Call $f: [a,b] \to \mathbb{R}$ a step function if there exist a partition $P=\{x_0, \ldots, x_n \}$ of $[a,b]$ such that $f$ is constant on the interval $[x_i, x_{i+1})$.

I'm having some issues proving that f is Riemann integrable. If $f$ was defined such that it was constant on the closed interval $[x_i, x_{i+1}]$ then it would be trivial to define a partition such that the supremum and infinum coincide, and the upper and lower sums would cancel. But since we are working with a half-open interval, I'm having issues defining the partitions that will allow me to prove integrability. All proofs I've seen online rely on theorems we haven't proven in class, so I'm assuming I must be missing something simple.


This can be done by using the characterization that the difference between upper and lower sums can be made arbitrarily small, which is one of the very first results when studying Riemann integrability.

The idea is to "manage the bad points", the bad points being where you have the jump discontinuities, since as you point out, the supremums and infimums kind of mess up the Riemann sums at these points.

Let $\epsilon>0$ be given, let $M=\max_{x,y\in[a,b]}|f(x)-f(y)|$, and since we have $n+1$ jumps, set $\delta=\frac{\epsilon}{M(n+1)}$. Now consider the partition $P'=\{x_0,x_0+\delta,x_1\pm\delta,x_2\pm\delta,...,x_{n-1}\pm\delta,x_n-\delta,x_n\}$. Can you bound $U(P',f)-L(P',f)$ by something small?


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.