In this sheet we use many results about limits of sequences, though we will not prove each of them explicitly. We remark that most theorems about limits of functions have completely analogous statements for limits of sequences (and exactly the same proofs). Specifically, recall that increasing sequences that are bounded above converge, as do decreasing sequences that are bounded below.

Definition 13.1. An infinite series (or, simply, a series) is a sequence (an)(a_n) of real numbers that we intend to sum, if possible, and thus write as:

∑n=1∞an.\sum_{n=1}^{\infty} a_n.

We may start the series at n=0n = 0 or at any other point after which ana_n is defined. We define the sequence of partial sums (pn)(p_n) of the series by:

pn=a1+⋯+an=∑i=1nai.p_n = a_1 + \dots + a_n = \sum_{i=1}^{n} a_i.

We say that the series converges if there exists L∈RL \in \mathbb{R} such that lim⁡n→∞pn=L\lim_{n \to \infty} p_n = L. When this is the case, we write this as:

∑n=1∞an=L,\sum_{n=1}^{\infty} a_n = L,

and we say that LL is the sum of the series. When there does not exist such an LL, we say that the series diverges.

Theorem 13.2. If ∑n=1∞an=L\sum_{n=1}^{\infty} a_n = L and ∑n=1∞bn=M\sum_{n=1}^{\infty} b_n = M and c∈Rc \in \mathbb{R}, then

∑n=1∞(an+bn)=L+M,∑n=1∞(c⋅an)=c⋅L.\sum_{n=1}^{\infty} (a_n + b_n) = L + M, \qquad \sum_{n=1}^{\infty} (c \cdot a_n) = c \cdot L.

Lemma 13.3. Let x∈Rx \in \mathbb{R} with ∣x∣<1|x| < 1. Given the sequence an=xna_n = x^n, we have lim⁡n→∞an=0\lim_{n \to \infty} a_n = 0.

Theorem 13.4 (Geometric Series). Let −1<x<1-1 < x < 1. Then:

∑n=1∞xn=x1−x.\sum_{n=1}^{\infty} x^n = \frac{x}{1 - x}.

Theorem 13.5. If ∑n=1∞an\sum_{n=1}^{\infty} a_n converges, then lim⁡n→∞an=0\lim_{n \to \infty} a_n = 0.

The converse of this theorem is however not true, as we see from the next theorem.

Theorem 13.6 (Harmonic Series). The series ∑n=1∞1n\sum_{n=1}^{\infty} \frac{1}{n} diverges.

Definition 13.7. We say that the series ∑n=1∞an\sum_{n=1}^{\infty} a_n converges absolutely if the series ∑n=1∞∣an∣\sum_{n=1}^{\infty} |a_n| converges.

Theorem 13.8. If a series converges absolutely, then it converges.

Theorem 13.9 (Alternating Series Test). Suppose that (an)(a_n) is a decreasing sequence of positive numbers and that lim⁡n→∞an=0\lim_{n \to \infty} a_n = 0. Then ∑n=1∞(−1)n+1an\sum_{n=1}^{\infty} (-1)^{n+1} a_n converges.

The following theorem will be useful to prove more specialized tests for convergence of series:

Theorem 13.10 (Comparison Test). Suppose that (cn)(c_n) is a sequence of positive numbers and that (an)(a_n) is a sequence such that ∣an∣≤cn|a_n| \leq c_n for all n>N0n > N_0, where N0N_0 is some fixed integer. If ∑n=1∞cn\sum_{n=1}^{\infty} c_n converges, then ∑n=1∞an\sum_{n=1}^{\infty} a_n converges.

Theorem 13.11 (Ratio Test). Let (an)(a_n) be a sequence such that each an>0a_n > 0 and suppose that:

lim⁡n→∞an+1an<1.\lim_{n \to \infty} \frac{a_{n+1}}{a_n} < 1.

Then ∑n=1∞an\sum_{n=1}^{\infty} a_n converges.

