Week 6 Discussion Notes
Table of Contents
Power Series
Power series are a special type of infinite series:
Definition
Let F ( x ) = ∑ n = 0 ∞ a n ( x − c ) n \displaystyle F\p{x} = \sum_{n=0}^\infty a_n\p{x - c}^n F ( x ) = n = 0 ∑ ∞ a n ( x − c ) n . F F F is called a power series and we call c c c the center of F F F .
Power series are great because they have a bunch of nice properties (during discussion, I only talked about the first property):
Proposition (properties of power series)
Let F ( x ) = ∑ n = 0 ∞ a n ( x − c ) n \displaystyle F\p{x} = \sum_{n=0}^\infty a_n\p{x - c}^n F ( x ) = n = 0 ∑ ∞ a n ( x − c ) n be a power series.
There exists a number R ∈ [ 0 , ∞ ] R \in \br{0, \infty} R ∈ [ 0 , ∞ ] called the radius of convergence of F F F such that if ∣ x − c ∣ < R \abs{x - c} < R ∣ x − c ∣ < R , then F F F converges absolutely and if ∣ x − c ∣ > R \abs{x - c} > R ∣ x − c ∣ > R , then F F F diverges. If R = 0 R = 0 R = 0 , then the series only converges at its center and if R = ∞ R = \infty R = ∞ , then F F F converges absolutely everywhere.
The coefficients can be expressed in terms of the derivatives of F F F :
a n = F ( n ) ( c ) n ! a_n = \frac{F^{\p{n}}\p{c}}{n!} a n = n ! F ( n ) ( c )
You can differentiate term-by-term, i.e., you can pretend the sum isn't there and take the derivative like normal:
F ′ ( x ) = ∑ n = 1 ∞ n a n ( x − c ) n − 1 F'\p{x} = \sum_{n=1}^\infty na_n\p{x - c}^{n-1} F ′ ( x ) = n = 1 ∑ ∞ n a n ( x − c ) n − 1
F ′ F' F ′ is also a power series and it has the same radius of convergence as F F F (but not necessarily the same interval of convergence).
You can also integrate term-by-term:
∫ F ( x ) d x = C + ∑ n = 0 ∞ a n n + 1 ( x − c ) n + 1 \int F\p{x} \,\diff{x} = C + \sum_{n=0}^\infty \frac{a_n}{n+1} \p{x - c}^{n+1} ∫ F ( x ) d x = C + n = 0 ∑ ∞ n + 1 a n ( x − c ) n + 1
Like F ′ F' F ′ , the antiderivative of F F F is a power series with the same radius of convergence as F F F (but not necessarily the same interval of convergence).
Example 1.
We've already seen an example of a power series: if we replace r r r with x x x in the geometric series, then we get
∑ n = 0 ∞ r n = 1 1 − r ⇝ ∑ n = 0 ∞ x n = 1 1 − x , \sum_{n=0}^\infty r^n = \frac{1}{1 - r}
\rightsquigarrow \sum_{n=0}^\infty x^n = \frac{1}{1 - x}, n = 0 ∑ ∞ r n = 1 − r 1 ⇝ n = 0 ∑ ∞ x n = 1 − x 1 ,
and this converges if and only if ∣ x ∣ < 1 \abs{x} < 1 ∣ x ∣ < 1 . Thus, the geometric series above is an example of a power series centered at 0 0 0 and with interval of convergence ( − 1 , 1 ) \p{-1, 1} ( − 1 , 1 ) .
A common type of problem involving power series is finding its interval of convergence:
Example 2.
Find the interval of convergence of ∑ n = 1 ∞ 3 n x n n 3 \displaystyle\sum_{n=1}^\infty \frac{3^n x^n}{n^3} n = 1 ∑ ∞ n 3 3 n x n .
Solution.
For these types of questions, there's a step-by-step way to solve them:
Use the ratio or root test to find the radius of convergence.
Check whether the series converges at the endpoints.
For this series, we'll use the ratio test:
lim n → ∞ ∣ a n + 1 a n ∣ = lim n → ∞ 3 n + 1 ∣ x ∣ n + 1 ( n + 1 ) 3 n 3 3 n ∣ x ∣ n = lim n → ∞ 3 ∣ x ∣ ( n n + 1 ) 3 = 3 ∣ x ∣ . \begin{aligned}
\lim_{n\to\infty} \abs{\frac{a_{n+1}}{a_n}}
&= \lim_{n\to\infty} \frac{3^{n+1} \abs{x}^{n+1}}{\p{n+1}^3} \frac{n^3}{3^n \abs{x}^n} \\
&= \lim_{n\to\infty} 3 \abs{x} \p{\frac{n}{n+1}}^3 \\
&= 3\abs{x}.
