Solution: Proof of Convex Hull's Convexity and Intersection Property

  • Thread starter Thread starter Chris L T521
  • Start date Start date
Click For Summary
SUMMARY

The convex hull of a subset \( A \) in a vector space \( V \) is defined as the set of all convex combinations of vectors from \( A \). This solution proves that the convex hull is indeed convex and demonstrates that it serves as the intersection of all convex sets containing \( A \). The proof utilizes the properties of convex combinations, specifically the constraints \( t_i \in [0,1] \) and \( \sum_{i=1}^n t_i = 1 \), to establish these characteristics definitively.

PREREQUISITES
  • Understanding of vector spaces and their properties
  • Familiarity with convex combinations and their mathematical definitions
  • Knowledge of convex sets and their intersection properties
  • Basic proficiency in mathematical proofs and logic
NEXT STEPS
  • Study the properties of convex sets in detail
  • Explore the concept of convex combinations in various vector spaces
  • Learn about the applications of convex hulls in computational geometry
  • Investigate the relationship between convex hulls and linear programming
USEFUL FOR

Mathematicians, computer scientists, and students studying geometry or optimization who are interested in the properties of convex sets and their applications in various fields.

Chris L T521
Gold Member
MHB
Messages
913
Reaction score
0
Here's this week's problem.

-----

Problem: The convex hull of a subset $A$ in a vector space $V$ is the set of all convex combinations of vectors from $A$, that is,\[t_1x_1+t_2x_2+\cdots+t_nx_n,\]
where $t_i\in[0,1]$ and $\sum_{i=1}^n t_i=1$, $x_1,x_2,\ldots,x_n\in A$ and $n\in\mathbb{N}$ any natural number. Prove that the convex hull of $A$ is convex and that it is the intersection of all convex sets that contain $A$.

-----

 
Physics news on Phys.org
No one answered this week's question. Here's my solution:

Denote the convex hull of $A$ by\[C=\left\{\lambda_1x_1+\lambda_2x_2+\cdots+\lambda_nx_n : \lambda_i\in[0,1],\,\sum_{i=1}^n \lambda_i=1.\right\}\]
We first show that $C$ is convex. Let $\sum_i\alpha_ix_i,\sum_i \beta_ix_i\in C$. Consider the combination
\[t\left(\sum_i \alpha_ix_i\right) + (1-t)\left(\sum_i \beta_ix_i\right) = \sum_i x_i(t\alpha_i + (1-t)\beta_i)\]
for $t\in[0,1]$. Note that for all $i$,
\[t\alpha_i + (1-t)\beta_i\geq 0\]
since $\alpha_i,\beta_i\in[0,1]$. Furthermore,
\[\sum_i t\alpha_i + (1-t)\beta_i = t\left(\sum_i \alpha_i\right) + (1-t)\left(\sum_i \beta_i\right) = t(1)+(1-t)(1) = 1\]
Thus, $\sum_i x_i(t\alpha_i + (1-t)\beta_i)\in C$. Therefore $C$ is convex.Now, assume that $S$ is the intersection of all convex sets containing $x_1,\ldots,x_i$. We just showed that $C$ is convex, so we must have $S\subseteq C$. To show $C\subseteq S$, we proceed by induction on $i$. The base is trivial. Suppose that convex combinations of $n-1$ of the $x_i$'s are in $S$. WLOG, let $y=t_1x_1+\ldots + t_{n-1}x_{n-1}+t_nx_n$ for $x_i\in A$. If one of the $t_i=0$, then we are done by the inductive hypothesis. Otherwise, we note that
\[y=(1-t_n)\left( \frac{1}{1-t_n} (t_1x_1+\ldots + t_{n-1}x_{n-1})\right) + t_nx_n.\]
Since $\sum_{i=1}^n t_i=1$, we know that
\[1-t_n= t_1+\ldots +t_{n-1};\]
thus, the equation for $y$ reduced to $y=(1-t_n)z+t_nx_n$ where $z$ is a convex combination of $n-1$ of the $x_i$'s. This now reduces to the case where $n=2$, implying that $(1-t_n)z+t_n x_n$ is convex. Thus, $C\subseteq S$ which now implies that the convex hull is the intersection of all convex sets that contain $A$.$\hspace{.25in}\blacksquare$
 

Similar threads

  • · Replies 4 ·
Replies
4
Views
2K
  • · Replies 1 ·
Replies
1
Views
2K
  • · Replies 1 ·
Replies
1
Views
3K
  • · Replies 1 ·
Replies
1
Views
1K
  • · Replies 8 ·
Replies
8
Views
2K
  • · Replies 24 ·
Replies
24
Views
4K
  • · Replies 4 ·
Replies
4
Views
4K
  • · Replies 3 ·
Replies
3
Views
2K
  • · Replies 5 ·
Replies
5
Views
3K
  • · Replies 25 ·
Replies
25
Views
3K