Real Analysis Sequences and Series of Functions: Theorem 2: Assume that R > 0. suppose that 0 < r < R.

By [REDCATED] Then the power series converges uniformly and absolutely on

Pointwise Convergence: The sequence (fn ) converges |x − c| ≤ r to a continuous function f , i.e.
pointwise to the function f , iff for every x in the domain we
∑
∞
Introduction have f (x) = an (x − c)n
lim fn (x) = f (x) n=0
n→∞
Continuous at a: ∀ϵ > 0, ∃δ > 0 such that |x−a| < δ implies
|f (x) − f (a)| < ϵ. Uniform Convergence: The sequence (fn ) converges uni- defines a continous function f : (c − R, c + R) → R.
Supremum: The lest upper bound. The unique real number formly on a set E with limit f if for every ϵ > 0 there exists Lemma: The two power series
M such that: an N ∈ N such that for all n ≥ N and x ∈ E:
∑
∞ ∑
∞

• a ≤ M for all a ∈ A and |fn (x) − f (x)| < ϵ an (x − c)n and nan (x − c)n−1
n=1 n=1
• If M ′ is any real number such that a ≤ M ′ for all a ∈ A,
then M ≤ M ′ If fn converges uniformly then it converges pointwise. have the same radius of convergence.
Weierstrass M-test: Suppose there exists a sequence Theorem 3: f (z) defined above is infinitely differentiable on
Convergence: an converges to a if for every ϵ > 0 there exists (M n )n∈N such that |fn (x)| ≤ Mn , ∀x ∈ E and n ∈ N and
an N ∈ N such that n ≥ N implies |an − a| < ϵ. ∑∞ ∑∞ |x − c| < R where R is the radius of convergence, and for such
n=1 Mn < ∞. Then the series n=1 fn (x) converges uni- x:
Cauchy Sequence: If for every ϵ > 0 there exists an N ∈ N formly on E. ∑
∞
such that m, n ≥ N implies |am − an | < ϵ. Lemma: Uniform convergence iff: f ′ (x) = nan (x − c)n−1
Note: Convergence says that the numbers are getting closer n=0
and closer to a as n gets bigger, while Cauchy says that the lim sup |fn (x) − f (x)| = 0 and the series converges absolutely and uniformly on [c−r, c+r]
n→∞ x∈E
numbers are getting closer and closer to each other as n gets for any r < R. Moreover:
bigger. Lemma: If fn is continuous on some interval, and f isn’t,
Bolzano-Weierstrass Theorem: Every sequence of real then the convergence isn’t uniform. f (n) (c)
an =
numbers which is bounded must have a convergent subse- Uniformly Cauchy: A sequence of functions fn is uniformly n!
quence. Cauchy if for all ϵ > 0 there exists N ∈ N such that if m, n ≥ N
|a | Uniformly Continuous: Let I be an interval and let f : I →
Ratio Test: If a sequence satisfies limn→∞ |an+1 = r < 1, then ∀x ∈ E we have |fm (x) − fn (x)| < ϵ
∑ n| R be a function. We say that f is uniformly continuous on I
then the series n an converges. If r > 1 then the series di- Power Series: is a series of the form if for every ϵ > 0 there is a δ > 0 such that for all x, y ∈ I,
verges, if r = 1 it is inconclusive. |x − y| < δ implies that |f (x) − f (y)| < ϵ.
Intermediate Value Theorem: For a continuous function ∑
∞
an (x − c)n Lemma: Let I be an open interval in R. Suppose f : I → R
f : R → R, IVT says that for every value c strictly between n=0 is differentiable and its derivative f ′ is bounded on I. Then f
f (a) and f (b) there is some x ∈ [a, b] such that f (x) = c. is uniformly continuous on I.
Extreme Value Theorem: For a continuous function f : Where an are the coefficients and c is its centre. Proof: Suppose that |f ′ (ζ)| ≤ M for all ζ ∈ I. By MVT we
R → R, there are values c and d in [a, b] at which f takes on Radius of Convergence: R is defined by have f (x) − f (y) = (x − y)f ′ (ζ) for some ζ between x and y.
the extreme values m and M where m ≤ f (x) ≤ M for all So |f (x) − f (y)| = |x − y||f ′ (ζ)| ≤ M |x − y|. Let ϵ > 0 and let
x ∈ [a, b]. Meaning f (c) = m and f (d) = M . R = sup{r ≥ 0 : (an rn ) is bounded}
δ = ϵ/M . If now |x − y| < δ we have M |x − y| < M δ = ϵ.