\end{aligned} n → ∞ lim ∣ ∣ a n a n + 1 ∣ ∣ = n → ∞ lim ( n + 1 ) 3 3 n + 1 ∣ x ∣ n + 1 3 n ∣ x ∣ n n 3 = n → ∞ lim 3 ∣ x ∣ ( n + 1 n ) 3 = 3 ∣ x ∣ .
The ratio test tells us that the series will converge absolutely if
3 ∣ x ∣ < 1 ⟺ ∣ x ∣ < 1 3 ⟺ x ∈ ( − 1 3 , 1 3 ) . 3\abs{x} < 1
\iff \abs{x} < \frac{1}{3}
\iff x \in \p{-\frac{1}{3}, \frac{1}{3}}. 3 ∣ x ∣ < 1 ⟺ ∣ x ∣ < 3 1 ⟺ x ∈ ( − 3 1 , 3 1 ) .
This tells us that the radius of convergence is 1 3 \frac{1}{3} 3 1 . Now we need to check the endpoints:
x = − 1 3 x = -\frac{1}{3} x = − 3 1 :
We get
∑ n = 1 ∞ 3 n ( − 1 3 ) n n 3 = ∑ n = 1 ∞ ( − 1 ) n n 3 , \sum_{n=1}^\infty \frac{3^n \p{-\frac{1}{3}}^n}{n^3}
= \sum_{n=1}^\infty \frac{\p{-1}^n}{n^3}, n = 1 ∑ ∞ n 3 3 n ( − 3 1 ) n = n = 1 ∑ ∞ n 3 ( − 1 ) n ,
which converges, for example, by the alternating series test. (Alternatively, you can say that it converges absolutely by the p p p -test.)
x = 1 3 x = \frac{1}{3} x = 3 1 :
We get a similar series in this case:
∑ n = 1 ∞ 3 n ( 1 3 ) n n 3 = ∑ n = 1 ∞ 1 n 3 , \sum_{n=1}^\infty \frac{3^n \p{\frac{1}{3}}^n}{n^3}
= \sum_{n=1}^\infty \frac{1}{n^3}, n = 1 ∑ ∞ n 3 3 n ( 3 1 ) n = n = 1 ∑ ∞ n 3 1 ,
and this converges by the p p p -test.
Thus, the interval of convergence of this power series is
[ − 1 3 , 1 3 ] . \boxed{\br{-\frac{1}{3}, \frac{1}{3}}}. [ − 3 1 , 3 1 ] .
Convergence Tests
Example 3.
Determine whether ∑ n = 0 ∞ n n n 4 + ln n \displaystyle \sum_{n=0}^\infty \frac{n\sqrt{n}}{n^4 + \ln{n}} n = 0 ∑ ∞ n 4 + ln n n n converges.
Solution.
Heuristically, ln n \ln{n} ln n grows very, very slow, so if n n n is very large,
n n n 4 + ln n ≈ n n n 4 = 1 n 5 / 2 . \frac{n\sqrt{n}}{n^4 + \ln{n}}
\approx \frac{n\sqrt{n}}{n^4}
= \frac{1}{n^{5/2}}. n 4 + ln n n n ≈ n 4 n n = n 5/2 1 .
So, we'll try to apply the limit test with b n = 1 n 5 / 2 b_n = \frac{1}{n^{5/2}} b n = n 5/2 1 , which is summable by the p p p -test.
lim n → ∞ n n n 4 + ln n 1 n 5 / 2 = lim n → ∞ n 4 n 4 + ln n ⋅ 1 n 4 1 n 4 = lim n → ∞ 1 1 + ln n n 4 = 1 , \begin{aligned}
\lim_{n\to\infty} \frac{\frac{n\sqrt{n}}{n^4 + \ln{n}}}{\frac{1}{n^{5/2}}}
&= \lim_{n\to\infty} \frac{n^4}{n^4 + \ln{n}} \cdot \frac{\frac{1}{n^4}}{\frac{1}{n^4}} \\
&= \lim_{n\to\infty} \frac{1}{1 + \frac{\ln{n}}{n^4}} \\
&= 1,
\end{aligned} n → ∞ lim n 5/2 1 n 4 + l n n n n = n → ∞ lim n 4 + ln n n 4 ⋅ n 4 1 n 4 1 = n → ∞ lim 1 + n 4 l n n 1 = 1 ,
so the limit test tells us that because ∑ n = 1 ∞ 1 n 5 / 2 \sum_{n=1}^\infty \frac{1}{n^{5/2}} ∑ n = 1 ∞ n 5/2 1 converges, ∑ n = 1 ∞ n n n 4 + ln n \sum_{n=1}^\infty \frac{n\sqrt{n}}{n^4 + \ln{n}} ∑ n = 1 ∞ n 4 + l n n n n also converges.
Example 4.
