# Differentiating arctan

#### ACLerok

http://www.eden.rutgers.edu/~cjjacob/images/arctan.gif [Broken]

I was given this problem yesterday. It's asking me to differentiate a function containing arctan. Should I find a common denominator and combine them into one term, should I just differentiate each term separately? Please I need help with this thing! I've been told it comes out to 0 but I'm not sure if he is correct or not. Thanks for replying if you choose to!

Last edited by a moderator:
Related Introductory Physics Homework Help News on Phys.org

#### Doc Al

Mentor
ACLerok said:
Should I find a common denominator and combine them into one term, should I just differentiate each term separately?
Should you try to combine the two arctan terms into one? Yikes!

No, just differentiate each term separately. If you know how to differentiate arctan(x) and how to use the chain rule, this should be a snap.

#### ShawnD

According to maple, f'(x) is this

$$\left( 1+{x}^{2} \right) ^{-1}-{\frac {1}{{x}^{2} \left( 1+{x}^{-2} \right) }}$$

I can't help with how to get that though; I've never dealt with arctan before.

It simplifies to 0.

Last edited:

Staff Emeritus
Gold Member

#### Janitor

Draw a right triangle with a base of length 1 and a side of length x. Note that one of the acute angles is equal to arctan(x) and the other acute angle is equal to arctan(1/x). Since the three interior angles of any triangle add up to pi radians, and since the remaining angle in this triangle is pi/2, then the sum arctan(x) + arctan(1/x) has to equal the constant value pi - pi/2 = pi/2. Its derivative is thus zero.

$$y = \arctan{x}$$
$$\tan{y} = x$$
$$\frac{dy}{dx}\sec^2{y} = 1$$
$$\frac{dy}{dx} = \cos^2{y}$$

Draw a triangle based on the second line, tan(y) = x. This means that the opposite side is x and the adjacent side is 1, giving the hypotenuse a value of $\sqrt{x^2 + 1}$. This means that cos(y) will be 1 over the hypotenuse, i.e.

$$\cos{y} = \frac{1}{\sqrt{x^2 + 1}}$$

which means that our derivative is

$$\frac{dy}{dx} = \frac{1}{x^2 + 1}$$
$$\frac{d}{dx}\Big(\arctan{x}\Big) = \frac{1}{x^2 + 1}$$

The rest follows from the chain rule.

#### ACLerok

$$y = \arctan{x}$$
$$\tan{y} = x$$
$$\frac{dy}{dx}\sec^2{y} = 1$$
$$\frac{dy}{dx} = \cos^2{y}$$

Draw a triangle based on the second line, tan(y) = x. This means that the opposite side is x and the adjacent side is 1, giving the hypotenuse a value of $\sqrt{x^2 + 1}$. This means that cos(y) will be 1 over the hypotenuse, i.e.

$$\cos{y} = \frac{1}{\sqrt{x^2 + 1}}$$

which means that our derivative is

$$\frac{dy}{dx} = \frac{1}{x^2 + 1}$$
$$\frac{d}{dx}\Big(\arctan{x}\Big) = \frac{1}{x^2 + 1}$$

The rest follows from the chain rule.

ok i think I can see what you have done. can you explain how you got from tan(y)=x to the line below it? did you just differentiate both sides with respect to y?

And then at the end are you saying to just substitute the equation for the derivative of arctan into the orginal equation?

Thanks for all your help guys!

Last edited:

Yes. The third line is reached by differentiating the second with respect to x.

The derivative of arctan(1/x) is found in a similar manner, or you can differentiate it using a substitution and the chain rule. Both methods require just about the same amount of work.

#### Janitor

I had a buddy in high school who liked to use non-calculus techniques to get the answer to homework problems that the calculus teacher assigned. The kid knew darn well that the teacher would not give credit for getting the right answer by "devious" means, but that didn't stop him from doing that. I think he had some sort of persecution complex, and he loved to wail to anyone who would listen about how unfair the teacher was.

#### Janitor

Thanks, Cookiemonster! But I had a guilty conscience since I figured the teacher wouldn't give him any credit for it, since it doesn't teach him how to do derivatives.

Haha, why not? He still has to differentiate pi/2. =]

True!

. .

#### ACLerok

Yes. The third line is reached by differentiating the second with respect to x.

The derivative of arctan(1/x) is found in a similar manner, or you can differentiate it using a substitution and the chain rule. Both methods require just about the same amount of work.

ok so the derivative of arctan x is 1/[(x^2)+1] but what is the derivative of arctan(1/x) do i just plug 1/x into 1/[(x^2)+1]? it seems like the logical thing to do but then again, i may be totally wrong.. Thanks.