Proof:

Suppose that lim⁡n→∞an+1an=L<α<1\lim_{ {n} \to {\infty} } \frac{a_{n+1}}{a_{n}}=L<\alpha<1

Then when n≥N0n\geq N_{0}, aN0+1aN0<α\frac{a_{N_{0}+1}}{a_{N_{0}}}<\alpha. i.e. aN0+1<αaN0a_{N_{0}+1}<\alpha a_{N_{0}}
In the same way, aN0+2<αaN0+1<α2aN0a_{N_{0}+2}<\alpha a_{N_{0}+1}<\alpha^2a_{N_{0}}
Following this pattern, we have: aN0+k<αkaN0a_{N_{0}+k}<\alpha^ka_{N_{0}}
Let αkaN0=xk\alpha^ka_{N_{0}}=x_{k}, by Theorem 13.4, an∑k=1∞xka_{n}\sum_{k=1}^{\infty}x_{k} converges.
By Theorem 13.10, ∑N0∞an\sum_{N_{0}}^{\infty}a_{n} converges.
Since ∑n=1∞an=∑n=1N0an+∑n=N0∞an\sum_{n=1}^{\infty}a_{n}=\sum_{n=1}^{N_{0}}a_{n}+\sum_{n=N_{0}}^{\infty} a_{n}, and ∑n=1N0an\sum_{n=1}^{N_{0}}a_{n} is a constant, ∑n=1∞an\sum_{n=1}^{\infty} a_n converges..

What I learned:
- Find a α\alpha between LL and 11, because LL cant be used because it's hard to explain ϵ→0\epsilon\to0 while if we use 1, we couldn't create a converged RHS.
- Ratio form implies creating geometric sequence.
- Constant+converged sequence=converged sequence∎

Theorem 13.12 (Root Test). Let (an)(a_n) be a sequence such that each an>0a_n > 0 and suppose that:

lim⁡n→∞ann<1.\lim_{n \to \infty} \sqrt[n]{a_n} < 1.

Then ∑n=1∞an\sum_{n=1}^{\infty} a_n converges.

Similar to the proof of Theorem 13.11, Suppose that lim⁡n→∞ann=L<α<1.\lim_{ {n} \to {\infty} }\sqrt[n]{ a_{n} }=L<\alpha<1.
Then when n≥Nn\ge N, ann<α\sqrt[n]{ a_{n} }<\alpha i.e. an<αna_{n}<\alpha^n.
In the same way with proof above, using Theorem 13.4 and 13.10, we could finally prove that ∑n=1∞an\sum_{n=1}^{\infty} a_n converges.

Definition 13.13. For n∈Nn \in \mathbb{N}, we define the factorial of nn to be the product of all natural numbers less than or equal to nn. We denote this by the formula:

n!=n⋅(n−1)⋅(n−2)⋯3⋅2⋅1.n! = n \cdot (n-1) \cdot (n-2) \cdots 3 \cdot 2 \cdot 1.

By convention, we also set 0!=10! = 1.

Exercise 13.14. Prove that

∑n=0∞1n!\sum_{n=0}^{\infty} \frac{1}{n!}

converges. The number that it converges to is called ee.

We will now turn to sequences and series of functions.

Definition 13.15. Let A⊂RA \subset \mathbb{R}, and consider X={f:A→R}X = \{ f: A \to \mathbb{R} \}, the collection of real-valued functions on AA. A sequence of functions (on AA) is an ordered list (f1,f2,f3,…)(f_1, f_2, f_3, \ldots) which we will denote (fn)(f_n), where each fn∈Xf_n \in X. (More formally, we can think of the sequence as a function F:N→XF: \mathbb{N} \to X, where fn=F(n)f_n = F(n) for each n∈Nn \in \mathbb{N}, but this degree of formality is not particularly helpful.)

Remark that we can take the sequence to start at any n0∈Zn_0 \in \mathbb{Z} and not just at 1, just like we did for sequences of real numbers.

Definition 13.16. The sequence (fn)(f_n) converges pointwise to a function f:A⟶Rf: A \longrightarrow \mathbb{R} if for all p∈Ap \in A and ϵ>0\epsilon > 0, there exists N∈NN \in \mathbb{N} such that:

if n>N, then ∣fn(p)−f(p)∣<ϵ.\text{if } n > N, \text{ then } |f_n(p) - f(p)| < \epsilon.

In other words, we have that for all p∈Ap \in A, lim⁡n→∞fn(p)=f(p)\lim_{n \to \infty} f_n(p) = f(p).

Definition 13.17. The sequence (fn)(f_n) converges uniformly to a function f:A⟶Rf: A \longrightarrow \mathbb{R} if for all ϵ>0\epsilon > 0, there exists N∈NN \in \mathbb{N} such that:

if n>N, then ∣fn(p)−f(p)∣<ϵfor every p∈A.\text{if } n > N, \text{ then } |f_n(p) - f(p)| < \epsilon \quad \text{for every } p \in A.

