BrandonS15
Badges: 14
Rep:
?
#1
Report Thread starter 3 weeks ago
#1
Hello,
I’ve found the Fourier series for e^x and now I realise I can form two closed-form values for two infinite summations if I let x=0 and x=pi (sin(xn) = 0 for all n) For the case x = pi I’m getting the following result:
Name:  49C1B86C-0A62-4D7C-B118-8A204AAD3BB0.jpg.jpeg
Views: 8
Size:  25.6 KB

I checked the summation from 1 to inf of 1/(1+n^2) on Wolfram and it gives me 1.07 something while my value is 0.5 something,
I’m not sure where I went wrong, I feel like there is a restriction to what values I can put for x (due to periodic) but I’m not sure.
Any help is appreciated, many thanks
0
reply
mqb2766
Badges: 19
Rep:
?
#2
Report 3 weeks ago
#2
What is the actual function you're finding the Fourier series of?

You realise that the Fourier series does not converge pointwise at the jumps (discontinuities), which is what you may be asking about?
https://en.m.wikipedia.org/wiki/Gibbs_phenomenon
Last edited by mqb2766; 3 weeks ago
0
reply
BrandonS15
Badges: 14
Rep:
?
#3
Report Thread starter 3 weeks ago
#3
(Original post by mqb2766)
What is the actual function you're finding the Fourier series of?

You realise that the Fourier series does not converge pointwise at the jumps (discontinuities), which is what you may be asking about?
https://en.m.wikipedia.org/wiki/Gibbs_phenomenon
Yes sorry, it’s of e^x and I’m evaluating at x=pi because like x=0 It gets rid of the second infinite sum of sines
Last edited by BrandonS15; 3 weeks ago
0
reply
BrandonS15
Badges: 14
Rep:
?
#4
Report Thread starter 3 weeks ago
#4
I’m evaluating at x = pi because (like x=0) it removed the other infinite summation with sines so just like x=0 I’m trying to use x=pi to form the infinite summation of (1/(1+n^2)) as shown in the photo
0
reply
mqb2766
Badges: 19
Rep:
?
#5
Report 3 weeks ago
#5
(Original post by BrandonS15)
Yes sorry, it’s of e^x and I’m evaluating at x=pi because like x=0 It gets rid of the second infinite sum of sines
What is the periodic/repetition interval?
-pi to pi
or
0 to pi
Or ...
Last edited by mqb2766; 3 weeks ago
0
reply
BrandonS15
Badges: 14
Rep:
?
#6
Report Thread starter 3 weeks ago
#6
(Original post by mqb2766)
What is the periodic/repetition interval?
-pi to pi
or
0 to pi
Or ...
-pi to pi as I did in the integrals for computing the Fourier coefficients
0
reply
mqb2766
Badges: 19
Rep:
?
#7
Report 3 weeks ago
#7
(Original post by BrandonS15)
-pi to pi as I did in the integrals for computing the Fourier coefficients
So at x=pi you're evaluating it at a point where the series does not converge to e^x?
Last edited by mqb2766; 3 weeks ago
0
reply
BrandonS15
Badges: 14
Rep:
?
#8
Report Thread starter 3 weeks ago
#8
(Original post by mqb2766)
So at x=pi you're evaluating it at a point where the series does not converge to e^x?
How do I know the series doesn’t converge to e^x at x = pi but does at x = 0?
0
reply
mqb2766
Badges: 19
Rep:
?
#9
Report 3 weeks ago
#9
(Original post by BrandonS15)
How do I know the series doesn’t converge to e^x at x = pi but does at x = 0?
See the link in #2. It does not converge at x=(2k+1)pi for this problem (the discontinuities) where k is an integer.
Last edited by mqb2766; 3 weeks ago
0
reply
DFranklin
Badges: 18
Rep:
?
#10
Report 3 weeks ago
#10
It *will* converge to the average of the values at -pi, pi; whether you're allowed to quote that in your course I don't know.

Edit: although f(x)+f(-x) will be continuous at pi and so that series will converge, so it's not hard to prove if you need to.
Last edited by DFranklin; 3 weeks ago
0
reply
davros
  • Study Helper
Badges: 16
Rep:
?
#11
Report 3 weeks ago
#11
(Original post by BrandonS15)
I’m evaluating at x = pi because (like x=0) it removed the other infinite summation with sines so just like x=0 I’m trying to use x=pi to form the infinite summation of (1/(1+n^2)) as shown in the photo
What's the actual problem you've been asked to solve here? Can you upload a picture of the question?
0
reply
BrandonS15
Badges: 14
Rep:
?
#12
Report Thread starter 3 weeks ago
#12
(Original post by davros)
What's the actual problem you've been asked to solve here? Can you upload a picture of the question?
No problem, I’m just experimenting and I thought I could plug pi into the Fourier series of e^x to obtain and infinite series in closed form but I don’t fully understand why it’s discontinuous at x = pi on the -pi to pi interval
0
reply
BrandonS15
Badges: 14
Rep:
?
#13
Report Thread starter 3 weeks ago
#13
(Original post by DFranklin)
It *will* converge to the average of the values at -pi, pi; whether you're allowed to quote that in your course I don't know.