Mean Value Theorem: For a function f : R → R, contin- Theorem: Suppose f : [a, b] → R is continuous. Then it is
Unless (an rn ) is bounded for all r ≥ 0, in which case we declare
uous on a closed interval [a, b] and differentiable on an open uniformly continuous.
R=∞
interval (a, b), then there exists a point c ∈ (a, b) such that Proof: Assume f isn’t unif cont. Then there is ϵ > 0 and there
It is a general fact that the radius of convergence of a power
f ′ (c) = f (b)−f (a)
are sequences xn , yn with |xn −yn | → 0 but |f (xn )−f (yn )| ≥ ϵ.
b−a series is given by:
Rolles Theorem: If f is continuous on a closed interval [a, b] Bolzano-Weierstrass tells us that (xn ) has a convergent subse-
lim inf |ak |−1/k
and differentiable on the open interval (a, b), and f (a) = f (b), n→∞ k≥n quences xnk → x ∈ [a, b]. Since |xn − yn | → 0, ynk also
then there exists at least one c ∈ (a, b) such that f ′ (c) = 0. Root Test: Let converges to x as k → ∞. Continuity of f at x gives that
√
Chain Rule: Given a composition of differentiable functions L = lim sup n
|an | limk→∞ f (xnk ) = f (x) and similarly limk→∞ f (ynk ) = f (x).
ϕ(x) = L ◦ E(x) we have ϕ′ (x) = L′ (E(x))E ′ (x) n→∞ So limk→∞ |f (xnk ) − f (ynk )| = 0. But this contradicts the
Cauchy-Schwarz Inequality: If L < 1 then the series converges absolutely. If L > 1 then it fact that |f (xn ) − f (yn )| ≥ ϵ for all n. ∑
∞
√ √ diverges. if L = 1 and the limit approaches strictly from above Series Converge Pointwise: Series n=1 fn (x) converges
x1 y1 + ... + an bn ≤ a21 + ... + a2n b21 + ... + b2n then the series diverges. pointwise to a function s : E → R on E iff for every x ∈ E

1
 ∑ ϵ > 0 there is an N ∈ N such that for k ≥ N we have
and ϵ > 0. There exists N such that n ≥ N and ∀x ∈ J we have |fn (y) − f (y)| < ϵ/2 for all y. Let y = xn .
| kn=1 fn (x) − s(x)| < ϵ. ∑∞ we have |fn (x) − f (x)| < ϵ/3. Continuity of fN at a So if n ≥ N we have |fn (xn ) − f (xn )| < ϵ/2. Since
Series Converge Uniform: Series n=1 fn (x) converges implies that ∃δ > 0 such that for |x − a| < δ we have xn → x, and f is continuous, there is an M ∈ N such
uniformly to s : E → R on E iff for every ϵ > 0 there |fN (x) − fN (a)| < ϵ/3. So for |x − a| < δ we have: that n ≥ M implies |fn (xn ) − f (xn )| < ϵ/2. So if
∑ ∈ N such that for k ≥ N and all x ∈ E we
is an N we take n ≥ max{N, M } we have: |fn (xn ) − f (x)| ≤
have | kn=1 fn (x) − s(x)| < ϵ. Alternatively iff for every |f (x) − f (a)| ≤ |fn (xn ) − f (xn )| + |f (xn ) − f (x)| < ϵ/2 + ϵ/2 = ϵ
∑
∑ is an N ∈ N such that for k ≥ N we have
ϵ > 0 there
|f (x) − fN (x)|+|fN (x) − fN (a)|+|fN (a) − f (a)| < 3ϵ/3 • Proof ∞ 1
n=1 np converges for p > 1:
supx∈E | kn=1 fn (x) − s(x)| < ϵ From integral test we want f continuous, positive, de-
So f is continuous at a. creasing on [1, ∞) such that an = f (n).