Exercise 13.18. For each n∈Nn \in \mathbb{N}, let fn:[0,1]⟶Rf_n: [0,1] \longrightarrow \mathbb{R} be given by fn(x)=xnf_n(x) = x^n. Determine what function the sequence (fn)(f_n) converges to pointwise. Does this sequence converge uniformly to this function?

f(x)={x=0,0=<x<1x=1,x=1f(x)=\begin{cases} x=0,&0=<x<1 \\x=1,&x=1 \end{cases}

It does not converge uniformly to ff.

Proof:
Let ϵ=0.5\epsilon=0.5 and p=1p=1, then
Theorem 13.19. Let (fn)(f_n) be a sequence of functions, and suppose that each fn:A⟶Rf_n: A \longrightarrow \mathbb{R} is continuous. If (fn)(f_n) converges uniformly to f:A⟶Rf: A \longrightarrow \mathbb{R}, then ff is continuous.∎

Theorem 13.20. Suppose that (fn)(f_n) is a sequence of integrable functions on [a,b][a,b] and suppose that (fn)(f_n) converges uniformly to f:[a,b]⟶Rf: [a,b] \longrightarrow \mathbb{R}. Then

∫abf=lim⁡n→∞∫abfn.\int_a^b f = \lim_{n \to \infty} \int_a^b f_n.

Theorem 13.21. Let (fn)(f_n) be a sequence of functions on [a,b][a,b] such that each fnf_n is differentiable and fn′f_n' is integrable on [a,b][a,b]. Suppose further that (fn)(f_n) converges pointwise to ff and that (fn′)(f_n') converges uniformly to a continuous function gg. Then ff is differentiable and

f′(x)=lim⁡n→∞fn′(x).f'(x) = \lim_{n \to \infty} f_n'(x).

Definition 13.22. We define series of functions the same way we defined series of numbers. That is, given a sequence (fn)(f_n), define the sequence of partial sums (pn)(p_n) by

pn(x)=f1(x)+⋯+fn(x)p_n(x) = f_1(x) + \dots + f_n(x)

and say that ∑n=1∞fn\sum_{n=1}^{\infty} f_n converges pointwise or converges uniformly to ff if the sequence (pn)(p_n) does.

Theorem 13.23 (Weierstrass MM-test). Suppose that fn:A⟶Rf_n: A \longrightarrow \mathbb{R} is a sequence of functions and that there exists a sequence of positive real numbers (Mn)(M_n) such that for all x∈Ax \in A, we have ∣fn(x)∣≤Mn|f_n(x)| \leq M_n. If ∑n=1∞Mn\sum_{n=1}^{\infty} M_n converges, then for each x∈Ax \in A, the series of numbers ∑n=1∞fn(x)\sum_{n=1}^{\infty} f_n(x) converges absolutely. Furthermore, ∑n=1∞fn\sum_{n=1}^{\infty} f_n converges uniformly to the function ff defined by:

f(x)=∑n=1∞fn(x).f(x) = \sum_{n=1}^{\infty} f_n(x).

The previous theorem raises the question of what functions defined as series look like.

Definition 13.24. A function of the form

f(x)=∑n=0∞cn(x−a)n,cn∈Rf(x) = \sum_{n=0}^{\infty} c_n (x - a)^n, \qquad c_n \in \mathbb{R}

is called a power series. The power series is centered at aa and the numbers cnc_n are called the coefficients.

The next theorem gives a way to determine where a given power series is well-defined.

Theorem 13.25. Let

f(x)=∑n=0∞cnxnf(x) = \sum_{n=0}^{\infty} c_n x^n

be a power series centered at 0. Suppose that x0x_0 is a real number such that the series

f(x0)=∑n=0∞cnx0nf(x_0) = \sum_{n=0}^{\infty} c_n x_0^n

converges. Let rr be any number such that 0<r<∣x0∣0 < r < |x_0|. Then the following series of functions converges uniformly on [−r,r][-r, r] (and absolutely for each x∈[−r,r]x \in [-r, r]):

f(x)=∑n=0∞cnxn.f(x) = \sum_{n=0}^{\infty} c_n x^n.

The same is true for

g(x)=∑n=1∞ncnxn−1.g(x) = \sum_{n=1}^{\infty} n c_n x^{n-1}.

Furthermore, ff is differentiable on [−r,r][-r, r] and f′=gf' = g.

Theorem 11.26
Exercise 10.25