Determine whether ∑ n = 0 ∞ cos n π n \displaystyle \sum_{n=0}^\infty \frac{\cos{n\pi}}{\sqrt{n}} n = 0 ∑ ∞ n cos nπ converges.
Solution.
The trick for this is to notice that cos n π = ( − 1 ) n \cos{n\pi} = \p{-1}^n cos nπ = ( − 1 ) n , so the series is
∑ n = 0 ∞ cos n π n = ∑ n = 0 ∞ ( − 1 ) n n , \sum_{n=0}^\infty \frac{\cos{n\pi}}{\sqrt{n}}
= \sum_{n=0}^\infty \frac{\p{-1}^n}{\sqrt{n}}, n = 0 ∑ ∞ n cos nπ = n = 0 ∑ ∞ n ( − 1 ) n ,
and this converges by the alternating series test.
Example 5.
Determine whether ∑ n = 0 ∞ n sin ( 1 n 2 ) \displaystyle \sum_{n=0}^\infty n\sin\p{\frac{1}{n^2}} n = 0 ∑ ∞ n sin ( n 2 1 ) converges.
Solution.
Recall from last week the fact that lim x → 0 sin x x = 1 \lim_{x\to0} \frac{\sin{x}}{x} = 1 lim x → 0 x s i n x = 1 . Heuristically, this means that if x ≈ 0 x \approx 0 x ≈ 0 , then
sin x x ≈ 1 ⟹ sin x ≈ x . \frac{\sin{x}}{x} \approx 1
\implies \sin{x} \approx x. x sin x ≈ 1 ⟹ sin x ≈ x .
Specializing to our problem, if n n n is large enough, then 1 n 2 ≈ 0 \frac{1}{n^2} \approx 0 n 2 1 ≈ 0 , so if we plug in x = 1 n 2 x = \frac{1}{n^2} x = n 2 1 , we get
sin ( 1 n 2 ) ≈ 1 n 2 ⟹ n sin ( 1 n 2 ) ≈ 1 n . \sin\p{\frac{1}{n^2}} \approx \frac{1}{n^2}
\implies
n\sin\p{\frac{1}{n^2}} \approx \frac{1}{n}. sin ( n 2 1 ) ≈ n 2 1 ⟹ n sin ( n 2 1 ) ≈ n 1 .
This tells us to try b n = 1 n b_n = \frac{1}{n} b n = n 1 with limit comparison:
lim n → ∞ n sin ( 1 n 2 ) 1 n = lim n → ∞ sin ( 1 n 2 ) 1 n 2 = 1. \lim_{n\to\infty} \frac{n\sin\p{\frac{1}{n^2}}}{\frac{1}{n}}
= \lim_{n\to\infty} \frac{\sin\p{\frac{1}{n^2}}}{\frac{1}{n^2}}
= 1. n → ∞ lim n 1 n sin ( n 2 1 ) = n → ∞ lim n 2 1 sin ( n 2 1 ) = 1.
But ∑ n = 1 ∞ 1 n \sum_{n=1}^\infty \frac{1}{n} ∑ n = 1 ∞ n 1 diverges by the p p p -test, so by the limit comparison test, ∑ n = 1 ∞ n sin ( 1 n 2 ) \sum_{n=1}^\infty n\sin\p{\frac{1}{n^2}} ∑ n = 1 ∞ n sin ( n 2 1 ) also diverges.
Example 6.
Determine whether ∑ n = 0 ∞ ( 1 − cos ( 1 n ) ) \displaystyle \sum_{n=0}^\infty \p{1 - \cos\p{\frac{1}{n}}} n = 0 ∑ ∞ ( 1 − cos ( n 1 ) ) converges.
Solution.
Like the previous example, we'll use a "well-known" limit to do this problem:
lim x → 0 1 − cos x x 2 = 1 2 . \lim_{x\to0} \frac{1 - \cos{x}}{x^2} = \frac{1}{2}. x → 0 lim x 2 1 − cos x = 2 1 .
You can prove this as follows:
lim x → 0 1 − cos x x 2 ⋅ 1 + cos x 1 + cos x = lim x → 0 sin 2 x x 2 1 1 + cos x = lim x → 0 ( sin x x ) 2 1 1 + cos x = 1 2 . \begin{aligned}
\lim_{x\to0} \frac{1 - \cos{x}}{x^2} \cdot \frac{1 + \cos{x}}{1 + \cos{x}}
&= \lim_{x\to0} \frac{\sin^2{x}}{x^2} \frac{1}{1 + \cos{x}} \\
&= \lim_{x\to0} \p{\frac{\sin{x}}{x}}^2 \frac{1}{1 + \cos{x}} \\
&= \frac{1}{2}.