Sequence and Series Examples: • If fn is uniformly continuous and converges uniformly to
∑
∞ ∫ ∞
f on J then f is uniformly continuous. Proof: consider an converges ⇐⇒ f (x)dx converges
• From FPM: nxn → 0 as n → ∞ any ϵ > 0. There exists N such that n ≥ N and ∀x ∈ J n=1 1
we have |fn (x) − f (x)| < ϵ/3. Uniform continuity of fN ∫∞
• f : I → R is uniformly continuous on I iff when- implies that ∃δ > 0 such that for x, y ∈ J, |x−y| < δ =⇒ So define f (x) = x1 , then 1
f (x)dx = [ln(x)]∞
1 =
ever sn , tn ∈ I are such that |sn − tn | → 0, then |fN (x) − fN (y)| < ϵ/3. So for any x, y ∈ J, |x − y| < δ limn→∞
∑ ln(u) − 0 = ∞.
|f (sn ) − f (tn )| → 0. we have: So ∞ 1
n=1 n doesn’t converge.
Proof: Suppose f : I → R is unif cont on I and that |f (x) − f (y)| ≤
sn , tn ∈ I are such that |sn − tn | → 0 as n → ∞. Let
ϵ > 0. By unif cont of f there is a δ > 0 such that |f (x) − fN (x)|+|fN (x) − fN (y)|+|fN (y) − f (y)| < 3ϵ/3 Integration of Real Functions
|x − y| < δ =⇒ |f (x) − f (y)| < ϵ. Since |sn − tn | → 0 So f is uniformly continuous on J.
there is an N such that n ≥ N implies |sn − tn | < δ. Characteristic Function: With E ⊆ R define χE : R → R
• Let fn (x) = nx(1 − x2 )n for 0 ≤ x ≤ 1. Find limit func- by χE (x) = 1 if x ∈ E and χE (x) = 0 if x ∈ / E. Let I be a
So if n ≥ N we have |f (sn ) − f (tn )| < ϵ. Now suppose
tion and if uniform. bounded interval, and:
that f is continuous but not unif cont. So there is an
Proof: If x = 0, 1 then fn (0) = 0. If 0 < x < 1 then ∫
ϵ > 0 such that taking δ = 1/n there are sn , tn ∈ I with
0 < 1 − x2 < 1 and so n(1 − x2 )n → 0. So fn con- χI := length(I) = |I| = b − a
|sn − tn |δ but |f (sn ) − f (tn )| ≥ ϵ. So |sn − tn | → 0 but
verges pointiwse to the zero ∫function. ∫ If it was uni-
|f (sn ) − f (tn )| doesn’t tend to zero. 1 1
form convergence we’d have 0 fn → 0 0 = 0. But .
• Let f (x) = x2 when x is rational and f (x) = 0 when x ∫1
f = 2(n+1)
0 n
n
→ 1/2 ̸= 0. So convergence isn’t uni- Step Function: We say ϕ : R → R is a step function with
is irrational. f is discontinuous everywhere except zero, respect to {x0 , x1 , ..., xn } if there exists x0 < x1 < ... < xn for
√ form on [0, 1]. If a ≤ x ≤ 1 then 1 − x2 ≤ 1 − a2 so that
since if ϵ > 0 then |x| < ϵ implies |f (x) = f (0)| < ϵ. some n ∈ N such that:
|fn (x)| ≤ n(1−a2 )n → 0 since a > 0. So the convergence
Therefore it can’t be differentiable at any point except
is uniform on such intervals. • ϕ(x) = 0 for x < x0 and x > xn
possibly zero (It is differentiable at zero).
∑∞ 2 n • Is f (x) = 1/x on (0, ∞) uniformly continuous? No. • ϕ is constant on (xj−1 , xj ) 1 ≤ j ≤ n
• Radius of convergence of n=0 an x : the sequence is
n ∞
(an r )n=0 is bounded for r < R and unbounded for Proof: Take ϵ = 1. Consider the sequences xn = 1/n In other words, ϕ is a step function with respect to
r > R. So (a2n r2n ) is bounded for r2 < R2 and un- and yn = 1/(n + 1). Then |f (xn ) − f (yn )| = 1 so there is {x0 , x1 , ..., xn } iff there exists c0 , c1 , ..., cn such that
bounded for r2 > R2 so that (a2n sn ) is bounded for no δ > 0 such that |x − y| < δ implies |f (x) − f (y)| < 1.
