0

I’m trying to figure out a proof that if $g(x)$ is the antiderivative of $f(x)$ where $f$ is Riemann integrable then $g$ is continuous.

In other words: if $ \int_{0}^{x}f \space dt = g(x) $ and $f$ is Riemann integrable on all intervals but not necessarily continuous, does it follow that $g(x)$ is continuous ?

FD_bfa
  • 3,989
Gracie
  • 37

1 Answers1

0

Consider the standard calculus definition of continuity: The function $f$ is continuous at $x$ iff

$$f(x+h) \xrightarrow[h\to0\,]{} f(x)\,.\vphantom{\int}$$

Now apply this to our function $g$ and see what happens. The Riemann integrable $f$ should be defined on some closed interval. Let $x$ be an inner point of this interval and consider $g(x+h)$ for some small $h$. (If $x$ were a boundary point we should only check right or left continuity.) If we want to be careful we can split $g(x+h)$ into two parts:

$$g(x+h) = \int_0^{x+h}f(t)dt = \int_0^{x}f(t)dt + \int_x^{x+h}f(t)dt \,.$$

If $h$ is negative the last term is defined as $\left(-\int_{x+h}^x f(t)dt\right)$.

Now, $$g(x+h) = g(x) + \int_x^{x+h}f(t)dt$$

and the only thing we need, to get what we want, is

$$\int_x^{x+h}f(t)dt\ \xrightarrow[h\to 0\,\,]{} \int_x^{x}f(t)dt\ = 0 \,,$$

which looks very plausible. To be formal, we can choose $a<x<b$ such that $f$ is defined on $[a,b]$ and $M>0$ such that $M\geq |f(t)| \;\forall t\in [a,b].$ Then

$$\left|\int_x^{h}f(t)dt\right|\leq \int_x^h |f(t)|dt \leq M|h| \xrightarrow[h\to 0\,]{} 0\,.$$

Thus $$g(x+h) = g(x) + \int_x^{x+h}f(t)dt \xrightarrow[h\to 0\,]{} g(x)\,,$$

which means that $g(x)$ is continuous at $x$.

Lorents
  • 319
  • Decent answer, but you need to show rigorously that you can interchange the limit and the integration, or at least reference a theorem that allows you to do so. – K.defaoite Mar 17 '23 at 12:30
  • @K.defaoite Where do I interchange a limit and integration? My argument boils down to $\lim_{h\to 0 } \int_0^{x+h} fdt= \int_0^{x} fdt$. – Lorents Mar 17 '23 at 12:42
  • @K.defaoite You probably mean something like "move the limit up to the bound of integration". Adding a comment about this. – Lorents Mar 17 '23 at 15:11
  • ...or incorporating it into the argument... – Lorents Mar 17 '23 at 15:46