This problem is taken from an exam to acced to PhD at my University in 2012.
Find the maximum on $[0,1]$ of the function defined by
$$\tag{1}f(x) =\int_0^x \sqrt{t^4+(1-x)^2(x-t)^2}\, \mathrm d t$$
Clearly $f(x) \ge 0$ for every $x \in [0,1]$, $f(0)=0$ and $f(1) = \frac{1}{3}$.
My attempt
My first idea was to eliminate the $x$ on the extrema, so with substitution $u = \frac{t}{x}$, we get
$$ f(x) = x \int_0^1 \sqrt{ x^2u^4 + (1-x)^2(1-u)^2}\, \mathrm d u $$
But this seems not very promising.
Another idea that came to my mind was to use the Leibniz integral formula in formula $(1)$:
$$ f'(x) = x^2 + \int_0^x \frac{\partial }{\partial x} \sqrt{t^4+(1-x)^2(x-t)^2} \operatorname \, \mathrm d t $$
But I believe there is a simpler way, which I cannot find.
Could you provide me any help?
I believe that your idea to use the Leibniz integral formula was exactly right and also simpler than you may have realized.
Proceeding with that idea, we have:
$$f'(x) = x^2 + \int_0^x \frac{(1-x)(x-t)(1+t)}{\sqrt{t^4 + (1-x)^2(x-t)^2}} \, dt.$$
The integrand should always be non-negative for any $x \in [0,1]$. Hence, the integral is guaranteed to be non-negative. Similarly, $x^2$ is non-negative, so $f'(x)$ is non-negative. This guarantees that, for all $x \in [0,1]$, $f(x)$ is non-decreasing. Hence, the maximum must occur precisely at $x = 1$. Thus, since you already found that $f(1) = \frac{1}{3}$, the maximum is ⅓.