\end{aligned} x → 0 lim x 2 1 − cos x ⋅ 1 + cos x 1 + cos x = x → 0 lim x 2 sin 2 x 1 + cos x 1 = x → 0 lim ( x sin x ) 2 1 + cos x 1 = 2 1 .
So, we get the following heuristic: if x ≈ 0 x \approx 0 x ≈ 0 , then
1 − cos x x 2 ≈ 1 2 ⟹ 1 − cos x ≈ x 2 2 . \frac{1 - \cos{x}}{x^2} \approx \frac{1}{2}
\implies 1 - \cos{x} \approx \frac{x^2}{2}. x 2 1 − cos x ≈ 2 1 ⟹ 1 − cos x ≈ 2 x 2 .
If we plug in x = 1 n x = \frac{1}{n} x = n 1 , then
1 − cos ( 1 n ) ≈ 1 2 n 2 . 1 - \cos\p{\frac{1}{n}} \approx \frac{1}{2n^2}. 1 − cos ( n 1 ) ≈ 2 n 2 1 .
From here, you can use b n = 1 n 2 b_n = \frac{1}{n^2} b n = n 2 1 (or b n = 1 2 n 2 b_n = \frac{1}{2n^2} b n = 2 n 2 1 ; both will work) in the limit comparison test. Then
lim n → ∞ 1 − cos ( 1 n ) 1 n 2 = lim n → ∞ 1 − cos ( 1 n ) ( 1 n ) 2 = 1 2 . \lim_{n\to\infty} \frac{1 - \cos\p{\frac{1}{n}}}{\frac{1}{n^2}}
= \lim_{n\to\infty} \frac{1 - \cos\p{\frac{1}{n}}}{\p{\frac{1}{n}}^2}
= \frac{1}{2}. n → ∞ lim n 2 1 1 − cos ( n 1 ) = n → ∞ lim ( n 1 ) 2 1 − cos ( n 1 ) = 2 1 .
The p p p -test tells us that ∑ n = 1 ∞ 1 n 2 \sum_{n=1}^\infty \frac{1}{n^2} ∑ n = 1 ∞ n 2 1 converges, so by limit comparison, ∑ n = 1 ∞ ( 1 − cos ( 1 n ) ) \sum_{n=1}^\infty \p{1 - \cos\p{\frac{1}{n}}} ∑ n = 1 ∞ ( 1 − cos ( n 1 ) ) also converges.
Example 7.
True or false:
If ∑ n a n \sum_n a_n ∑ n a n converges absolutely, then ∑ n a n 2 \sum_n a_n^2 ∑ n a n 2 converges.
Solution.
This is true. Since ∑ n ∣ a n ∣ \sum_n \abs{a_n} ∑ n ∣ a n ∣ converges, we know that
lim n → ∞ ∣ a n ∣ = 0. \lim_{n\to\infty} \abs{a_n} = 0. n → ∞ lim ∣ a n ∣ = 0.
This means that eventually, ∣ a n ∣ ≤ 1 \abs{a_n} \leq 1 ∣ a n ∣ ≤ 1 . Mathematically, this means there exists N > 0 N > 0 N > 0 such that if n ≥ N n \geq N n ≥ N , then ∣ a n ∣ ≤ 1 \abs{a_n} \leq 1 ∣ a n ∣ ≤ 1 . So, if n ≥ N n \geq N n ≥ N , then
a n 2 = ∣ a n ∣ 2 = ∣ a n ∣ ⋅ ∣ a n ∣ ≤ 1 ⋅ ∣ a n ∣ = ∣ a n ∣ . \begin{aligned}
a_n^2
&= \abs{a_n}^2 \\
&= \abs{a_n} \cdot \abs{a_n} \\
&\leq 1 \cdot \abs{a_n} \\
&= \abs{a_n}.
\end{aligned} a n 2 = ∣ a n ∣ 2 = ∣ a n ∣ ⋅ ∣ a n ∣ ≤ 1 ⋅ ∣ a n ∣ = ∣ a n ∣ .
Since ∑ n = N ∞ ∣ a n ∣ \sum_{n=N}^\infty \abs{a_n} ∑ n = N ∞ ∣ a n ∣ converges, direct comparison tells us that ∑ n = N ∞ a n 2 \sum_{n=N}^\infty a_n^2 ∑ n = N ∞ a n 2 also converges, so
∑ n = 1 ∞ a n 2 = ∑ n = 1 N − 1 a n 2 + ∑ n = N ∞ a n 2 , \sum_{n=1}^\infty a_n^2
= \sum_{n=1}^{N-1} a_n^2 + \sum_{n=N}^\infty a_n^2, n = 1 ∑ ∞ a n 2 = n = 1 ∑ N − 1 a n 2 + n = N ∑ ∞ a n 2 ,
i.e., this is the sum of two convergent series, so the entire series converges.