What about [a, ∞) for a > 0? Yes. ∑
n
s < R2 and unbounded for s > R2 . So R2 is the ra- ϕ(x) = cj χ(xj−1 ,xj ) (x)
dius of convergence. Proof: Let ϵ > 0. Consider |f (x) − f (y)| for a ≤ x, y.
This equals |x − y|/|xy| ≤ a−2 |x − y| for such x, y. So
j=1
∑
• Radius of convergence of ∞ n=0 a2n x
2n
cannot be deter- if a > 0, we take δ < ϵa2 we have |x − y| < δ =⇒ Integrals of Step Functions: If ϕ is a step function with
mined only from R. It might happen that a2n = 0 and |f (x) − f (y)| < a−2 ϵa2 = ϵ. respect to {x0 , x1 , ..., xn } which takes the value on cj on
the radius of convergence is infinite, or the radius of con-
• Let fn converge uniformly to f . Let (xn ) be a se- (xj−1 , xj ), then
vergence could be R. It could be any value in [R, ∞).
∑∞ n quence of real numbers which converge to x ∈ R. Show ∫ ∑n
• If R is the radius of convergence for n=0 an x then
∑∞ 2n
√ fn (xn ) → f (x). ϕ := cj (xj − xj−1 )
the radius
∑∞ of convergence for n=0 a n x is ρ = R Proof: From triangle inequality |fn (xn ) − f (x)| ≤ j=1
and n=0 a2n xn is ρ = R2 |fn (xn ) − f (xn )| + |f (xn ) − f (x)|. Since it’s implied
• If fn is continuous and converges uniformly to f on that f is continuous, let ϵ > 0. Then by uniform conver-
and ∫ ∫ ∫
J then f is continuous. Proof: Let a ∈ J and let gence of fn to f , there exists N ∈ N such that n ≥ N (αϕ + βψ) = α ϕ+β ψ

2
Riemann Integrable Def 1: Let f : R → R we say that • max{f, g} and min{f, g} are R-I Integral Examples:
f is Riemann Integrable if for every ϵ > 0 there exists step • f g is R-I
functions ϕ and ψ such that: • If f : R → R is R-I, then f must be bounded and have
Theorem: If f is not zero outside some bounded interval then bounded support.
• ϕ≤f ≤ψ it is not integrable.
∫ ∫ Proof: If f is R-I then taking ϵ = 1, there exists step
• ψ− ϕ<ϵ Theorem 4: If g : [a, b] → R is continuous, and f is defined functions such that ϕ ≤ f ≤ ψ. Then |f | ≤ max{|ϕ, |ψ|},
by f (x) = g(x) for a ≤ x ≤ b, f (x) = 0 for x ∈
/ [a, b], then f is
Theorem
∫ 1: A function f : R → R is R-I iff: ∫
which as a step function, takes only finitely many val-
R-I.
sup{ ϕ : ϕ is a step function and ϕ ≤ f } = inf{ ψ : ues, therefore is bounded. So f is bounded. Moreover,
Fundamental Theorem of Calculus: Let g : [a, b] → R be there are M, N ∈ R+ such that ϕ(x) = 0 for |x| > M
ψ is a step function and ψ ≥ f }
R-I. For a ≤ x ≤ b define and ψ(x) = 0 for |x| > N , so that ϕ(x) = ψ(x) = 0 for
Integral Definition:
∫ if f is R-I, and ϕ, ψ are step functions,
∫ x
then we define f : |x| > max{M, N }. Since ϕ ≤ f ≤ ψ this forces f (x) = 0
G(x) = g for |x| > max{M, N }, and so f has bounded support.
∫ ∫ ∫ a
f := sup{ ϕ : ϕ ≤ f } = inf{ ψ : ψ ≥ f } • Not zero outside some bounded interval =⇒ not inte-
G is differentiable on (a, b) and its derivative is g(x).
grable.
