eigenmath/doc/ross/ross-18.2.tex

26 lines
958 B
TeX
Raw Permalink Normal View History

2004-10-14 00:54:52 +02:00
\beginsection 18.2
Reread the proof of Theorem 18.1 with $[a,b]$ replaced by $(a,b)$.
Where does it break down? Discuss.
\medskip
2004-10-26 17:44:55 +02:00
Try going through the proof line by line replacing $[a,b]$ with $(a,b)$.
2004-10-14 00:54:52 +02:00
2004-10-26 17:44:55 +02:00
1. Assume that $f$ is not bounded on $(a,b)$. {\it Ok.}
2004-10-14 00:54:52 +02:00
2004-10-26 17:44:55 +02:00
2. Then to each $n\in N$ there corresponds an $x_n\in (a,b)$ such that
2004-10-14 00:54:52 +02:00
$|f(x_n)|>n$. {\it Ok, because f is unbounded.}
2004-10-26 17:44:55 +02:00
3. By the B-W Theorem 11.5, $(x_n)$ has a subsequence $(x_{n_k})$ that converges
2004-10-14 00:54:52 +02:00
to some real number $x_0$.
{\it Ok, because $x_n$ is bounded on $(a,b)$.}
2004-10-26 17:44:55 +02:00
4. The number $x_0$ must also belong to the open interval $(a,b)$.
2004-10-14 00:54:52 +02:00
{\it Uh-oh, here's a problem, $x_0$ may be equal to $a$ or $b$ in the limit.}
2004-10-26 17:44:55 +02:00
5. Since $f$ is continuous at $x_0$, we have $\lim f(x_{n_k})=f(x_0)$,
2004-10-14 00:54:52 +02:00
but we also have $\lim|f(x_{n_k})|=+\infty$ which is a contradiction.
{\it If $x_0$ is equal to $a$ or $b$ then $f$ doesn't have to be continuous
there so we cannot say that $\lim f(x_{n_k})=f(x_0)$.}