2
$\begingroup$

In chapter $13$ in Spivak Calculus in the beginning of Integrals chapter he introduced a new definition which is as follow :

Suppose $f$ is bounded on $[a,b]$ and $P$ is a partition of $[a,b]$ . Let

$m_{i}=inf\{ f(x):t_{i-1}\le x \le t_{i}\}$

$M_{i}=sup\{ f(x):t_{i-1}\le x \le t_{i}\}$

then the Lower sum denoted by $L(f,P) =\sum_{i=1}^{n} m_{i}(t_{i}-t_{i-1})$

and the Upper sum denoted by $U(f,P)=\sum_{i=1}^{n} M_{i}(t_{i}-t_{i-1})$

My question is as follows :Is the using of $inf$ and $sup$ insted of $max$ and $min$ indicates that the function may not be well defined at the end points of the interval which is $a$ and $b$ $?$

because I read that the function to be riemann integrable must be defined on the whole interval $[a,b]$ so does the definition in Spivak mean something else or not

$\endgroup$

1 Answer 1

6
$\begingroup$

The definition uses $\inf$ and $\sup$ because the sets $\{f(x)\mid t_{i-1}\leqslant x\leqslant t_i\}$ is bounded (since $f$ is bounded) and non-empty, and therefore they have a supremum and and an infimum. But those sets don't have to have a maximum or a minimum. For instance, if $a=0$, $b=1$, $P=\{0,1\}$, and$$f(x)=\begin{cases}\cos(\pi x)&\text{ if }x\in[0,1]\setminus\Bbb Q\\0&\text{ otherwise,}\end{cases}$$then $\{f(x)\mid0\leqslant x\leqslant1\}$ has no maximum or minimum. But its supremum is $1$ and its infimum is $-1$.

$\endgroup$
5
  • $\begingroup$ @BernardPan I've edited my answer. Thank you. $\endgroup$ Aug 15 at 22:09
  • 2
    $\begingroup$ or another example (I'm sure you know this; this is just for OP) $f(x)=x$ for $x\in (0,1)$ and $f(0)=f(1)=\frac{1}{2}$. This is a bounded function on $[0,1]$ with no minimum/maximum, but infimum/supremum of $0,1$ respectively. $\endgroup$
    – peek-a-boo
    Aug 15 at 22:11
  • $\begingroup$ @JoséCarlosSantos No problem. Very nice construction! $\endgroup$ Aug 15 at 22:12
  • $\begingroup$ @JoséCarlosSantos So we use $inf$ and $sup$ because sometimes the function can't have a $max$ or $min$ but the function must be defined on the whole interval $[a,b]$ to be riemann integrable on the interval $[a,b]$, is that right ? $\endgroup$ Aug 15 at 22:27
  • $\begingroup$ Yes, that is correct. $\endgroup$ Aug 15 at 22:29

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.