Intersection of a closed convex set

Click For Summary
SUMMARY

The discussion focuses on the intersection of a closed convex set C within a real Banach space X, utilizing the Hahn-Banach Theorem to demonstrate that C equals the intersection of half-spaces defined by linear functionals in the set Lc. The corollary of the Hahn-Banach theorem is applied to establish the existence of a non-zero linear functional f in the dual space X* that satisfies specific inequalities for points in C and an external point a. The proof confirms that C indeed contains the intersection of these half-spaces, establishing a definitive relationship between convex sets and linear functionals.

PREREQUISITES
  • Understanding of real Banach spaces
  • Familiarity with the Hahn-Banach Theorem
  • Knowledge of dual spaces and linear functionals
  • Concept of convex sets and their properties
NEXT STEPS
  • Study the implications of the Hahn-Banach Theorem in functional analysis
  • Explore the properties of convex sets in real Banach spaces
  • Learn about the Minkowski functional and its applications
  • Investigate the role of dual spaces in optimization problems
USEFUL FOR

Mathematicians, particularly those specializing in functional analysis, as well as students and researchers interested in the properties of convex sets and their applications in optimization and duality theory.

Funky1981
Messages
21
Reaction score
0
Let X be a real Banach Space, C be a closed convex subset of X.

Define Lc = {f: f - a ∈ X* for some real number a and f(x) ≥ 0 for all x ∈ C} (X* is the dual space of X)

Using a version of the Hahn - Banach Theorem to show that

C = ∩ {x ∈ X: f(x) ≥ 0} with the index f ∈ Lc under the intersection Could someone help me to solve this problem, i can't see how Hahn - Banach can imply the above statement ( I used the separation version to obtain g(x)<a<g(y) for some linear functional g)
 
Physics news on Phys.org
You will need the following corollary of the Hahn--Banach theorem (with sublinear functional): if ##U## is a convex open set, and ##a\notin U## then there exists a non-zero linear functional ##f\in X^*## such that ##f(x)\le f(a)## for all ##x\in U##. To prove the corollary you first notice that because of translation invariance, you can assume without loss of generality that ##0\in U##.

Then you define a sublinear functional ##q## to be the Minkowski functional of ##U##, and define ##f_0## to be the linear functional on the one-dimensional space spanned by ##a## such that ##f_0(a)=1##. Extending ##f_0## by Hahn--Banach you get a functional ##f## such that ##f(x)\le q(x)## for all ##x\in X##. Then for all ##x\in U## $$ f(x) \le q(x) \le 1 = q(a),$$ so we get the desired inequality.
We still need to show that ##f\in X^*## (i.e. that ##f## is bounded), but that is easy: ##U## is open, so an open ball ##B_\varepsilon \subset U##, so ##q(x), q(-x)\le \frac1\varepsilon \|x\|## (##B_\varepsilon## is an open ball of radius ##\varepsilon## centered at ##0##), and therefore ##|f(x)|\le \frac1\varepsilon \|x\|##.

To get the statement you need, take ##a \notin C##. Then for some ##\varepsilon>0## the whole open ball ##x+B_\varepsilon## does not intersect ##C##, or, equivalently, the (open) set ##U:= C+B_\varepsilon## does not contain ##a##. Applying the above corollary yo get that there exists ##f\in X^*## such that
$$ f(x)\le f(a)$$ for all ##x\in U##, or, equivalently, $$f(x) \le f(a) + f(y) \qquad \forall x\in C, \ \forall y\in B_\varepsilon.$$ Noticing that for any non-zero ##f\in X^*## the point ##0## is an interior point of ##f(B_\varepsilon)## (in fact ##f(B_\varepsilon)## is an open set), we conclude that there exists real ##b## such that $$f(x) \le b <f(a) \qquad \forall x\in C.$$ The last statement means that ##C## contains the intersection of the half-spaces. The opposite inclusion is trivial.
 

Similar threads

  • · Replies 4 ·
Replies
4
Views
3K
  • · Replies 9 ·
Replies
9
Views
3K
  • · Replies 24 ·
Replies
24
Views
5K
  • · Replies 1 ·
Replies
1
Views
3K
  • · Replies 2 ·
Replies
2
Views
3K
  • · Replies 2 ·
Replies
2
Views
2K
  • · Replies 9 ·
Replies
9
Views
2K
  • · Replies 1 ·
Replies
1
Views
2K
  • · Replies 12 ·
Replies
12
Views
6K
  • · Replies 0 ·
Replies
0
Views
2K