Question about a certain part of Weierstrass extreme value theorem

38 Views Asked by At

The lemma says that if $A \in \mathbb{R}^{n}$ is compact and $f:A \to \mathbb{R}$ is continuous in $A$, then there's a real $M$ such that $\forall X \in A,f(X) \leq M$.

My textbook's proof goes this way: if $f(X)$ didn't have such an upper bound, then there would be a sequence $\{X_k\}_{k \in \mathbb{N}} \subseteq A$ such that $\forall k \in \mathbb{N},f(X_k) \geq k$. But since $A$ is compact, there's a convergent subsequence $\{X_{k_j}\}$ such that $X_{k_j} \to P \in A$. Therefore $f(X_{k_j}) \geq k_j$. But this is impossible since $f$ is continuous in $A$.

Does this last step amount to showing that $f(P)$, if defined, would have to be greater than every natural number (which contradicts the Archimedean property)?

I also don't understand how to justify that last step. A previous lemma stated that convergent sequences of $A$ whose images under $f$ were bounded implied that the image of the limit of those sequences was also bounded. But in this case we're "pushing forward" the bound every time. How's this justified?

1

There are 1 best solutions below

0
On BEST ANSWER

By assumption $f(P)$ is a real number. Hence $k_j >f(P)+1$ for $j$ sufficiently large. But $f(X_{k_j}) \to f(P)$ and $f(X_{k_j})\geq k_j>f(P)+1$ for $j$ sufficiently large. This is a contradiction, right?