Riemann Integrable Def 2: A function f : R → R is R-I iff Theorem ∫ x 5: Let g : [a, b] → R be R-I. For a ≤ x ≤ b let Not bounded =⇒ not integrable.
there exists sequences of step functions ϕn and ψn such that: G(x) = a g. Suppose g is continuous at x for some x ∈ [a, b].
Proof: Since every step function is bounded and vanish-
∫ ∫ Then G is differentiable at x and G′ (x) = g(x).
ing outside a bounded interval, the fact that ϕn ≤ f ≤
ϕn ≤ f ≤ ψn for all n, and ψn − ϕ n → 0 Theorem 6: Suppose f : [a, b] → R has continuous derivative
ψn implies the same for f .
f ′ on [a, b]. Then ∑N r N +1 −1
∫ b • n
n=1 r = r−1
provided −1 < r < 1
If ϕn and ψn are any sequences of step functions satisfying
above, then as n → ∞ f ′ = f (b) − f (a) • Converges only if: p > 1
a
∫ ∫ ∫ ∫ ∑∞
1
ϕn → f and ψn → f Theorem 7: Suppose that fn : R → R is a sequence of R-I p
n
functions which converges uniformly to a function f . Suppose n=1
that fn and f are zero outside some common interval [a, b].
Lemma 1: Let f : R → R be a bounded function with bounded • This function is R-I:
Then f is R-I and {
support [a, b]. The following are equivalent: x = 1/n2 , n ∈ N
∫ ∫ f (x) =
1
• f is Riemann-Integrable f = lim fn 0 otherwise
n→∞
• For every ϵ > 0 there exists a = x0 < ... < xn = b such Proof: Takes infintely many different values so not a step
that if Mj and mj denote the supremum and infimimum Corollary: Suppose∑that fn : R → R is a sequence of R-I function. Define ϕ = 0 and ψ(x) = 1 for 0 ≤ x ≤ 1/N 2
values of f on [xj−1 , xj ] respectively then functions such that n fn converges uniformly to a function and x = 1/n2 , 1 ≤ n ≤ ∫N and ψ(x)
f . Suppose that fn and ∫ = 0 otherwise.
∑
n ∑f are zero outside some common in- We get ϕ ≤ f ∫≤ ψ, and ϕ = 0, ψ = 1/N 2 . So f
terval [a, b]. Then f = n fn is R-I and integrable and f = 0.
(Mj − mj )(xj − xj−1 ) < ϵ
∫ ∑ ∑∫
j=1 • χQ∩[0,1] Is not R-I.
fn = fn Proof: let there be step functions such that ϕ ≤
• For every ϵ > 0 there exists a = x0 < ... < xn = b such n n
χQ∩[0,1] ≤ ψ. Then on any interval of positive length
that, with Ij = (xj−1 , xj ) for j ≥ 1: Integral Test: Suppose (an )∞
is a non negative sequence on which ϕ is constant, the value of ϕ must in fact
n=1

∑
n of numbers and f : [1, ∞) → (0, ∞) is a function such that: be non-positive. This is because any interval of posi-
∫n tive length must contain irrationals, and we have that
sup |f (x) − f (y)| |Ij | < ϵ • 1 f ≤ K for some K and all n
j=1
x,y∈Ij χQ∩[0,1] (x) = 0 for irrational x. Thus ϕ(x) ≤ 0 ex-
• an ≤ f (x) for n ≤ x < n + 1 cept for possibly finitely many values of x, and therefore
∑ ∫
Theorem 3: Suppose f and g are R-I, and α, β are real num- Then n an converges ∑n to a real number which is at most K. ϕ ≤ 0. Similarly any interval of positive length must
bers. Then: Proof: Take ϕ = k=1 ak χ[k,k+2) is a step function which contain rationals, ψ must be at least 1 on any interval
• αf + βg is R-I and the integral is what you expect satisfies ϕ ≤ f χ[1,n+1) so that: of positive ∫length meeting [0,
∫ 1] on∫ which it is constant.
