# Show integral limit - easy I think

I'm going to show this:

$$\lim_{x\rightarrow ∞}\,\,\,\int_0^x \! \frac{\arctan{t}}{t} \, \mathrm{d} t = ∞$$

We're working on L.Hopital and integrals... I guess it's very easy. But I struggle... Don't need to solve it for me if you give just give me guidelines. Thank you so much!

-
$$\arctan\,u=\frac{\pi}{2}-\arctan\frac1{u}$$ might be useful... – J. M. Oct 27 '11 at 15:05

Hint: $\arctan(t)\ge\frac{\pi}{4}$ when $t\ge1$
Here's one possible hint: $\arctan t \to \pi/2$, so $\arctan t \ge 1$ from some point on (say for $t\ge a$). What can you say about $\lim_{x\to\infty} \int_a^x \frac{dt}{t}$?
@IQlessThan70 That the function never becomes 0 is not enough to show that the integral diverges. But you can easily find $\int_0^x \mathrm{constant}/t\;dt$. – Michael Hardy Oct 27 '11 at 17:09
Seriously? Can I just set that arctan t to a constant?! ln to doesn't work. Can I change the integral to: $$\lim_{x\rightarrow ∞}\,\,\,\int_1^x \! \frac{c}{t} \, \mathrm{d} t = ∞$$ ? Thank you so much! – IQlessThan70 Oct 27 '11 at 18:57
@IQlessThan70 Just to be clear, you're not really setting $\arctan(t)$ to a constant, you're using the fact that eventually, for x large enough (say $x>b$), $\arctan(t)>1$. Now you have that $arctan(t) / t > 1/t$, and you can integrate both sides from $b$ to $x$. At this point, you're using the comparison theorem to show that since $\int_1^{\infty} \frac {dt} t$ diverges, the integral you're interested in also diverges. Geometrically this is because the area under $\arctan(t)/t$ has larger area than the area under $1/t$ between $b$ and any $x>b$. – process91 Oct 27 '11 at 23:17
@IQlessThan70 You can also prove it rigorously, "from the ground up". Taking off from what I said in my earlier comment, you integrate the inequality from $b$ to $x$: $$\int_b^x \frac {dt} t < \int_b^x \frac {\arctan(t)} t dt$$ $$\implies \log (x) - \log(b) < \int_b^x \frac {\arctan(t)} t dt$$ Since the left side is unbounded above as $x\to\infty$, you can use the definition of limit and prove the result by contradiction. – process91 Oct 28 '11 at 2:36