Question about arctan addition

  • Thread starter Thread starter issacnewton
  • Start date Start date
  • Tags Tags
    Addition
Click For Summary

Homework Help Overview

The discussion revolves around proving identities involving the arctangent function, specifically the addition of angles. The original poster presents an attempt to derive the identity for the sum of two arctangents, while also exploring the implications of the domains and ranges of the functions involved.

Discussion Character

  • Exploratory, Conceptual clarification, Mathematical reasoning, Assumption checking

Approaches and Questions Raised

  • The original poster attempts to use the tangent addition formula to derive the identity for arctangents, questioning the validity of their reasoning and the implications of the conditions under which the identity holds.
  • Some participants question the necessity of proving that the product of the inputs is less than one and the implications of the ranges of the arctangent function.
  • Others suggest testing specific values of x and y to explore when the identity holds and when it does not, particularly in relation to the discontinuities of the arctangent function.
  • There is discussion about the need for graphical analysis to understand the behavior of the functions involved, especially when considering cases where the sum exceeds the range of the arctangent function.

Discussion Status

The discussion is ongoing, with participants exploring various cases and attempting to clarify the conditions under which the identities hold. Some guidance has been offered regarding the need to consider the ranges of the functions, but there is no explicit consensus on a complete resolution.

Contextual Notes

Participants note that the arctangent function has a range of (-π/2, π/2) and discuss the implications of this range on the identities being explored. There is also mention of the need to consider cases where the product of the inputs exceeds one and how this affects the validity of the identities.

issacnewton
Messages
1,035
Reaction score
37
Homework Statement
Prove the following identity . If ## f(x) = \arctan(x) + \arctan(y) ## then prove that
$$f(x) =
\begin{cases}
\arctan\bigg( \frac{x+y}{1 - xy} \bigg), \qquad xy < 1 \\
\pi + \arctan\bigg( \frac{x+y}{1 - xy} \bigg), \quad x>0, y > 0, xy > 1 \\
-\pi + \arctan\bigg( \frac{x+y}{1 - xy} \bigg), \quad x < 0, y < 0, xy > 1
\end{cases} $$
Relevant Equations
Equations of inverse tangent functions
Now following is my attempt for the very first one. Let ##\arctan(x) = \alpha## and ## \arctan(y) = \beta ##. So, I have ## x = \tan(\alpha)## and ##y = \tan(\beta)##. Since the domain is restricted to define inverse function, I have that ##\alpha \in (-\pi/2, \pi/2) ## and ##\beta \in (-\pi/2, \pi/2)##. Now, I use the following identity for addition of angles

$$ \tan(\alpha + \beta) = \frac{\tan(\alpha) + \tan(\beta) }{ 1- \tan(\alpha) \tan(\beta) } $$

On the left hand side, we have ##\alpha + \beta ## as the argument for the ##tan## function. But since the domain of the tan function is restricted, we must have that ##\alpha + \beta \in (-\pi/2, \pi/2) ##. This further restricts the values which ##\alpha## and ##\beta## can take. So, we have ##\alpha \in (-\pi/4, \pi/4) ## and ##\beta \in (-\pi/4, \pi/4)##. This means the following

$$ - \frac{\pi}{4} < \alpha < \frac{\pi}{4} $$
$$ - \frac{\pi}{4} < \beta< \frac{\pi}{4} $$

Now, I use the fact that ##\tan## is an increasing function. So, I get ## - 1 < \tan(\alpha) < 1 ## and ## -1 < \tan(\beta) < 1 ##. Which means that ##x \in (-1,1) ## and ##y \in (-1,1) ##. From this, its easy to prove that ## xy < 1## and plugging for x and y, we get the required first identity.

$$ \alpha + \beta = \arctan \Big( \frac{x+y}{1 - xy} \Big) $$
$$ \arctan(x) + \arctan(y) = \arctan \Big( \frac{x+y}{1 - xy} \Big) $$

Is my reasoning right ? For the other two identities, I have no clue how to proceed.
Thanks
 
