts around $S$, it outlines the boundary of the convex hull of $S$. Or to use another analogy, the convex hull of $S$ fills in all the holes in the inside of $S$ and fills out all the dents in the boundary of $S$.

EXAMPLE 2
a. The convex hulls of sets $S$ and $T$ in ${\mathbb{R}}^2$ are shown below.

b. Let $S$ be the set consisting of the standard basis for ${\mathbb{R}}^3$, S = { e 1 , e 2 , e 3 } S=\{\boldsymbol e_1,\boldsymbol e_2,\boldsymbol e_3\} S={e1​,e2​,e3​}. Then $convS$ is a triangular surface in ${\mathbb{R}}^3$, with vertices ${\mathbf{e}}_1$, ${\mathbf{e}}_2$, and ${\mathbf{e}}_3$. See Figure 2.

EXAMPLE 3
Let

Show that the convex hull of $S$ is

See Figure 3.

SOLUTION
Consider any point $\mathbf{p}$ in the shaded region of Figure 3, say

The line through $0$ and $\mathbf{p}$ has the equation $y=(b/a)t$ for $t$ real. That line intersects $S$ where $t$ satisfies $(b/a)t=t^2$, that is, when $t=b/a$. Thus, $\mathbf{p}$ is on the line segment from $0$ to $\left[\begin{array}{c}b/a\\ b^2/a^2\end{array}\right]$, which shows that Figure 3 is correct.

---

The following theorem is basic in the study of convex sets. It was first proved by Constantin Caratheodory in 1907. If $\mathbf{p}$ is in the convex hull of $S$, then, by definition, $\mathbf{p}$ must be a convex combination of points of $S$. But the definition makes no stipulation(规定) as to how many points of $S$ are required to make the combination. Caratheodory’s remarkable theorem says that in an $n$-dimensional space, the number of points of $S$ in the convex combination never has to be more than $n+1$.

PROOF
Given $\mathbf{p}$ in $convS$, one may write $\mathbf{p}=c_1{\mathbf{v}}_1+...+c_k{\mathbf{v}}_k$, where ${\mathbf{v}}_i\in S$, $c_1+...+c_k=1$, and $c_i\ge 0$, for some $k$ and $i=1,...,k$. The goal is to show that such an expression exists for $\mathbf{p}$ with $k\le n+1$.

If $k>n+1$, then { v 1 , . . . , v k } \{\boldsymbol v_1,...,\boldsymbol v_k\} {v1​,...,vk​} is affinely dependent, by Exercise 12 in Section 8.2. Thus there exist scalars $d_1,...,d_k$, not all zero, such that

Consider the two equations

and

By subtracting an appropriate multiple of the second equation from the first, we now eliminate one of the ${\mathbf{v}}_i$ terms and obtain a convex combination of fewer than $k$ elements of $S$ that is equal to $\mathbf{p}$.

Since not all of the $d_i$ coefficients are zero, we may assume (by reordering subscripts if if necessary) that $d_k>0$ and that $c_k/d_k\le c_i/d_i$ for all those $i$ for which $d_i>0$. For $i=1,...,k$, let $b_i=c_i-(c_k/d_k)d_i$ . Then $b_k=0$ and

Furthermore, each $b_i\ge 0$. Indeed, if $d_i\le 0$, then $b_i\ge c_i\ge 0$. If $d_i>0$, then $b_i=d_i(c_i/d_i-c_k/d_k)\ge 0$. By construction,

Thus $\mathbf{p}$ is now a convex combination of $k-1$ of the points ${\mathbf{v}}_1,...,{\mathbf{v}}_k$. This process may be repeated until $\mathbf{p}$ is expressed as a convex combination of at most $n+1$ of the points of $S$.

The following example illustrates the calculations in the proof above.

EXAMPLE 4
Let

and S = { v 1 , v 2 , v 3 , v 4 } S=\{\boldsymbol v_1,\boldsymbol v_2,\boldsymbol v_3,\boldsymbol v_4\} S={v1​,v2​,v3​,v4​}. Then

Use the procedure in the proof of Caratheodory’s Theorem to express $\mathbf{p}$ as a convex combination of three points of $S$.
SOLUTION
The set $S$ is affinely dependent. Use the technique of Section 8.2 to obtain an affine dependence relation

Next, choose the points ${\mathbf{v}}_2$ and ${\mathbf{v}}_4$ in (3), whose coefficients are positive. For each point, compute the ratio of the coefficients in equations (2) and (3). The ratio for ${\mathbf{v}}_2$ is $1/24$, and that for ${\mathbf{v}}_4$ is $1/48$. The ratio for ${\mathbf{v}}_4$ is smaller, so subtract $1/48$ times equation (3) from equation (2) to eliminate ${\mathbf{v}}_4$:

This result cannot, in general, be improved by decreasing the required number of points. Indeed, given any three non-collinear points in ${\mathbb{R}}^2$, the centroid(质心) of the triangle formed by them is in the convex hull of all three, but is not in the convex hull of any two.