∫ ∫ ∫ ∫ ∫ ∫ Therefore ψ ≥ 1. Hence ψ − ϕ ≥ 1. So not true
• If f ≥ 0 then f ≥ 0, if f ≤ g then f ≤ g ∑
n n+1
that ∀ϵ > 0 there exist∫ step ∫functions ϕ and ψ such that
∫ ∫ ak = ϕ≤ f χ[1,n+1) = f ≤K
• |f | is R-I and | f | ≤ |f | k=1 1 ϕ ≤ χQ∩[0,1] ≤ ψ and ψ − ϕ < ϵ.

3
Metric Spaces Completeness and Contraction Compactness in Metric Spaces
Metric Space: is a set X with a function d : X × X → R Converging: (xn ) converges in X if there is a point a ∈ X Cover: A covering of X is a collection of sets whose union is
which satisfies the following for all x, y, z ∈ X: such that for every ϵ > 0 there is an N ∈ N such that: X. An open covering of C is a collection of open sets whose
union is X.
• Positive Definite: d(x, y) ≥ 0 with d(x, y) = 0 iff x = y Compact: For Q to be compact: for every open cover of Q
n≥N =⇒ d(xn , a) < ϵ
• Symmetric: d(x, y) = d(y, x) there is a finite subcover.
Negation: There exists some open cover {Uα } of Q which has
• Triangle Inequality: d(x, y) ≤ d(x, z) + d(z, y) Cauchy: (xn ) is Cauchy if for every ϵ > 0 there is an N ∈ N no finite subcover.
such that: More in Depth Compactness Definition: For every col-
The Usual Metric: Every Euclidean space Rn is a metric
space with metric d(x, y) = ||x − y||. lection of open sets {Uα } in R2 such that Q ⊆ ∪α Uα there is a
n, m ≥ N =⇒ d(xn , xm ) < ϵ finite subcollection {Uα1 , ..., Uαk } such that Q ⊆ ∪kj=1 Uαj .
The Discrete Metric: R is a metric space with metric
Proposition: A subset A ⊆ Rn is compact iff it is closed and
{
0 x=y Complete Metric Space: If every Cauchy sequence of points bounded.
σ(x, y) =
1 x ̸= y in the metric space has a limit that is also in the metric space. Corollary: A compact set is always closed. A closed subset
Or that every Cauchy sequence in M converges in M. of a compact set is compact.
Examples: Contraction: Let (X, d) be a metric space. A function Theorem: Suppose (X, d) is a complete metric space, which
∑n f : X → X is called a contraction if there exists a number for all r > 0 we can cover X by finitely many closed balls of
• d1 (x, y) = |xi − yi | (looks like unit diamond radius r > 0. Then X is compact.
i=1 α with 0 < α < 1 such that
∑ 1 Proof Sort Of: Suppose X is compact. Consider the open
• d2 (x, y) = ||x − y|| = ( n i=1 |xi − yi | )
n n
(looks like the
unit circle) d(f (x), f (y)) ≤ αd(x, y) for all x, y ∈ X cover given by {B(x, r) : x ∈ X}. This has a finite subcover
{B(xj , r) : 1 ≤ j ≤ n}. Then the closed balls of radius r with
• d∞ (x, y) = max1≤i≤n |xi −yi | (looks like the unit square) centres at xj cover X.
Banach’s Contraction Mapping Theorem: If (X, d) is a
Lemma: If X is compact then it is sequentially compact. i.e
Function Metric Spaces: on C([0, 1]) we have: complete metric space and if f : X → X is a contraction, then
every sequence in X has a convergent subsequence.
there is a unique point x ∈ X such that f (x) = x. This point
• ∫ Compact Functions: The direct image of a compact metric
1 x is called a fixed point of f .
space by a continuous function is compact. i.e let X and Y
d1 (f, g) := |f − g| Proof: Pick x0 ∈ X, and let x1 = f (x0 ), x2 =
0 be metric spaces, with X compact, and let f : X → Y be a
f (x1 ), ..., xn+1 = f (xn ). Consider d (xn+1 , xn ) = continuous surjective map. Then Y is compact.
