I want to show the following proposition

Let $[a,b]$ be a compact interval and $f: [a,b] \to \mathbb R$ be a bounded, piecewise monotone function. Then $f$ is Riemann integrable.

Our definition of piecewise monotonicity is

Let $[a,b]$ be a compact interval and $f: [a,b] \to \mathbb R$ a function. $f$ is piecewise monotone iff there exists a partition $\xi = \{a=x_0 \lt \ldots \lt x_n = b\}$ of $[a,b]$, such that $f_k := f|_{(x_{k-1},x_k)}$ is monotone for all $k \in \{1,\ldots,n\}$.

I am allowed to use the fact that a monotone function on a compact interval is Riemann integrable, as well as the additivity property, and, that changing one point of a Riemann integrable function doesn't change it's integrability.

My attempt:

Since $f_k$ is monotone, it's Riemann integrable. We use the additivity of the Riemann integral to combine $f_1$ and $f_{2}$, then we can combine those with $f_3$ and so on. Therefore

$$f(x) = \begin{cases} f_1(x) & \text{if } x \in [x_0,x_1) \\ f_2(x) & \text{if } x \in [x_1,x_2) \\ \ldots \\ f_n(x) & \text{if } x \in [x_{n-1},x_n] \end{cases}$$

is Riemann integrable.

I don't feel comfortable with this argument, it seems too simple. And also I think the additivity property requires that the edges of the adjacent intervals match up (i.e. $f_k(x_k) = f_{k+1}(x_k)$) which might not necessarily be true. Can somebody help me out?

  • $\begingroup$ I was about to say that "bounded" is not needed (or: is automatic), but it seems that $f$ may be increasing and decreasing alternatingly. $\endgroup$ – Hagen von Eitzen Dec 31 '17 at 16:39
  • $\begingroup$ @HagenvonEitzen We are given the following example function which is apparently Riemann integrable: imgur.com/a/4lGBi (I should add it's defined on $[0,1] \to [0,1]$) $\endgroup$ – philmcole Dec 31 '17 at 16:41
  • 1
    $\begingroup$ If a function is Riemann integrable on finitely many disjoint subintervals of $[a,b]$ whose union is $[a,b]$, then it is Riemann integrable on $[a,b]$. Hence it suffice to prove that a monotonic $f:[a,b] \to \mathbb{R}$ is Riemann integrable. $\endgroup$ – AlohaSine Dec 31 '17 at 21:46
  • $\begingroup$ For the function $g$ of the image, one needs infinitely many subintervals on which $g$ is monotone (decreasing). That doesn't qualify as piecewise monotone by the definition you mention, which requires that $[a,b]$ be partitioned into only finitely many parts. (I don't know if this is a problem for what you want to show.) $\endgroup$ – coffeemath Dec 31 '17 at 21:46

There is nothing wrong with your argument. It is simple because this is a simple problem.

Integration does not depend on the value of the function at isolated points, so you can define $f_k$ however you please at the partition points and still have that $f_k$ is integrable on $[x_{k-1},x_k]$ even though monotonicity fails at the endpoints.


Just extend the definition of $f_k$ at end points of the subinterval $[x_{k-1},x_{k}]$ so that $f_k$ is monotone on the compact subinterval $[x_{k-1},x_{k}]$. Then $f_k$ is integrable on this subinterval. And since $f$ differs from $f_{k} $ possibly at the end points of this subinterval, it follows that $f$ is also integrable on this subinterval. By additivity it is integrable on whole of $[a, b] $.


You can also use that a monotone function is continuous a.e. Continuity of Monotone Functions

and since a continuous function is bounded on any compact interval $[x_{k-1}, x_k]$ by the Extreme Value Theorem, it is also Riemann integerable there. Just apply this to every sub-interval in the partition.


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.