16
$\begingroup$

Question: Let $f:[0,1]\to\mathbb{R}$ be a continuous function such that $$\int_0^1f(x)dx=\int_0^1xf(x)dx.$$ Show that there is a $c\in(0,1)$ such that $$f(c)=\int_0^cf(x)dx.$$

My solution: Define the function $g:[0,1]\to\mathbb{R}$, such that $$g(x)=x\int_0^x f(t)dt-\int_0^x tf(t)dt, \forall x\in[0,1].$$

Now since $f$ is continuous on $[0,1]$, thus we can conclude by Fundamental Theorem of Calculus that $g$ is differentiable $\forall x\in[0,1]$ and $$g'(x)=\int_0^x f(t)dt+xf(x)-xf(x)=\int_0^xf(t)dt, \forall x\in[0,1].$$

Observe that $g(0)=g(1)=0$. Hence by Rolle's Theorem we can conclude that $\exists b\in(0,1)$, such that $g'(b)=0$, i.e $$\int_0^b f(t)dt=0.$$

Now define $h:[0,1]\to\mathbb{R}$, such that $$h(x)=e^{-x}g'(x), \forall x\in[0,1].$$

Now $h'(x)=-e^{-x}g'(x)+g''(x)e^{-x}=e^{-x}(g''(x)-g'(x)), \forall x\in[0,1].$

Observe that $h(0)=h(b)=0$. Hence by Rolle's Theorem we can conclude that $\exists c\in(0,b)\subseteq (0,1)$, such that $h'(c)=0$. This implies that $$e^{-c}(g''(c)-g'(c))=0\\\implies g''(c)-g'(c)=0\hspace{0.3 cm}(\because e^{-c}\neq 0)\\\implies f(c)=\int_0^cf(x)dx.$$

Is this solution correct? And is there a better solution that this?

$\endgroup$
3
4
$\begingroup$

Your proof is correct. This is another one.

We may assume that $f$ is not identically zero (otherwise it is trivial). Since $f$ is continuous and $$\int_0^1(1-x)f(x)\,dx=0$$ we have that $M=\max_{x\in [0,1]}f(x)>0$ and $m=\min_{x\in [0,1]}f(x)<0$. Moreover $\exists x_M,x_m\in [0,1]$ such that $f(x_M)=M$ and $f(x_m)=m$. Let us consider the following continuous map $$F(x):= f(x) - \int_0^xf(t)\,dt.$$ If $x_M<1,$ then $$F(x_M)=M-\int_0^{x_M}f(t)\,dt\geq M- Mx_M >0.$$ If $x_M=1$ then, $$F(x_M)=M-\int_0^{1}f(t)\,dt> 0$$ because $f$ strictly less than $M$ in an interval of positive length containing $x_m$. In both cases we conclude that $F(x_M)>0$. In a similar way, we show that $F(x_m)<0$.

Finally, since $F$ is continuous on $[0,1]$, it follows, by the Intermediate Value Theorem, that there exists $c$ strictly between $x_M$ and $x_m$, and therefore $c\in (0,1)$, such that $F(c)=0$, that is $$f(c)=\int_0^cf(t)\,dt.$$

$\endgroup$
2
  • $\begingroup$ +1 It appears that the approach would also work for the question linked in my comment to the current question. $\endgroup$ Apr 23 '20 at 2:02
  • 1
    $\begingroup$ Another observation: what you have proved is that if a continuous $f$ changes sign in $[0,1]$ then we have a $c\in(0,1)$ such that $f(c) =\int_{0}^{c}f(x)\,dx$. $\endgroup$ Apr 23 '20 at 2:08
1
$\begingroup$

As noted by RobertZ, your proof is correct. Here is another proof that follows the same outline as yours: first we find another zero for the antiderivative of $f$ and then we use Rolle's theorem in an appropriate way. This approach is admittedly more long-winded but doesn't make use of the $e^{-x}$ trick.

Define $F: [0,1] \to \mathbb{R}$ as $F(x) =\displaystyle \int_{0}^{x}f(t)dt.$ Note that the given condition can be stated as $\displaystyle \int_{0}^{1}F(t)dt =0$

Claim 1: There exists $b \in (0,1)$ such that $F(b) =0.$

Proof of claim 1: By the mean value theorem for integrals there exists $b \in (0,1)$ such that $F(b)= \displaystyle \int_{0}^{1}F(t)dt$, which implies $F(b)=0.$

Now, we look for an appropriate sub-interval of $[0,b]$ on which we can apply Rolle's theorem to $g.$

Let $G(x)=\displaystyle \int_{0}^{x}F(t)dt$ and define $g:[0,b] \to \mathbb{R}$ by $g(x)= G(x) -F(x).$

Claim 2: $g$ is not injective on $[0, b].$

Proof of claim 2: Suppose not. Then $g$ is injective and since it is clearly continuous too, $g$ is monotone. WLOG, let $g$ be monotone increasing. Then since $g$ is differentiable, $g'(x) \geq 0 \, \forall \, x \in [0,1].$ If there exists at least one $x$ for which $g'(x) =0$ we are done so assume $g'(x)>0.$ Since $g(0) =0,$ we have $g(x)>0$ for all $x \in (0,b].$

Let $x_{0}$ be a point of maximisation for $F.$ Assume $F$ is not identically $0$ or else $f$ is and the problem is trivial. We claim that there exists $c \in (0, b)$ such that $F(c)<0.$ If $x_{0}=0$ or $b$ then $F\leq 0$ so if $F$ is not identically $0$ choose an other point of $(0, b)$ to be $c.$ If $x_{0} \in (0, b)$ then since $g({x}_{0})>0, F(x_{0})< \displaystyle \int_{0}^{x_{0}}F(t)dt \leq x_{0}F(x_{0}).$

If $F(x_{0}) \neq 0$ we get $x_{0} \geq 1,$ a contradiction. Hence $F(x_{0})=0$ and since $F$ is not identically $0$ there exists some $c \in (0, b)$ such that $F(c)<0.$

Since $F$ is a continuous function on a closed and bounded interval $[0, b]$, it attains its bounds. In particular $\exists \, d \in [0, b]$ such that $F(d)\leq F(x) \, \forall \, x \in [0,b].$ Clearly $d\neq 0, 1$ or else $F(x) \geq 0 \, \forall x \in [0,b]$ contradicting the fact that $F(c) <0.$ Therefore $d \in (0,b)$ and since it is a point of minimisation, $F'(d) =0.$ Then $F(d)= F(d) -F'(d) =g'(d)>0> F(c)$ contradicting the fact that $d$ is a point of minimisation of $F.$ Therefore our hypothesis that $g$ is injective is false and hence $g$ is not injective and there exists $a, a' \in [0, b]$ with $a \neq a'$ such that $g(a) =g(a').$

Then since $g$ restricted to $[a, a']$ satisfies the conditions for Rolle's Theorem, there exists some $x_0 \in (a,a')$ such that $g'(x_0)=0$ which implies $F(x_0)=F'(x_0)$ from which it follows that $f(x_{0}) = \displaystyle \int_{0}^{x_{0}}f(x)dx$

Note that the proof follows almost identically if we assume $g$ to be monotone decreasing in the proof of the claim.

$\endgroup$

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.