• d (f (xn ) , f (xn−1 ) ≤ αd (xn , xn−1 )). Repeating we get
(∫ 1 )1/2 Continuity: For a mapping f : X → Y , it is continuous if:
d (xn+1 , xn ) ≤ αn d (x1 , x0 ).
d2 (f, g) := |f − g|2
0 So that when m ≥ n (∀ϵ > 0)(∀x ∈ X)(∃δ > 0) (∀x′ ∈ X)
• dX (x, x′ ) < δ =⇒ dY (f (x), f (x′ )) < ϵ
d∞ (f, g) = sup |f (x) − g(x)| d (xm , xn ) ≤ d (xm , xm−1 ) + . . . + d (xn+1 , xn )
x Uniform Continuity: For a mapping f : X → Y , it is uni-
( ) formly continuous if:
Strongly Equivalent: if between two metrics d and ρ on a ≤ αm−1 + . . . + αn d (x1 , x0 )
set X, there exists positive numbers A and B such that: (∀ϵ > 0)(∃δ > 0)(∀x ∈ X) (∀x′ ∈ X)
αn
≤ d (x1 , x0 ) dX (x, x′ ) < δ =⇒ dY (f (x), f (x′ )) < ϵ
d(x, y) ≤ Aρ(x, y) and ρ(x, y) ≤ Bd(x, y) for all x, y ∈ X 1−α
Since α < 1. This shows that (xn ) is a Cauchy sequence Proposition: Let f be a continuous mapping of a compact
Equivalent: if between two metrics d and ρ on a set X, for in X, and since X is complete, there exists x ∈ X to metric space X into a metric space Y . Then f is uniformly
every x ∈ X and every ϵ > 0 there exists a δ > 0 such that: which it converges. Now a contraction map is continuous, continuous.
so continuity of f at x shows that f (x) = f (limn→∞ xn ) = Cluster Point: a ∈ X is a cluster point iff Bδ (a) contains
d(x, y) < δ =⇒ ρ(x, y) < ϵ limn→∞ f (xn ) = limn→∞ xn+1 = x, so indeed f (x) = x. Fi- infinitely many points for each δ > 0
nally, if there are x, y ∈ X with f (x) = x and f (y) = y, we Bolzano-Weierstrass Property: X satisfies the Bolzano-
and have d(x, y) = d(f (x), f (y)) ≤ αd(x, y). Which since α < 1, Weierstrass Property iff every bounded sequence xn ∈ X has
ρ(x, y) < δ =⇒ d(x, y) < ϵ forces d(x, y) = 0 =⇒ x = y. a convergent subsequence.

4
 ∪∞ { 1 }
Heine-Borel: Let X be a separable metric space which sat- • Let (X, d) be a discrete metric space, then it is complete. • E = n=1 n × [0, 1] Is neither open nor closed.
isfies the Bolzano-Weierstrass Property, and H be a subset of Proof: Suppose (xn ) is Cauchy in (X, d). Take ϵ = 1. intE = ∅, ∂E = E = E ∪ {0} × [0, 1]. Since it is not
X. Then H is compact iff it is closed and bounded. Then ∃N such that for m, n ≥ N we have d(xm , xn ) < 1. closed it is not compact. It is not connected, we can
But this means for m, n ≥ N we have that xm = xn , i.e take U = {(x, y) : x < 3/4} and V = {(x, y) : x > 3/4}.
that xn is constant for n ≥ N . Hence (xn ) converges These are two open disjoint sets that separate E.
Metric Examples: and so (X, d) is complete.
∫ (∫ 2 )1/2 (∫ 2 )1/2
• Proof of Cauchy-Shwarz
∫ fg ≤ f g . Definitions
Just expand (f − αg) ≥ 0 and find the determinant
2 Spaces:
when it has at most one real root. • Injective: f (a) = f (b) =⇒ a = b
Interior: Let E be a subset of a metric space X. The interior
• On the complete metric space C([0, 1]) with the metric • Surjective: ∀y ∈ Y, ∃x ∈ X such that y = f (x)
of E is:
d∞ (f, g) = sup)t ∈ [0, 1]|f (t) − g(t)|. Consider the map- ∪ • Closed: A set F ⊂ X is closed iff the complement X\F
ping T : C([0, 1]) → C([0, 1]) given by E o := {V : V ⊆ E and V is open in X} is open. That is for any x ∈ X\F there is r > 0 such
∫ s
that B(x, r) ⊂ X\F .
