# Homework Help: Riemann sums

1. Mar 22, 2006

### pureouchies4717

hi, is it possible to find the riemann sum of (cos1)^x?

it looks divergent to me

can someome please help me... even if it is convergent, i dont know how to find the sum of a trigonometric function

2. Mar 22, 2006

### arildno

Eeh, do you mean (cos(1))^x?

3. Mar 22, 2006

### d_leet

Did you really mean the riemann sum? Or did you mean the sum of the infinite series? Well since cos(0) = 1 and cos(pi) = -1 etc.. Then for any x that is not a multiple of pi |cos(x)| will be less than 1. Since this absolute value is less than 1 the geometric series will converge, specifically it will converge to a/(1-r) where a is the first term, and r is the common ratio.

4. Mar 22, 2006

### pureouchies4717

thanks alot for the input guys

can someone please tell me if 1.175 is right? thanks

Last edited: Mar 22, 2006
5. Mar 22, 2006

### d_leet

Where is the sum from? From 0 to infinity or what?

6. Mar 22, 2006

### arildno

1.175 is definitely not "right". What you've got is a geometric series, and you should be able to get a simple, exact expression for the sum, rather than an approximation.

7. Mar 22, 2006

### nrqed

yes, that is correct.

8. Mar 22, 2006

### pureouchies4717

ah i think i got it

i think the sum would be .540

9. Mar 22, 2006

### nrqed

well, you are right. But his answer is correct within 4 sig figs. I assumed he wanted a numerical, approximate value.

10. Mar 22, 2006

### pureouchies4717

thanks so much for the help guys :!!)

but now im kinda confused:rofl:

11. Mar 22, 2006

### arildno

Since there isn't said anywhere whether cos(1) is the cosine to 1 degree or to 1 radian, for example, the numerical answers are just meaningless.

12. Mar 22, 2006

### pureouchies4717

13. Mar 22, 2006

### arildno

And why is it "right" just to keep 4 significant figures? Why not 15, or just 1?
Nothing in what you posted gives any suggestion of what appoximation scheme you ought to use.

14. Mar 22, 2006

### pureouchies4717

ok heres what i did to get the 1.7... figure:

you said that it converges to a/(1-r)

A. a1= cos1
r= cos1 (right?)

so then: cos1/(1-cos1)

this comes out to 1.7

B. this is how i got the .5... answer:
a/(1-r)
a1= cos1
r= (cos1)^k

cos1/(1-(cos1)^k)

then i took the limits (cos1)^k becomes 0

so its: cos1/(1)= cos1 or .5...

15. Mar 22, 2006

### arildno

To be to the point:
Why are you so reluctant to give your answer as the exact expression
cos(1)/(1-cos(1)) ??

16. Mar 22, 2006

### pureouchies4717

o ok

hmm i just felt like rounding. thanks guys/girls

17. Mar 22, 2006

### d_leet

Well for case B, if you even just write out the first few terms of the eries you should see that that definitely cannot be teh answer because you have

cos(1) + cos2(1) + cos3(1) + ...

As you can see the first term in that series is about .54... and since none of those terms are ever negative the sum of the series must be larger than the first term about .54...

Case A gives you an approximation of the correct sum of this series.

18. Mar 22, 2006

### arildno

I'm glad you finally posted this, since finally, we can get rid of some misconceptions you show here.
A geometric series is of the form:
$$S=a*r^{0}+ar^{2}+++ar^{k}+$$
where a and r are real numbers, and |r|<1
$$\cos(1)*(\cos(1))^{0}+\cos(1)*(\cos(1))^{1}+++\cos(1)*(\cos(1))^{k}+++$$
Thus, a=r=cos(1)

19. Mar 22, 2006

### pureouchies4717

thanks everyone :)

and also,

i dont want to flood the board with topics, so can you guys please help with something else?

how is arctan(2n) divergent when, if you make n a big number, it clearly has a limit of 1.57?

20. Mar 22, 2006

### d_leet

I'm assuming you mean the infinite series with a general term of arctan(2n). Well it's usually the first test for convergence/divergence that you learn: that if the series converges then the limit of the terms as n goes to infinity must be 0, thus if the terms don't go to 0 then the series diverges.

21. Mar 22, 2006

### arildno

1. It doesn't have a limit at 1.57.

2. It isn't divergent.

22. Mar 22, 2006

### arildno

Just a question:
Whatever did this thread have to do with RIEMANN sums??

23. Mar 22, 2006

### pureouchies4717

o im sorry for the confusion... i meant normal sums

so arctan2n doesnt have a sum right? because its divergent?

24. Mar 22, 2006

### nrqed

That's the answer. There is nothing else to do..

(that comes out to 1.17534265.....but I have been chastised for writing an approcimate numerical value so I will say that the answer is cos(1)/(1-cos(1)) )

25. Mar 22, 2006

### HallsofIvy

The point is that you still haven't told us precisely what the problem was!

The geometric sum $\Sigma_{i=0}^\infty (cos(1))^n$ is a "geometric series", not a "Riemann Sum" (those are the finite sums used to define an integral). As Arildno said, you haven't stated whether that is 1 radian or 1 degree (although, I would say that, as long as we are talking about the cosine function as opposed to a trigonometry problem, radians should be assumed) and you haven't justified keeping only 4 significant figures.

Since the the geometric series $\sigma_{i= 0}{\infty}ar^n$ has sum $\frac{a}{1- r}$, as long as |r|< 1, as is cos(1). In your problem, a= cos(1) (since the sum starts at k= 1, not 0) and r= cos(1) (not "(cos(1)k)" . The sum is $\frac{cos(1)}{1- cos(1)}$.

Since the common ratio is cos(1) rather than cos(1)k, you do not take the limit as k goes to infinity. Indeed, if that were correct, for |r|< 1, the sum of $\sigma_{i=0}^{\infty}ar^n$ would always be just a.

Last edited by a moderator: Mar 22, 2006