Let $f$ be continuous on $\mathbb{R}$. Due to FTC I, we know that a function of the form∗ $F(x) = \int_a^xf(t)\operatorname dt$ is always an antiderivative of $f(x)$. In this question you will investigate whether all antiderivatives of $f(x)$ can be expressed in this form∗. For simplicity, let us further assume $f$ is non-negative $(i.e. ∀x ∈ \mathbb{R}, f(x) ≥ 0)$.

(a) Suppose$\lim\limits _{A\rightarrow\infty}\int_0^Af(t)\operatorname dt$ or $\lim\limits _{A\rightarrow-\infty}\int_A^0f(t)\operatorname dt$ is finite, show there is an antiderivative $G(x)$ of $f(x)$ which does not equal $\int_a^xf(t)\operatorname dt$ for any a $\in \mathbb{R}$

(b)Suppose $\lim\limits _{A\rightarrow\infty}\int_0^Af(t)\operatorname dt=\infty$ and $\lim\limits _{A\rightarrow-\infty}\int_A^0f(t)\operatorname dt=\infty$, show for any antiderivative $G(x)$ of $f(x), ∃a ∈ \mathbb{R} \text{ s.t.} G(x) = \int_a^xf(t)\operatorname dt$

Hint: Think about whether antiderivatives of f(x) need to have zeroes.

What I have tried so far:

Look thorugh (a) and (b), it's saying if $f$ is continuous on $\mathbb{R}$ we have:

$(\lim\limits _{A\rightarrow\infty}\int_0^Af(t)\operatorname dt=\pm\infty \wedge \lim\limits _{A\rightarrow-\infty}\int_A^0f(t)\operatorname dt=\pm\infty )\leftrightarrow \forall G(x), ∃a ∈ R \text{ s.t.} G(x) = \int_a^xf(t)\operatorname dt$

(This is a stronger version of the question, since negation of finite also include $-\infty$, I'm not sure if this is still true, but this should implies what the question is asking to prove)

By assumption, $f$ is non-negative, then we don't need to consider the $-\infty$ cases, just show the following would be sufficient:

$(\lim\limits _{A\rightarrow\infty}\int_0^Af(t)\operatorname dt=\infty \wedge \lim\limits _{A\rightarrow-\infty}\int_A^0f(t)\operatorname dt=\infty )\leftrightarrow \forall G(x), ∃a ∈ R \text{ s.t.} G(x) = \int_a^xf(t)\operatorname dt$

I don't have the Intuition of why this is true, at least it's not very trivial to me..

So, first I tried to break it into definitions:

1.$\lim\limits _{A\rightarrow\infty}\int_0^Af(t)\operatorname dt=\pm\infty$

$\Leftrightarrow \forall N\in \mathbb{R},\exists M\in \mathbb{R} s.t. A>M\rightarrow(\int_0^Af(t)\operatorname dt>N\vee \int_0^Af(t)\operatorname dt<N)$

2.$\lim\limits _{A\rightarrow-\infty}\int_A^0f(t)\operatorname dt=\pm\infty$

$\Leftrightarrow \forall N\in \mathbb{R},\exists M\in \mathbb{R} s.t. A<M\rightarrow(\int_0^Af(t)\operatorname dt>N\vee \int_0^Af(t)\operatorname dt<N)$

3.$\forall G(x), ∃a ∈ R \text{ s.t.}G(x) = \int_a^xf(t)\operatorname dt$ (not sure about this one)

$\Leftrightarrow\forall G(x), ∃a ∈ R \text{ s.t.}\forall n \in \mathbb{R}, \forall \varepsilon>0, \exists \delta>0\text{ s.t. } \exists P\in \mathbb{P}$ s.t.

$( \text{$P$ is a partition of [a,n]} \wedge l(P)<\delta)\rightarrow|S(f(t),P)-G(n)|<\varepsilon$

But those doesn't looks like very useful...where should I start?

Any help or hint or suggestion would be appreciated.

  • $\begingroup$ Hint: Consider when $a = \pm\infty$. $\endgroup$
    – Jakobian
    Jul 13 '19 at 21:29

Well, at first, let $f:\mathbb{R}\to\mathbb{R}$ be some continuous function. Then, as you stated, every function of the form: $$G(x)=\int_a^xf(t)dt,$$ is an antiderivative of $f$. Let $F(x)$ be some antiderivative of $f$. Then, we have $F'(x)=f(x)$ for every $x\in\mathbb{R}$. Thus, there exists some constant $c_a\in\mathbb{R}$ such that: $$F(x)=\int_a^xf(t)dt+c_a.$$ Inversely, it is evident that a function of the form $\int_a^xf(t)dt+c$ is an anti-derivative of $f$. So, we have proved the following:

Let $f:\mathbb{R}\to\mathbb{R}$ be a continuous function. Then the set: $$\mathcal{A}:=\left\{\int_a^xf(t)dt+c:a,c\in\mathbb{R}\right\}$$ contains exactly all the anti-derivatives of $f$.

In general, since $f(x)\geq0$, we have that: $$\int_0^xf(t)dt\geq0,\ \forall\ x>0,$$ and,similarly: $$\int_x^0f(t)dt\geq0,\ \forall\ x<0.$$

Also, since $f(x)\geq0$, we get that any anti-derivative of $f$ is increasing.

For the first question, let, W.L.O.G. $$\lim_{x\to+\infty}\int_0^xf(t)dt=L<+\infty.$$ Also, let $$F(x)=\int_0^xf(t)dt.$$ Since $F$ is an anti-derivative of $f$, $F$ is increasing and: $$F(x)\leq L.$$

From the above, we also have that $L\geq0$. Then, the function: $$G(x)=\int_0^xf(t)dt-L-1$$ is an anti-derivative of $f$ with $G(x)\leq-1<0$ for each $x>0$. So, $G$ has no roots, thus, cannot be of the form: $$\int_a^xf(t)dt,$$ since any such function has at least one root ($a$ is always a root).

For the second question, let $G$ be an antiderivative of $f$. Then, $G$ can be written in the form: $$G(x)=\int_a^xf(t)dt+c.$$ We can now do the following trick: $$G(x)=\int_a^xf(t)dt+c=\int_0^xf(t)dt+\underbrace{\int_a^0f(t)dt+c}_{C}=\int_0^xf(t)dt+C.$$

Now, the two given assumptions imply that: $$\lim_{x\to+\infty}G(x)=+\infty\text{ and }\lim_{x\to-\infty}G(x)=-\infty,$$ and, since $G$ is continuous, we get that $G(\mathbb{R})=\mathbb{R}.$ Particularly, this implies that there exists some $x_0\in\mathbb{R}$ such that $G(x_0)=0$, or, equivalently: $$\int_0^{x_0}f(t)dt+C=0\Leftrightarrow C=\int_{x_0}^0f(t)dt.$$ Thus, we have: $$G(x)=\int_0^xf(t)dt+\int_{x_0}^0f(t)dt=\int_{x_0}^xf(t)dt,$$ which was our goal.


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.