Author 
Message 
Bart science forum beginner
Joined: 07 Jul 2005
Posts: 39

Posted: Wed Jun 21, 2006 8:36 am Post subject:
trigonometric identity



In order to understand a certain step in a proof, I would like to
proove the following identity:
exp(Pi*I*N*theta)*sum_{k=1}^{N}exp(2*Pi*I*k*theta)
= sin(Pi*N*theta)cot(Pi*theta)
where I is the imaginary unit. However, I can't get there
Can anybody give me some hints?
Thanks,
Bart 

Back to top 


Robert Low science forum Guru
Joined: 01 May 2005
Posts: 1063

Posted: Wed Jun 21, 2006 8:42 am Post subject:
Re: trigonometric identity



Bart wrote:
Quote:  In order to understand a certain step in a proof, I would like to
proove the following identity:
exp(Pi*I*N*theta)*sum_{k=1}^{N}exp(2*Pi*I*k*theta)
= sin(Pi*N*theta)cot(Pi*theta)
where I is the imaginary unit. However, I can't get there
Can anybody give me some hints?

You might think of the sum as sum_{k=1}^N (exp(2pi i theta))^k
and use the formula for the sum of a (finite) geometric series. 

Back to top 


Bart science forum beginner
Joined: 07 Jul 2005
Posts: 39

Posted: Wed Jun 21, 2006 11:22 am Post subject:
Re: trigonometric identity



On 20060621, Robert Low <mtx014@coventry.ac.uk> wrote:
Quote: 
Bart wrote:
In order to understand a certain step in a proof, I would like to
proove the following identity:
exp(Pi*I*N*theta)*sum_{k=1}^{N}exp(2*Pi*I*k*theta)
= sin(Pi*N*theta)cot(Pi*theta)
where I is the imaginary unit. However, I can't get there
Can anybody give me some hints?
You might think of the sum as sum_{k=1}^N (exp(2pi i theta))^k
and use the formula for the sum of a (finite) geometric series.

Hmm... strange... I have used the exponential sum formula (2) from
http://mathworld.wolfram.com/ExponentialSumFormulas.html (note
that my sum runs from 1 to N and the sum on this site runs from 0
to N1, so be careful here!) and I arrive at
exp(Pi*I*N*theta)*sum_{k=1}^{N}exp(2*Pi*I*k*theta)
= sin(Pi*N*theta)*cot(Pi*theta) + I*sin(Pi*N*theta)
^^^^^^^^^^^^^^^^^
So apparently i have an extra term there, namely I*sin(Pi*N*theta).
Did I make a mistake, or is the original equality as I posted it
not valid?
Thanks,
Bart 

Back to top 


Robert Low science forum Guru
Joined: 01 May 2005
Posts: 1063

Posted: Wed Jun 21, 2006 11:37 am Post subject:
Re: trigonometric identity



Bart wrote:
Quote:  On 20060621, Robert Low <mtx014@coventry.ac.uk> wrote:
Bart wrote:
In order to understand a certain step in a proof, I would like to
proove the following identity:
exp(Pi*I*N*theta)*sum_{k=1}^{N}exp(2*Pi*I*k*theta)
= sin(Pi*N*theta)cot(Pi*theta)
You might think of the sum as sum_{k=1}^N (exp(2pi i theta))^k
and use the formula for the sum of a (finite) geometric series.
... I arrive at
exp(Pi*I*N*theta)*sum_{k=1}^{N}exp(2*Pi*I*k*theta)
= sin(Pi*N*theta)*cot(Pi*theta) + I*sin(Pi*N*theta)

I get that too, so either we both made similar mistakes or
the original statement is wrong. Maybe it was just the real
part of the expression you were meant to compute? 

Back to top 


Bart science forum beginner
Joined: 07 Jul 2005
Posts: 39

Posted: Wed Jun 21, 2006 12:02 pm Post subject:
Re: trigonometric identity



On 20060621, Robert Low <mtx014@coventry.ac.uk> wrote:
Quote: 
I get that too, so either we both made similar mistakes or
the original statement is wrong. Maybe it was just the real
part of the expression you were meant to compute?

For as far as I understand the proof, I do not only need the real
part here. Let me try to show more context...
The paper defines:
e(x) = exp(2*Pi*I*x)
A_q = (2*q+1)!/((q!)^2)
K_q(theta) = 1/N * sum_{k=1}^{N} A_q*(k/N)^q*(1k/N)^q*e(k*theta)
And then states that
e^{Pi*I*N*theta}K_0(theta) = 1/N*(sin(Pi*N*theta)*cot(Pi*theta))
For as far as I interpret all this, the extra term is somehow
forgotten here... right?
Best wishes,
Bart 

Back to top 


Robert Low science forum Guru
Joined: 01 May 2005
Posts: 1063

Posted: Wed Jun 21, 2006 1:12 pm Post subject:
Re: trigonometric identity



Bart wrote:
Quote:  On 20060621, Robert Low <mtx014@coventry.ac.uk> wrote:
I get that too, so either we both made similar mistakes or
the original statement is wrong. Maybe it was just the real
part of the expression you were meant to compute?
The paper defines:
e(x) = exp(2*Pi*I*x)
A_q = (2*q+1)!/((q!)^2)
K_q(theta) = 1/N * sum_{k=1}^{N} A_q*(k/N)^q*(1k/N)^q*e(k*theta)
And then states that
e^{Pi*I*N*theta}K_0(theta) = 1/N*(sin(Pi*N*theta)*cot(Pi*theta))
For as far as I interpret all this, the extra term is somehow
forgotten here... right?

Beats me. As I said, it's always possible I messed up the algebra.
If so, I'm sure somebody will point it out fairly soon. 

Back to top 


Google


Back to top 


