This came up in a proof I was reading. Define $$\inf_{z \in K} \|x-z\| = d$$

Let $y_n\in K$ be a minimizing sequence How do we know that such a minimizing sequence exists? Here K is a closed convex subset of a Hilbert space H. Is this some axiom for from the reel numbers?

  • $\begingroup$ Just remember what the infimum of a set of real numbers is. $\endgroup$ – Thomas Dec 17 '12 at 10:24

As pointed out by Thomas, it follows from the definition of infimum.

A subset $A$ of the real numbers is said to be bounded below if there is $b \in \mathbb{R}$ such that $b \leq a$ for all $a \in A$. For such a set, the infimum of $A$ is the greatest lower bound for $A$ which is guaranteed to exist due to the completeness of $\mathbb{R}$.

As $\inf_{z\in K}\|x - z\| = d$ we know that $d \leq \|x - z\|$ for every $z \in K$. Now for each $n \in \mathbb{N}\setminus\{0\}$ let $y_n \in K$ be such that $d \leq \|x - y_n\| \leq d + \frac{1}{n}$. For each $n$, such a $y_n$ exists; if no such $y_n$ existed for a particular $n$, then we would have $d + \frac{1}{n} \leq \|x - z\|$ for all $z \in K$ which contradicts the fact that $\inf_{z\in K}\|x-z\|=d$.

So we have a sequence $(y_n)$ such that $d \leq \|x-y_n\| \leq d + \frac{1}{n}$. Taking limits we obtain $d \leq \displaystyle\lim_{n\to\infty}\|x-y_n\|\leq d$ and hence $\displaystyle\lim_{n\to\infty}\|x-y_n\| = d$.

  • $\begingroup$ When defining inf$_{z \in K} \lVert x-z \rVert = d$, how do we know that such a $d$ exists? $\endgroup$ – A Slow Learner Mar 21 at 18:23
  • $\begingroup$ @ASlowLearner: The set $\{\|x - z\| \mid z \in K\}$ is bounded below by zero. $\endgroup$ – Michael Albanese Mar 21 at 18:24

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.