# Homework Help: Integral problem help

1. Apr 3, 2009

### dirk_mec1

1. The problem statement, all variables and given/known data

Evaluate

$$\int_0^1\frac{\arctan t}{t\sqrt{1-t^2}}dt$$

3. The attempt at a solution
I suspect I need to subsitute something but what?

2. Apr 3, 2009

### CompuChip

Re: integral

I'm getting a little lost in trig identities, so I haven't looked into it very much. But have you tried t = tan(x) to get rid of the arctan, or t = sin(x) or cos(x) to get rid of the square root ?

3. Apr 3, 2009

### dirk_mec1

Re: integral

I've tried your suggestions and failed to get an integral which I can solve...

4. Apr 3, 2009

### n!kofeyn

Re: integral

I think I have a solution. I may be doing it by a method more complicated than required, but sometimes these integrals refuse to be done easily. I will give hints, and I can give more details if you need them.

With integrals that contain this type of square root on the bottom, you will almost always do a trigonometric substitution. The two identities that are useful are $cos^2\theta + \sin^2\theta = 1[/tex] and $1+\tan^2\theta=\sec^2\theta$. Since we have $$1-t^2$, the first identity is what we want to use. Let $t=\sin\theta$ so that you won't get negative signs when differentiating. Do this substitution and then simplify. Go ahead and post the integral that you get after this, and then we'll move on. 5. Apr 3, 2009 ### CompuChip Re: integral Hmm the answer comes out so nicely one thinks there must be some good substitution. But I'm seeing your problem: you are just two steps away from the solution but whatever you try seems to bring you just one step closer but no more. Actually I tried calculating the integral with Mathematica, the funny thing is that it gives an exact answer precisely for boundaries 0 and 1 and not, for example for 1/2 to 1 or 0 to 2. So probably there is a sneaky argument in there somewhere, although I am unable to see where the answer is coming from, except possible from the substitution t = sinh(x) but I don't see how that would help. In any case, I don't really know how to help you at this point, I've run out of possible ways to solve it. I hope someone else might be able to help you. n!kofeyn is making me curious, I'm following closely :) [/edit] Last edited: Apr 3, 2009 6. Apr 3, 2009 ### snipez90 Re: integral Hmmm I also pursued n!kofeyn's first step but I'm lost on how to deal with arctan(sinx). I can work with the limits of integration but there must be another clever step (or maybe easy) that I'm missing. I'll look at this later. 7. Apr 3, 2009 ### FedEx Re: integral Finally i scored one over a Homework Helper. Phew! That was good. Take [tex]t= tan\theta$$

Bring the entire integral in terms of sines and cosines.

Then substitute $$cos2\theta = {\alpha}^2$$

And then solve it by partial fractions.

PS Resolving to partial is quite tricky. And even after that there are more substitutions to come

8. Apr 4, 2009

### dirk_mec1

Re: integral

Ok I got (leaving out constants and integration limits):

$$\int \frac{\theta} { \sin( \theta) \sqrt{ \cos( 2 \theta)}}\ \mbox{d} \theta$$

9. Apr 4, 2009

### FedEx

Re: integral

Its cos(2Theta) and not cos(theta)

Oh crap!!!! I skipped a step.

$$cos2\theta = {\alpha}^2$$

You will have to apply parts before the above substitution. So that we will get rid of the $$\theta$$

Last edited: Apr 4, 2009
10. Apr 4, 2009

### dirk_mec1

Re: integral

Yes I've editted it. Now your substitution alpha^2 =cos (2* theta) is nasty are you sure this is right?

11. Apr 4, 2009

### FedEx

Re: integral

After doing that bring the entire term in terms of alpha. You will get

$$\int\frac{1}{(\frac{1-{\alpha}^2}{\alpha})(\sqrt{{\alpha}^2 +1})}$$

Do this with partial fractions.

PS I have not written the 2's and root 2's coming in the equation

12. Apr 4, 2009

### dirk_mec1

Re: integral

The theta = 0.5*cos^-1(alpha^2) where's that term?

13. Apr 4, 2009

### CompuChip

Re: integral

You get rid of it in the integration by parts, I presume.

14. Apr 4, 2009

### dirk_mec1

Re: integral

In the IBP: integrate w.r.t. to what?

15. Apr 4, 2009

### Cyosis

Re: integral

I am pretty sure FedEx has made a mistake integrating by parts and as a result his method seems to lead to a dead end. CompuChip's analysis with mathematica should give us a clue that there exists no primitive to the integral. However there is a smart trick to get an exact value for the definite integral from 0 to 1.

First we define the function $$f(a) = \int_0^1 \frac{\arctan(a t)}{t \sqrt{1-t^2}}$$ with $$a \geq 0$$. Then take the derivate of f(a) with respect to a and use the substitution $$t=\frac{1}{\sqrt{1+u^2}}$$.

\begin{align*} f'(a) & = \int_0^1 \frac{1}{a^2 t^2+1} \frac{1}{\sqrt{1-t^2}}\,dt \\ & = - \int_{\infty}^0 \frac{1}{(\frac{a^2}{1+u^2}+1) \sqrt{1 - \frac{1}{1+u^2}}} \frac{u}{(1+u^2)^{\frac{3}{2}}}\,du \\ &= \int_0^{\infty} \frac{du}{a^2+1+u^2} \end{align*}

You should be able to rearrange the integrand so you recognize the arctangent. Note that we now have f'(a). Now try to find the answer of your original integral with a=1.

Last edited: Apr 4, 2009
16. Apr 4, 2009

### FedEx

Re: integral

Precisely

17. Apr 4, 2009

### FedEx

Re: integral

With respect to alpha.

18. Apr 4, 2009

### Cyosis

Re: integral

FedEx could you show us how you did the integration by parts. I have tried to get rid of $$\theta$$ as you suggest, but it doesn't seem to lead to a manageable integral.

19. Apr 4, 2009

### dirk_mec1

Re: integral

I agree, Fedex you've made mistake somewhere.

20. Apr 4, 2009

### dirk_mec1

Re: integral

Nice! I'll look carefully at your derivation.