I have the following problem:
Suppose that $f$ is continuous on $[a,b]$ and suppose that for all $x \in [a,b]$, $f(x) \geq 0$ and $f(x)\leq \int_a^x f(t)dt$. Show that $f(x)=0$ for all $x \in [a,b]$.
By integrating repeatedly, we obtain: $$f(x) \leq \int_a^x f(t_1)dt_1 \leq \cdots \leq \int_a^x \int_a^{t_1} \cdots \int_a^{t_n} f(s)dsdt_n \cdots dt_1$$ Thus, if $M_x$ denotes the maximum of $f$ on the interval $[a,x]$, we have that $f(x) \leq M_x(x-a)^n$. Taking $n \to \infty$, we find that $f(x)=0$ on the interval $\left[a,a+1\right)$.
I am now having trouble proving that $f(x)=0$ on $[a,b]$, although I believe induction is involved. Any help would be appreciated!
To continue your work we have $f(x)=0$ for all $x\in[a,a+1]$ but this implies that: $$f(x)\leq\int_{a}^xf(t)dt= \int_{a+1}^xf(t)dt$$
and here you apply the same reasoning and you get $f(x)=0$ on $[a+1,a+2]$ again, and repeating the process you get the result.
Another method :
we have $f(a)=0$ now let $m=\max\{x\in[a,b]\mid\forall y\leq x f(y)=0\}$,
Assume that $m<b$ there exist $\epsilon$ such that $f(m+\epsilon)> 0$ for some $0<\epsilon< 1$ and $\epsilon \leq b-m $, and let $f(t)=\max_{x\in[m,m+\epsilon]}(f(x))$ we have:
$$0 \leq f(t)\leq \int_{a}^tf(t)dt=\int_{m}^tf(t)dt\leq \int_{m}^{m+\epsilon}f(t)dt\leq \epsilon f(t)$$
thus $f(t)=0$, this is absurd because $f(t)=0$ implies $f(m+\epsilon)=0$
Finally $m=b$ and $f=0$.