Edit: although f(x) f(-x) will be continuous at pi and so that series will converge, so it's not hard to prove if you need to.
How do you know it’s continuous or not? e^x is continuous for all x
Last edited by BrandonS15; 3 weeks ago
0
reply
mqb2766
Badges: 19
Rep:
?
#14
Report 3 weeks ago
#14
(Original post by BrandonS15)
No problem, I’m just experimenting and I thought I could plug pi into the Fourier series of e^x to obtain and infinite series in closed form but I don’t fully understand why it’s discontinuous at x = pi on the -pi to pi interval
You'd want the Fourier series to be equal to both
e^pi
And
e^(-pi)
At the point x=pi, by considering the boundaries of the neighbouring intervals [-pi,pi] and [pi,3pi]. It can't be multivalued, so converges to the midpoint at the interval boundaries.
Last edited by mqb2766; 3 weeks ago
0
reply
DFranklin
Badges: 18
Rep:
?
#15
Report 3 weeks ago
#15
(Original post by BrandonS15)
How do you know it’s continuous or not? e^x is continuous for all x
I was meaning the periodic Fourier expansion of the sum will be continuous (because if g(x) = f(x)+f(-x) then g(pi)=g(-pi)).
0
reply
BrandonS15
Badges: 14
Rep:
?
#16
Report Thread starter 3 weeks ago
#16
(Original post by mqb2766)
You'd want the Fourier series to be equal to both
e^pi
And
e^(-pi)
At the point x=pi, by considering the boundaries of the neighbouring intervals [-pi,pi] and [pi,3pi]. It can't be multivalued, so converges to the midpoint at the interval boundaries.
Does that mean I can only evaluate it at x = 0?
0
reply
mqb2766
Badges: 19
Rep:
?
#17
Report 3 weeks ago
#17
(Original post by BrandonS15)
Does that mean I can only evaluate it at x = 0?
Of course not. The only "difficult" points are where the periodic function is discontinuous (or multivalued) at
x = (2k+1)pi
Where is an integer.
At all other values (including 0) the Fourier series converges to the periodic version of e^x.
Last edited by mqb2766; 3 weeks ago
0
reply
BrandonS15
Badges: 14
Rep:
?
#18
Report Thread starter 3 weeks ago
#18
(Original post by mqb2766)
Of course not. The only "difficult" points are where the periodic function is discontinuous (or multivalued) at
x = (2k+1)pi
At all other values (including 0) the Fourier series converges to e^x.
The Fourier series repeats every -pi to pi right? So I’m not understanding why it’s dis continuous at those points
0
reply
DFranklin
Badges: 18
Rep:
?
#19
Report 3 weeks ago
#19
(Original post by BrandonS15)
The Fourier series repeats every -pi to pi right? So I’m not understanding why it’s dis continuous at those points
What is e^x when x=-pi?
What is e^x when x=pi?
Since the Fourier series (call it F(x)) representation has period 2pi, F(-pi) must equal F(pi).
Do you see the problem here?
0
reply
mqb2766
Badges: 19
Rep:
?
#20
Report 3 weeks ago
#20
(Original post by BrandonS15)
The Fourier series repeats every -pi to pi right? So I’m not understanding why it’s dis continuous at those points
The (repeated/periodic) function
e^x
Is not continuous at those points. Have you sketched/plotted it for few period intervals? You can see the discontinuities at those points?

The Fourier series is also periodic. It converges to e^x in the interior of each interval. It must jump at those points to represent the original function. That's what causes the Gibbs/ringing effect when you have a truncated Fourier seies. In the limit, the oscillations die down, but it still has to jump at the interval boundaries.
Last edited by mqb2766; 3 weeks ago
0
reply
X

Quick Reply

Attached files
Write a reply...
Reply
new posts
Back
to top
Latest
My Feed

See more of what you like on
The Student Room

You can personalise what you see on TSR. Tell us a little about yourself to get started.

Personalise

How do you prefer to get careers advice?

I like to speak to my friends and family (8)
10.53%
I like to do my own research online using careers specific websites (50)
65.79%
I like speaking to the careers advisors at school, college or uni (11)
14.47%
I prefer to listen watch videos or listen to podcasts of people in my chosen career (6)
7.89%
Something else (let us know in the thread) (1)
1.32%

Watched Threads

View All