If I have the function f(x)=3 and I take the integral with respect to x from 0 to 2, the answer is 6. Now what if I have the function f(x)=3 BUT x is undefined at 0. Well, this means that a line has been subtracted from our original area of 6. But a line has zero area so we still have an area of 6 from 0 to 2 even if 0 is undefined. So why is it that f(x) must be continuous from 0 to 2 if we want to take the integral from 0 to 2.

  • $\begingroup$ To talk about the integrability of a function we only need to want that it be bounded, not continuous, in that interval. $\endgroup$ – azif00 Sep 8 '19 at 1:50
  • $\begingroup$ So what you're saying is if I took the integral of $f(x)=3$ from $0$ to $2$ and $x=0.1, 0.3, 0.4,$ and $0.8$ were undefined I could still take the integral? $\endgroup$ – user532874 Sep 8 '19 at 1:58
  • $\begingroup$ Yes. More surprising that a function like this can be integrated :) $\endgroup$ – azif00 Sep 8 '19 at 2:10
  • $\begingroup$ I have taken the liberty to suppress tags "geometry" and "paradoxes" which are not meaningful here. $\endgroup$ – Jean Marie Sep 8 '19 at 2:54
  • $\begingroup$ @Azif00 I have one last question. What is the definition of bounded in this case? $\endgroup$ – user532874 Sep 8 '19 at 3:58

Your reasoning is right!

In principle we do not need to talk about the continuity of a function if we want to know whether or not it can be integrable in a certain closed interval. To talk about the integrability of a function we only need to want that it be bounded in that interval.

I leave you another, but similar, example. Take $f: [0,2] \to \Bbb R$ defined by $$f(x) = \left\{ \begin{align} 1 & \textrm{ if } x=1 \\ 0 & \textrm{ if } x\neq 1 \end{align} \right.$$ then $$\int_0^2 f(x)dx=0$$ To prove this rigorously, we turn to the definition of integral. Suppose $P=\{ t_0,t_1,\dots,t_n\}$ is a partition of the interval $[0,2]$ (this means $t_0=0$, $t_n=2$ and $t_{i-1}<t_i$ for all $i=1,\dots,n$) with $t_{j-1}<1<t_j$ for some $j$ . Let's call $$m_i=\inf \, \{ f(x): t_{i-1}\leq x\leq t_i \}$$ $$M_i=\sup \, \{ f(x): t_{i-1}\leq x\leq t_i \}$$ for $i=1,2,\dots,n$. Then $m_i=M_i=0$ if $i\neq j$, but $m_j=0$ and $M_j=1$. Since the lower and upper sums turns into $$L(f,P)=\sum_{i=1}^n m_i(t_i-t_{i-1})=0$$ $$U(f,P)=\sum_{i=1}^n M_i(t_i-t_{i-1})=t_j-t_{j-1}$$ it follows that, we can make the differences $U(f,P)-L(f,P)$ as small as possible (only choose a partition with $t_j-t_{j-1}$ enough small). This means that $f$ is integrable in that interval and since $$L(f,Q)\leq 0\leq U(f,Q) \quad \textrm{for all the partitions } Q$$ it follows $$\int_0^2 f(x)dx=0$$ as we want to show. (The latter is because the integral is defined as the infimum of all the upper sums and the supremum of all the lower sums.)


Your Answer

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