5. Function Limits & Continuity

4.2 Functional Limits

Delta-Epsilon

Pasted image 20250302115758.webp|214

Let f:AR and let c be a limit point in the domain of A. limxcf(x)=L if

ε>0,δ>0,0<|xc|<δ|f(x)L|<ε

Topological

Let f:AR and let c be a limit point in the domain of A. limxcf(x)=L if

Vε(L),Vδ(c) so that x(Vδ(c){c})Af(x)Vε(L)

Sequential Definition

Let f:AR and let c be a limit point in the domain of A. limxcf(x)=L iff for all sequences (xn)A satisfying xnc and f(xn)L

Proof


Assume limxcf(x)=L, (xn)c and xnc. Let ε>0, there exists Vδ(c) with the property that all xVδ(c) different from c satisfy f(x)Vε(L).

(xn)c so there exists an xN after which xnVδ(c). nN implies f(xn)Vε(L)


Contrapositive
If limxcf(x)L then there exists one particular ε0>0 for which no matter what δ>0 there will be

xVδ(c), with xc for which f(x)Vε0(L)

If δn=1n for any nN there is a xnVδn(c) with xnc and f(xn)Vε0(L). Here (xn)c with xnc where f(xn) does not converge to L.

Algebraic Limit Theorems for Functions

Let f and g be functions defined on AR and assume that limxcf(x)=L and limxcg(x)=M for some limit point c of A. Then,
(i) limxckf(x)=kL for all kR
(ii) limxc[f(x+g(x)]=L+M
(iii) limxc[f(x)g(x)]=LM
(iv) limxc(f(x)g(x))=LM provided M0

Divergence Criterion

Let f be defined on A and let c be a limit point of A. If there exists two sequences (xn) and (yn) in A with xnc and ync but limxn=limyn=c but

limf(xn)limf(yn)

Then limxcf(x) does not exist


4.3 Continuous Functions

Continuity

f:AR is continuous at a point cA if for all ε>0 there exists a δ>0 such that whenever |xc|<δ (and xA), it follows that |f(x)f(c)|<ε

limxcf(x)=f(c)

ε-Neighbourhood Definition:
Vε(c), Vδ(c) so that xVδ(c) (and xA) f(x)Vε(f(c))

Notation:

Thm: Functional Continuity and Sequences

f is continuous at cA iff for all (xn)A that converge to c, f(xn)f(c)
contrapositive useful for showing not continuous at c

Examples:

Polynomials

a0+a1x+a2x2++anxn=p(x)

limxcp(x)=p(x) so, p is continuous on R.

Rational Polynomials

p(x)q(x)p(c)q(c) as xc

as long as q(c)0
Generally rationals are not continuous at only a finite number of points.

Drichlet

f(x){1xQ0xRQ

Discontinuous everywhere

Let cRQ. For all δ>0 there exists xQ(cδ,c+δ,) since Q is dense. Choose ε<1, then |f(x)f(c)|=1ε

Similar if cQ

does there exist a function which is discontinuous on a dense set and continuous on a dense set?

Thomae

f(x){0xRQ1nx=mnQ,gcd(m,n)=1

Pasted image 20250225095810.webp|251

Proof:

Let c=mn and gcd(m,n)=1

Let (xn) be a sequence of irrational numbers tending to c.
So f(xn)=0 and f(x)=1n
Hence f is not continuous at cQ.

Let cRQ let ε>0
Suppose xQ
Let NN s.t 1N<ε

For (c1,c+1) there are only finitely many mn with n<N

Hence there exists δ>0 s.t (cδ,c+δ) contains only mn with n>N. Hence, if mn(cδ,c+δ) then

f(mn)=1n<1N

So |f(mn)f(c)|<1n1N<ε

If x is irrational then f(x)=0 for for any ε>0 |f(x)f(c)|=0<ε

Q: Does there exist a function continuous on Q but discontinuous on RQ ?

NO

Algebraic Continuity Theorems

Assume: f:AR and g:AR are continuous at cA then:

i) kf(x) is continuous at c for all kR
ii) f(x)+g(x) is continuous at c
iii) f(x)g(x) is continuous at c
iv) f(x)g(x) is continuous at c

Example:

sinx,cosx are continuous on R

Tool: sinxsiny=2sin(12(xy))cos(12(x+y))

|sin(x)sin(c)|2|sin(12(xc))||cos(12(xc))|2|12(xc)||xc|

Given ε>0 choose δ<ε then

|sin(x)sin(c)|<ε

Know cosx=sin(x+π2)

Thm: Composition of CTS Functions

Given f:AR and g:BR, assuming that f(A)B. If f is continuous at cA and if g is continuous at f(c)B then gf is continuous at c


