3

Let $f:[0,T]\to \mathbb{R}$ a Riemann integrable function that satisfies $f(t)>0$ for all $t\in [0,T]$.

Consider the integral $$g(t)=\int_{0}^{t}f^{2}(u)du$$

Is it true that $g(t)$ is strictly increasing?

user438991
  • 235
  • 2
  • 7

2 Answers2

3

The question is not as simple as it sounds. It is based on the following result:

If $f$ is Riemann integrable on $[a, b] $ and $f(x) >0$ for all $x\in[a, b] $ then $\displaystyle \int_{a} ^{b} f(x) \, dx>0$.

This can be proved using the following deep result :

If $f$ is Riemann integrable on $[a, b] $ then it is continuous at some point in $[a, b] $.

For your function $g$ note that $$g(y) - g(x) =\int_{x}^{y}f^{2}(u)\,du>0$$ if $x<y$ so that $g$ is indeed strictly increasing.


Update: On request of user gimusi via comment I am providing some explanation for the common approach to this problem and its limitations.

Let's relax our conditions and assume that $f$ is discontinuous only at a finite number of points say $t_{1}, t_{2}, \dots, t_{n}$ in interval $[0,T]$. Then $f^{2}$ is Riemann integrable on $[0,T]$ and $g$ is continuous on $[0,T]$. These points $t_{i} $ together with $0,T$ form a partition of interval $[0,T]$ and let this partition be $$P=\{x_{0}=0,x_{1},x_{2},\dots,x_{k}=T\} $$ Clearly $f^{2}$ is continuous on each open interval $(x_{i-1},x_{i})$ and therefore via Fundamental Theorem of Calculus $g'(t) =f^{2}(t)>0$ on each of these open intervals. Since $g$ is continuous on the closed interval $[x_{i-1}, x_{i}] $ it follows via mean value theorem that $g$ is strictly increasing in each closed interval $[x_{i-1},x_{i}]$ and hence it is strictly increasing on the whole interval $[x_{0},x_{k}]=[0,T]$.

Thus the usual approach based on sign of derivative can be used to show that $g$ is strictly increasing but only under the less general condition that $f$ has a finite number of discontinuities in $[0,T]$. For a general Riemann integrable function $f$ it is possible that the number of discontinuities is infinite (countably or uncountably). And then the above approach doesn't work.

On the other hand instead of counting the discontinuities of $f$ the proper approach is to measure them in the manner specified by Lebesgue and by doing so we get another deep theorem (with some effort):

If $f$ is bounded on $[a, b] $ then it is Riemann integrable on $[a, b] $ if and only if the set of discontinuities of $f$ on $[a, b] $ is of measure zero.

A trivial implication of the above result is that a Riemann integrable function must be continuous at infinitely many points which brings us back to the result mentioned (and linked) earlier.

  • Thanks for your observation! – user Dec 12 '17 at 07:02
  • If f is piecewise continuos can you concude that g is continuous and piecewise differentiable? in this case can you convlude that in each interval g’>0 and thus g is strictly increasing? Thanks. – user Dec 12 '17 at 07:43
  • @gimusi: the following is guaranteed : if $g(x) =\int_{a} ^{b} f(t) , dt$ then $g$ is continuous in $[a, b] $ and $g$ is differentiable at those points where $f$ is continuous and at those points $g'(x) =f(x) $. If $f$ is having simple discontinuity at $c$ then $g'(c) $ does not exist. If $f$ is having any other type of discontinuity then $g'(c) $ may or may not exist. – Paramanand Singh Dec 12 '17 at 07:57
  • @gimusi: if in current question if $f$ is discontinuous at a finite number of points then you can partition the whole interval into a number of subintervals and apply $g'>0$ in each open sub-interval. With $g$ continous this allows strictly increasing in whole of the interval. – Paramanand Singh Dec 12 '17 at 07:58
  • @gimusi : however Riemann integrable functions can be discontinuous at infinitely many points. In fact the set of discontinuities can be uncountable. The right idea here is not to count them, but measure them and we have the result that a bounded function is Riemann integrable on a closed interval if and only if it's set of discontinuities form a set of measure zero. This is bit technical to explain. – Paramanand Singh Dec 12 '17 at 08:00
  • Exactly I had in mind the case of f with a finite number of discontinuity points! I’ve to think better for all the other cases. Thanks a lot for the explanation! – user Dec 12 '17 at 08:05
  • @gimusi: I am glad that you took my comments in positive manner. I don't know exact level of your mathematical education, but this idea of Riemann integral is not very straightforward for discontinuous functions and typical intro calculus books do it only for continuous functions. If you have a chance to study real-analysis you will be more comfortable with whatever I have mentioned. – Paramanand Singh Dec 12 '17 at 08:07
  • @gimusi : you may wish to edit your answer with the disclaimer that it applies to finite number of discontinuities and I will reverse my downvote. After all the discussion I don't think you really deserve a downvote. – Paramanand Singh Dec 12 '17 at 08:10
  • Indeed I’m here to learn more from person like you. You should add your further explanation in your answer, they are very clear and useful. Thanks! – user Dec 12 '17 at 08:12
  • @gimusi : Ok will add more information in my answer. My update will take some time as I am going to get busy with some other work right now. – Paramanand Singh Dec 12 '17 at 08:14
  • Also for me, bye! – user Dec 12 '17 at 08:16
  • @gimusi : you may have a look at the updated answer when you get online. – Paramanand Singh Dec 12 '17 at 11:49
  • Great explanation. Thanks! – user Dec 12 '17 at 12:43
0

If $f$ is continuos then:

$$g(t)=\int_{0}^{t}f^{2}(u)du \implies g'(t)=f^2(t)>0$$

Thus $g$ is strictly increasing

NOTE

it also applies to finite number of discontinuities (see Paramanand Singh answer and comments)

the converse is not necessary true

user
  • 154,566
  • 2
    Perhaps you can't see deleted answers which made same mistake as yours. You are assuming continuity of integrand which is not given. – Paramanand Singh Dec 12 '17 at 06:30
  • g(t) is necessary continuos as it admits derivative thus as g'>0 it is strictly increasing – user Dec 12 '17 at 06:33
  • 2
    I am talking about continuity of $f$. By the way $g$ is guaranteed to be continuous but not necessarily differentiable. You need to know how Fundamental Theorem of Calculus works in more detail. The equation $g'(t) =f^{2}(t)$ may not hold if $f$ is not continuous. – Paramanand Singh Dec 12 '17 at 06:35
  • @ParamanandSingh Is it not true that if $g'(x)>0 \implies g$ strictly increasing? Could give a counter example please. Thanks. – user Dec 12 '17 at 06:38
  • 2
    The point is that $g$ is not guaranteed to be differentiable and hence one can't use something like $g'(x) >0$. You should read comments more carefully before replying. – Paramanand Singh Dec 12 '17 at 06:39
  • I have reversed my downvote to an upvote. Now your answer clearly mentions the assumptions under which your approach works and thus it is mathematically correct. – Paramanand Singh Dec 12 '17 at 11:59