As I said in my previous post, you can figure it out using the same method as for arctan(x), i.e.

tan(y) = 1/x

differentiate both sides implicitly to get

dy/dx = -cos^2(y)/x^2

then we build a triangle based on tan(y) = 1/x, which means the opposite side is 1 and the adjacent side is x, so the hypotenuse is sqrt(x^2 + 1), so the cos^2(y) term is x/sqrt(x^2 + 1), meaning

dy/dx = 1/x^2 * -x^2/(1+x^2) = -1/(x^2 + 1)

Alternatively, you could have made a substitution u = 1/x into the derivative, so we have

d/dx(arctan(u)) = df(u)/du * du/dx = 1/(1 + 1/x^2)*-1/x^2

which can be shown to be equivalent with a little bit of algebra.

#### ACLerok

As I said in my previous post, you can figure it out using the same method as for arctan(x), i.e.

tan(y) = 1/x

differentiate both sides implicitly to get

dy/dx = -cos^2(y)/x^2

then we build a triangle based on tan(y) = 1/x, which means the opposite side is 1 and the adjacent side is x, so the hypotenuse is sqrt(x^2 + 1), so the cos^2(y) term is x/sqrt(x^2 + 1), meaning

dy/dx = 1/x^2 * -x^2/(1+x^2) = -1/(x^2 + 1)

Alternatively, you could have made a substitution u = 1/x into the derivative, so we have

d/dx(arctan(u)) = df(u)/du * du/dx = 1/(1 + 1/x^2)*-1/x^2

which can be shown to be equivalent with a little bit of algebra.

so the derivative of the problem simplifies to -1/(x^2 + 1)? and that equals 0?

Yes. Both Janitor's solution and brute force show the solution to be 0. You wanted it to be 0 in the first place anyway, didn't you?

#### ACLerok

Yes. Both Janitor's solution and brute force show the solution to be 0. You wanted it to be 0 in the first place anyway, didn't you?

alright, i understand everything now except this one line:
dy/dx = 1/x^2 * -x^2/(1+x^2) = -1/(x^2 + 1)

how did you get this? i got up to dy/dx = -x^2+1/x^2
whatd you do to get that line? i tried using the quotient rule to differentiate it but i got 2x/x^4 but that is incorrect.

i never wanted the answer to be zero, i was just told that it is. i guess. thanks alot!

Last edited:

#### Doc Al

Mentor
ACLerok said:
so the derivative of the problem simplifies to -1/(x^2 + 1)? and that equals 0?
That's the derivative of the second term in your original expression. Add that to the derivative of the first term and the sum will equal zero.

#### Doc Al

Mentor
ACLerok said:
how did you get this? i got up to dy/dx = -x^2+1/x^2
whatd you do to get that line? i tried using the quotient rule to differentiate it but i got 2x/x^4 but that is incorrect.
I assume you can differentiate arctan(x)? That will be the first term of your answer.

Now use the chain rule to differentiate arctan(u), where u = 1/x. (What's the derivative of 1/x?) That will be the second term making up your answer.

#### ACLerok

for differentiating arctan 1/x, i got up to drawing the triangle to find cos(y) and i got dy/dx = -cos^2(y)/x^2. know what do i do with these two piece of information? do i just substitute sqrt(x^2+1) into dy/dx = -cos^2(y)/x^2 or what? please this is due tomorrow and i dont know how to finish it!

cos(y) can be expressed in terms of x. What does the triangle you drew tell you it should be?

#### ACLerok

sqrt(x^2+1). i dont know how you differentiated it. can you just explain the steps you took?

Differentiated what? I had two entire posts on the method to differentiate arctan(x) and arctan(1/x). Which parts are giving you trouble?

#### ACLerok

i dont understand what you did here and how you got this:
dy/dx = 1/x^2 * -x^2/(1+x^2) = -1/(x^2 + 1)

this is from differentiating arctan(1/x)
i know that you just substituted sqrt(x^2+1) into dy/dx = -cos^2(y)/x^2 but I have no clue how you went about finding the derivative or which method(s) you used. (quotient rule/chainrule/both/etc.) i know i'm giving you a hard time but calc isnt my piece of cake. sorry.

### Physics Forums Values

We Value Quality
• Topics based on mainstream science
• Proper English grammar and spelling
We Value Civility
• Positive and compassionate attitudes
• Patience while debating
We Value Productivity
• Disciplined to remain on-topic
• Recognition of own weaknesses
• Solo and co-op problem solving