4.4 Continuous Functions on Compact Sets

Thm: Preservation of Compact Sets

f:AR is continuous. If KA is compact then f(K) is compact

Proof

WTS that any sequence in f(K) has a convergent subsequence whose limit is in f(K).
Let (yn) be a sequence such that {yn}n=1f(K). Each yn=f(xn) for at least one xnK. K is compacts so there is subsequence (xnk)lK. f(l)f(K). Since f is continuous f(xnk)f(l)

Fails if discontinuous
counter

f(x)={1xx[1,1]{0}0x=0

[1,1](R{(1,1)}){0}

Thm: Extreme Value Theorem (compact)

If f:KR is continuous on a compact set KR then f attains a maximum and minimum value. There exists x0,x1K such that f(x0)f(x)f(x1) for all xK

Proof

f(K) is compact, thus it is closed and bounded. It also contains its limit points.
For every nN take

supf(K)1n<(xn)<supf(K)

(xn)supf(K)supf(K) is a max and exists in f(K).

Uniform continuity

f:AR is uniformly continuous on A if for every ε>0 there exists δ>0 so that for all x,yA

|xy|<δ|f(x)f(y)|<ε

Pasted image 20250227093914.webp|408

Example

sin2x

Set δ=ε2

|f(x)f(x)|=|sin2xsin2c|2|xc|<ε

Equivalencies

f:AR
(i) f is not uniformly continuous
(ii) ϵ0>0 so that for all δ>0 there is an x,uA so that |xu|<δ but |f(x)f(u)|>ε0

(iii) ϵ0>0 and (xn),(yn) so that |xnyn|0 but |f(xn)f(yn)|>ϵ0

Proof

i) iii)
Negation is f is not uniformly continuous on A if and only if there exists a ε0>0 such that for all δ>0 we can find two points x and y satisfying |xy|<δ but with |f(x)f(y)|ε0. Thus, if we set δn=1n, then there exists point x1 and y1 where |x1y1|<δ but |f(x1)f(y1)|ε0.

Other direction is obvious.

Thm: Compact and Uniform

let f:KR be cts then f is uniform cts on [a,b]

Proof

Proof by contrapositive:
Suppose f is not uniformly continuous

Then ε>0 so that δn=1n there exists xn,yn[a,b] so that |xnyn|<δn but |f(xn)f(yn)|>ε

xn is bounded BW (xnk)zK

|ynkz||ynkxnk|0+|xnkz|0limynk=z

So (xnk),(ynk)z[a,b] but |f(xnk)f(ynk)|>ε so not cts at z

Thm: Continuous Extension

f:(a,b)R is uniform continuous then there exists a cts extensions f~ on [a,b]

Lipschitz

f:AR is Lipschitz on A if there exists k>0 so that

|f(x)f(y)|k|xy|

for all x,yA

Thm: Lipschitz then Uniformly CTS

4.4.9


4.5 Intermediate Value Theorem

Thm: Intermediate Value Theorem

If we have a continuous function f:[a,b]R. If LR, f(a)<L<f(b) or f(a)>L>f(b), then there exists a point c(a,b) where f(c)=L

Proof:

Lemma:
f:[a,b]R is cts. Suppose f(a)<0 and f(b)>0 then there exist x[a,b] with f(x)=0.
2.1 Bisection Method

Set a1=a and b1=b, xn=bnan2

  1. f(xn)<0an+1=a and bn+1=xn
  2. f(xn)>0bn+1=b and an+1=xn
  3. f(xn)=0an+1=bn+1=(xn)

an,bn,xn[aN,bN] for all nN

|anbn|12|an1bn1|12n1|ba|

so |anam|,|bnbm|,|xnxm||aNbN|
(an),(bn),(xn) is Cauchy converges

Converges to 0 by squeeze theorem.

This completes the proof of IVT using the Nested Interval Property.

Now take f(a)<L<f(b)
Take g(x)=f(x)L
g(a)<0 and g(b)>0

Better: Use NIP Similar idea:
f(a)<0<f(b). Let I0=[a,b] and:

z=a+b2

If f(z)0 then set a1=a and b1=z. If f(z)<0 then a1=z and b1=b. Continuing like above. InIn1, In and let cn=1In.

Suppose for contradiction that f(c)<L; then there must be some ϵ>0 so that Vϵ(f(c))<L. Since f is continuous, there must also be some δ so that f(Vδ(c))<L - but this contradicts our construction in that one endpoint of In is mapped to a number greater than L, no matter how large n gets and how small In is as a result. A similar argument shows f(c) cannot be larger than L, and therefore f(c)=L.

Example

f(x)=xex2
f is cts on [0,1] and f(0)=2 , f(1)=e2>0 so there exists a root

Example

f(x)=x2

If c2ε>0 then |x2c2|<ε iff

c2εc<xc<c2+εcc1=1c2=3ε=12

|xc1|<δ|f(x)f(c1)|<ε iff δ1+ε10.2
|xc2|<δ|f(x)f(c2)|<ε iff δ9+ε30.08