Download Evaluating Infinite Series: Geometric, Telescoping, Taylor, and Fourier Methods and more Lecture notes Signals and Systems in PDF only on Docsity!
Methods for Evaluating Infinite Series
Charles Martin
March 23, 2010
Geometric Series
The simplest infinite series is the geometric series. Given real (or complex!) numbers a and r,
∑^ ∞
n=
arn^ =
{ (^) a 1 −r if^ |r|^ <^1 divergent otherwise
The mnemonic for the sum of a geometric series is that it’s “the first term divided by one minus the common ratio.” You’ll see why words are helpful in the examples below.
Example 1. Let’s evaluate ∑∞ n=
3(1/2)n
Be careful; the index doesn’t start at n = 0. We could do the following manipulations:
∑^ ∞ n=
3(1/2)n^ =
∑^ ∞
n=
3(1/2)n+2^ =
∑^ ∞
n=
(3/4)(1/2)n^ = 3 /^4 1 − 1 / 2
Alternatively, the mnemonic gives the answer:
∑^ ∞ n=
3(1/2)n^ = first term 1 − ratio
A useful property of absolutely convergent series is that Re
Re and Im
Im ; that is, we can take the real and imaginary parts of a series easily.
Example 2. Let’s evaluate ∑∞ n=
cos(nθ) 2 n
You can use the comparison test to show that this series converges absolutely. Here’s how we’ll approach the sum: (^) ∞ ∑ n=
cos(nθ) 2 n^ =
∑^ ∞
n=
Re einθ 2 n^ =
∑^ ∞
n=
Re
eiθ 2
)n = Re
1 − eiθ^ / 2 = Re^
2 − eiθ^.
To finish the computation, use the identities Re z = (z + z)/2 and eiθ^ + e−iθ^ = 2 cos θ.
∑^ ∞ n=
cos(nθ) 2 n^
2 − eiθ^
2 − e−iθ^ =^4 −^ 2 cos^ θ 5 − 4 cos θ
Once you know some series, other ones follow easily.
Example 3. Let’s evaluate ∑∞ n=
cos^2 (nθ) 2 n^
Let’s use the identity 2 cos^2 θ = 1 + cos 2θ. Since all the sums converge absolutely, we can write
∑^ ∞ n=
cos^2 (nθ) 2 n^
∑^ ∞
n=
1 + cos 2θ 2 n^
4 − 2 cos 2θ 5 − 4 cos 2θ
7 − 5 cos 2θ 5 − 4 cos 2θ
Telescoping Sums
Sometimes evaluating the partial sums of a series is easy.
Example 1. Let’s evaluate ∑∞ n=
n^2 + n
Using partial fractions gives us 1 n^2 + n
n(n + 1)
n
n + 1
so that
∑^ ∞ n=
n^2 + n = lim N →∞
∑^ N
n=
n
n + 1
= (^) Nlim →∞
2 −^
N −^
N + 1
= lim N →∞
N + 1
Notice the way that each term canceled with the previous one. When a sum does this, we say it ‘telescopes’. It’s important to rely on the definition of an infinite series when trying to telescope a series. This example shows why.
Example 2. Let’s evaluate ∑∞ n=
arctan(n + 2) − arctan n.
A handwaving approach might say “the sum clearly telescopes, so the answer is arctan(∞) − arctan 0 = π/2. But this is wrong! If we’re careful we see that
∑^ N n=
arctan(n + 2) − arctan n = arctan(N + 2) + arctan(N + 1) − arctan 1 − arctan 0
= arctan(N + 2) + arctan(N + 1) − π/ 4 ,
so the infinite sum is π/2 + π/ 2 − π/4 = 3π/4.
Taylor Series
Occasionally a series can be recognized as a special case of Taylor series.
Example 1. Let’s evaluate ∑∞ n=
2 n n!
This looks a lot like the series for ex. With a little adjustment:
∑^ ∞ n=
2 n n!
∑^ ∞
n=
2 n n! − 1 = e^2 − 1.
Sometimes a series looks similar enough to a known Taylor series that derivatives and integrals might save the day.
All of the previous techniques fail here, but suppose we remember the Fourier series
x^2 = π^2 3
cos x − cos 2x 22
cos 3x 32
, −π ≤ x ≤ π.
Setting x = 0 gives us
0 = π
2 3
∑^ ∞
n=
(−1)n n^2
so the sum is −π^2 /12.
Honestly, this technique requires a repository of Fourier series. The other theorem which comes in handy is Parseval’s identity: with the same definitions as before but specializing L = π, we have
1 π
∫ (^) π
−π
|f (x)|^2 dx = a^20 2 +
∑^ ∞
n=
(a^2 n + b^2 n).
Here’s the main example.
Example 2. Let’s evaluate ∑∞ n=
n^4
Again, using
x^2 = π^2 3
cos x − cos 2x 22
cos 3x 32
, −π ≤ x ≤ π,
we have a 0 = 2π^2 /3, and an = 4(−1)n/n^2 for all other n. Each bn is zero. Therefore
1 π
∫ (^) π
−π
x^4 dx = (2π^2 /3)^2 2
∑^ ∞
n=
n^4
Evaluating the integral and rearranging gives
∑^ ∞ n=
n^4
π^4 90
Miscellaneous
Here are some examples of series that don’t quite fit into the other categories.
Example 1. Let’s evaluate the double sum ∑∞ m=
∑^ ∞
n=
3 m+n^
We can treat this like a double integral:
∑^ ∞ m=
∑^ ∞
n=
3 m+n^
∑^ ∞
m=
3 m
∑^ ∞
n=
3 n
n=
3 n
m=
3 m
Speaking of double integrals, here’s an example where interchanging sums can be fruitful.
Example 2. Let’s evaluate (^) ∞ ∑
n=
∑^ n
m=
3 m+n^
While no means impossible, this sum is a bit messy. It’s nicer if we interchange the order first (this is allowed because all terms are positive). To see how to change the order, just think about how it’s done for integrals.
∑^ ∞ n=
∑^ n m=
3 m+n^ =
∑^ ∞
m=
∑^ ∞
n=m
3 m+n^ =
∑^ ∞
m=
3 m
∑^ ∞
n=m
3 n^ =
∑^ ∞
m=
3 m
1 / 3 m 1 − 1 / 3
∑^ ∞
m=
9 m^ =
