I'm trying to solve this limit over "its natural domain": $\Bbb R^2 \cap (x+y \neq 0)$, I suppose. $$ \lim_{(x,y)\to (0,0)}\int_x^y \frac{\arctan{(t^3)}}{x+y} dt\ $$

My attempt:
1. $ \lim_{(x,y)\to (0,0)} f(x,mx) \to 0$ with $(m \neq -1)$
2. $ \lim_{(x,y)\to (0,0)} f(x,x^2) \to 0$

I tried to evaluate the limt "near" the line $(-y,y)$ where there could be some problems:
3. $ \lim_{(x,y)\to (0,0)} f(-y\pm y^2,y) \to 0$ (if anybody wants to see steps ask me and I'll edit the post)

So I was thinking the limits could be $0$. I also tried to graph the function:
enter image description here

and I can't see any "strange curve" which I can exploit to disprove my conjecture.

So I tried to make such an estimation:
* $ |f(x,y| \le g(x,y) $
* $ \lim_{(x,y)\to (0,0)} g(x,y) \to 0$

For first I did: $$ \left|\int_x^y \frac{\arctan{(t^3)}}{x+y} dt\ \right| \le \frac{\pi}{2} \frac{|y-x|}{|x+y|} $$
But the last limit doesn't exist.

Then I tried the change of variables: $|x|=u^2$, $|y|=v^2$, separating the various cases; but I fail when I evaluate the cases $(x>0, y<0)$ and $(x<0, y>0)$. For example, the firts: $$ \left|\int_{u^2}^{-v^2} \frac{\arctan{(t^3)}}{u^2-v^2} dt\ \right| \le -\int_{u^2}^{-v^2} \left|\frac{\arctan{(t^3)}}{u^2-v^2}\right| dt\ \le -\frac{1}{|u^2-v^2|} \int_{u^2}^{-v^2} \left|t^3 \right| dt\ \le \frac{u^8+v^8}{|u^2-v^2|} $$ Again, I think the last limit doesn't exist.

Has anybody some hints to solve the limit?


1 Answer 1


I think you probably need to assume $x,y > 0$, otherwise if $x=-y$ it is infinite. With that assumption, since the derivative of arctan is 1 at 0, you can bound $arctan(t^3)$ by a constant times $t^3$, then just do the integration, and go from there.

  • $\begingroup$ Correct me if I am wrong, but he did do that in the 3rd step of the last integral attempt? $\endgroup$
    – Gareth Ma
    Apr 4, 2020 at 19:44
  • $\begingroup$ Yes, but he had $x<0<y$, which is a case I think you have to exclude. $\endgroup$ Apr 5, 2020 at 5:00
  • $\begingroup$ @Greg I don't think I can exclude other cases except for $x>0, y>0$; infact it is the second question of exercise. I have at least to show the limit doesn't exist at all.. $\endgroup$ Apr 5, 2020 at 8:30
  • 2
    $\begingroup$ Wait, I was wrong, if $x=-y$ then you get $0/0$, I overlooked the fact that the integral is $0$. But what about the following argument? The power series of $arctan(t^3)$ is $\sum_{n=0}^\infty \frac{(-1)^n t^{6n+3}}{2n}$. Integrate term by term to get $\sum_{n=0}^\infty \frac{(-1)^n(y^{6n+4}-x^{6n+4})}{2n(6n+4)(x+y)}$. Then you can factor $(y^{6n+4}-x^{6n+4})$ and cancel the $x+y$ in the denominator. Then you obtain $(x-y)(y^{6n+2} + y^6 x^2 + ... + x^{6n+2})$, and the modulus is $\leq |x-y| (3n+2) \max(|x|,|y|)^{6n+2}$. I think this limit should be $0$, no? $\endgroup$ Apr 5, 2020 at 20:08
  • $\begingroup$ At the end, I found another simpler solution. But, I really can't understand how you have factorized the numerator. Also, can you handle infinite sum into limits? $\endgroup$ Apr 6, 2020 at 8:12

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.