T (f )(s) = f (t)(s − t)dt + g(s) Closure: Let E be a subset of a metric space X. The closure
0 of E is: • Open Ball: Let a ∈ X and r > 0. Then the open ball
Nor for any 0 ≤ s ≤ 1 we have: ∩ with centre a and radius r is the set:
∫ s E := {B : B ⊇ E and B is closed in X}
T (f )(s) − T (h)(s) = (f (t) − h(t))(s − t)dt B(a, r) := {x ∈ X : d(x, a) < r}
0 E = {x ∈ X : x is a limit point of E}
∫ s Theorem: • Closed Ball: Let a ∈ X and r > 0. Then the closed
|T (f )(s) − T (h)(s)| ≤ |f (t) − h(t)|(s − t)dt ball with centre a and radius r is the set:
• Eo ⊆ E ⊆ E
∫ s 0
∫ s
|f (t) − g(t)|(s − t)dt ≤ sup |f (t) − g(t)| (s − t)dt • If V is open and V ⊆ E, then V ⊂ E o B(a, r) := {x ∈ X : d(x, a) ≤ r}
0 0≤t≤1 0 • If C is closed and E ⊆ C, the E ⊆ C
s2 1 Boundary: let E ⊆ X, then the boundary of E is: • Support: The support of a real valued function is a
= d∞ (f, g) =⇒ d∞ (T f, T h) ≤ d∞ (f, h)
2 2 subset of the domain containing elements which are not
Since α < 1 T is a contraction =⇒ ∃ unique fixed point ∂E := {x ∈ X : ∀r > 0, Br (x) ∩ E ̸= ∅ and Br (x) ∩ E c ̸= ∅} mapped to zero. If the domain is a topological space,
f ∈ C([0, 1]) of T . the support is instead the smallest closed set containing
Theorem: Let E ⊆ X. Then ∂E = E\E o
• Function metrics d1 and d∞ aren’t strongly equivalent. all points not mapped to zero.
Separable: Metric space X is separable iff it contains a count-
Proof: fn (x) = xn , d1 (fn , 0) = 1/(n+1) and d∞ (fn , 0) = able dense subset. i.e iff there is a countable set Z of X such • A countable union of countable sets is countable.
1 for all n. that for every point a ∈ X there is a sequence xk ∈ Z such • Compact (set): Closed and bounded
• Show that f (x) = 2 + x−2 on [2, ∞) is a contraction that xk → a as k → ∞
mapping [2, ∞) into itself. Relatively Open: A set U ⊆ E is relatively open in E iff • Connected (set): If path connected. i.e connected
Proof: For x ≥ 0 we have f (x) ≥ 2 and hence the map there is a set V open in X such that U = E ∩ V space is a topological space that cannot be represented
maps [2, ∞) into itself. We check that f is a contraction. Relatively Closed: A set U ⊆ E is relatively closed in E iff as the union of two or more disjoint non-empty open
Clearly: there is a set C closed in X such that A = E ∩ C subsets. i.e
A subset A ⊂ X is connected if there does not exist open
1 1 and disjoint sets U, V ⊂ X such that
|f (x) − f (y)| = − 2 = f ′ (c) |x − y|
x2 y Spaces Examples:
A ∩ U ̸= ∅, B ∩ V ̸= ∅, and A ⊂ U ∪ V
for some c ∈ [2, ∞) by MVT. Since |f ′ (c)| = 2
≤ 1
<1
|c 3 | 4
the map is a contraction. • R is closed and open. intR = R, R = R, ∂R = ∅. Not
• Radius of Convergence: The radius of convergence
compact, but connected since path connected.
• If d and ρ are strongly equivalent, then they are equiv- R of the given power series is the unique number R such
alent. • Q is not open, not closed. intQ = ∅, Q = R, ∂Q = R. Not that the series converges for |x| < R and diverges for
Proof: If ϵ > 0 then choosing δ = ϵ/B works for the first compact, not connected. |x| > R. We have R ∈ [0, ∞) ∪ {∞} where when R = 0
statement, and δ = ϵ/A works for the second statement • A ⊂ R is connected iff A is any type of interval (open, the series only converges at x = 0 while R = ∞ means
for all x. So δ = min{ϵ/A, ϵ/B} works for both. closed or semi-open). that the power series converges for all x ∈ R.