Physics news on Phys.org
IssacNewton said:
Homework Statement:: Prove the following identity . If ## f(x) = \arctan(x) + \arctan(y) ## then prove that
$$f(x) =
\begin{cases}
\arctan\bigg( \frac{x+y}{1 - xy} \bigg), \qquad xy < 1 \\
\pi + \arctan\bigg( \frac{x+y}{1 - xy} \bigg), \quad x>0, y > 0, xy > 1 \\
-\pi + \arctan\bigg( \frac{x+y}{1 - xy} \bigg), \quad x < 0, y < 0, xy > 1
\end{cases} $$
Relevant Equations:: Equations of inverse tangent functions

Now following is my attempt for the very first one. Let ##\arctan(x) = \alpha## and ## \arctan(y) = \beta ##. So, I have ## x = \tan(\alpha)## and ##y = \tan(\beta)##. Since the domain is restricted to define inverse function, I have that ##\alpha \in (-\pi/2, \pi/2) ## and ##\beta \in (-\pi/2, \pi/2)##. Now, I use the following identity for addition of angles

$$ \tan(\alpha + \beta) = \frac{\tan(\alpha) + \tan(\beta) }{ 1- \tan(\alpha) \tan(\beta) } $$

On the left hand side, we have ##\alpha + \beta ## as the argument for the ##tan## function. But since the domain of the tan function is restricted, we must have that ##\alpha + \beta \in (-\pi/2, \pi/2) ##. This further restricts the values which ##\alpha## and ##\beta## can take. So, we have ##\alpha \in (-\pi/4, \pi/4) ## and ##\beta \in (-\pi/4, \pi/4)##. This means the following

$$ - \frac{\pi}{4} < \alpha < \frac{\pi}{4} $$
$$ - \frac{\pi}{4} < \beta< \frac{\pi}{4} $$

Now, I use the fact that ##\tan## is an increasing function. So, I get ## - 1 < \tan(\alpha) < 1 ## and ## -1 < \tan(\beta) < 1 ##. Which means that ##x \in (-1,1) ## and ##y \in (-1,1) ##. From this, its easy to prove that ## xy < 1## and plugging for x and y, we get the required first identity.

$$ \alpha + \beta = \arctan \Big( \frac{x+y}{1 - xy} \Big) $$
$$ \arctan(x) + \arctan(y) = \arctan \Big( \frac{x+y}{1 - xy} \Big) $$

Is my reasoning right ? For the other two identities, I have no clue how to proceed.
Thanks
I don't see where you used ##xy < 1## here.

Do you understand why ##f(x)## is not always equal to ##g(x) = \arctan(\frac{x + y}{1 - xy})##, depending on ##x## and ##y##?
 
Well ##x## and ##y## are in ##\mathbb{R}## and since I reasoned that ##x, y \in (-1,1)##, we can take various cases of ##x## and ##y## and then prove that ##xy < 1##. I thought we are supposed to prove this.
 
IssacNewton said:
Well ##x## and ##y## are in ##\mathbb{R}## and since I reasoned that ##x, y \in (-1,1)##, we can take various cases of ##x## and ##y## and then prove that ##xy < 1##. I thought we are supposed to prove this.

If ##xy < 1##, you could have ##x = \frac 1 3, \ ##y = 2##.

What about my main question:

PeroK said:
Do you understand why ##f(x)## is not always equal to ##g(x) = \arctan(\frac{x + y}{1 - xy})##, depending on ##x## and ##y##?
 
No, I am confused about the question.
 
IssacNewton said:
No, I am confused about the question.

Try some numbers. See what you get.
 
Ok, so the domain of ##\arctan## should be ##(-\pi/2, \pi/2)##, right ? I will try some values
 
IssacNewton said:
Ok, so the domain of ##\arctan## should be ##(-\pi/2, \pi/2)##, right ? I will try some values
The domain of ##\arctan## is ##\mathbb R##. The range is ##(-\pi/2, \pi/2)##.

It's a tricky question.
 
I tried few values of ##x,y## so that ##xy < 1## on WolframAlpha and yes in this case, we have that
$$ \arctan(x) + \arctan(y) = \arctan\Big( \frac{x+y}{1-xy} \Big) $$

So, how do I proceed proving this ? How would my argument change ?
 
  • #10
IssacNewton said:
I tried few values of ##x,y## so that ##xy < 1## on WolframAlpha and yes in this case, we have that
$$ \arctan(x) + \arctan(y) = \arctan\Big( \frac{x+y}{1-xy} \Big) $$

So, how do I proceed proving this ? How would my argument change ?

What I meant was to try to find values of ##x, y## where this is not true and see why. The confusing thing, surely, is why that equation is not true for all ##x, y##. Why do you sometimes have to add ##\pm \pi##?
 
  • #11
So, for ##x=10## and ##y = 1/3##, I get ##\arctan(x) + \arctan(y) = 1.793##, which is greater than ##\pi/2## , so this is is beyond the range of arctan function and ##\arctan((x+y)/(1-xy)) = -1.35## and this is within the range of arctan function.
 
  • #12
IssacNewton said:
So, for ##x=10## and ##y = 1/3##, I get ##\arctan(x) + \arctan(y) = 1.793##, which is greater than ##\pi/2## , so this is is beyond the range of arctan function and ##\arctan((x+y)/(1-xy)) = -1.35## and this is within the range of arctan function.

... and the difference is ##1.793 - (-1.35) = 3.14 = \pi##.

What I would do first is draw the graph of ##\tan## and show various possibilities for ##\alpha, \beta## and ##\alpha + \beta##.

Because of the nature of the ##\arctan## function, this isn't going to come out purely algebraically. You're going to need the graph of ##\tan##.
 
  • #13
So, is there no algebraic proof of this ?
 
  • #14
IssacNewton said:
So, is there no algebraic proof of this ?

I said "purely" algebraically. The problem is knowing when to add ##\pm \pi##. That's not going to come out very easily without using the graph.
 
  • #15
Ok. Thanks
 
  • #16
If math analysis is allowed then one can consider derivative ##\frac{\partial f}{\partial x}##
 
  • #17
IssacNewton said:
is there no algebraic proof of this ?
The difficulty is that the arctan function is discontinuous. You know that ##\arctan(\frac{x+y}{1-xy})## will produce the right value modulo ##\pi##. It remains to show that for the second case it produces a value in the range ##(-3\pi/2, -\pi/2)##, etc.
wrobel said:
If math analysis is allowed then one can consider derivative ##\frac{\partial f}{\partial x}##
How does that help with the question asked by the OP, namely, dealing with the different ranges?
 
  • #18
haruspex said:
How does that help with the question asked by the OP, namely, dealing with the different ranges?
Let us regard ##y## as a parameter. We have two functions
$$f(x)=\arctan x+\arctan y,\quad g(x)=\arctan\Big(\frac{x+y}{1-xy}\Big).$$
Assume for example that ##y>0,\quad x>1/y##.
We obviously have ##g'(x)=f'(x)##. And these derivatives are continuous for ##x>1/y##. Thus for such ##x## it follows that ##g(x)=f(x)+const.## The constant may depend on the parameter ##y##.
Passing to a limit as ##x\to 1/y+## we get $$-\pi/2=\arctan(1/y)+\arctan y+const.$$
On the other hand
$$\frac{d}{dy}(\arctan(1/y)+\arctan y)=0$$ and thus (chek for ##y=1##) we get
$$\arctan(1/y)+\arctan y=\pi/2\Longrightarrow const=-\pi$$
That's all
 

Similar threads

  • · Replies 14 ·
Replies
14
Views
2K
  • · Replies 10 ·
Replies
10
Views
2K
Replies
2
Views
3K
  • · Replies 20 ·
Replies
20
Views
3K
Replies
15
Views
2K
  • · Replies 3 ·
Replies
3
Views
3K
  • · Replies 21 ·
Replies
21
Views
4K
  • · Replies 9 ·
Replies
9
Views
3K
  • · Replies 11 ·
Replies
11
Views
3K
  • · Replies 15 ·
Replies
15
Views
4K