Upload
tai-pham
View
217
Download
0
Embed Size (px)
Citation preview
7/25/2019 MR-6
1/63
1
Chapter 6
Introduction to Calculus
Archimedes of Syracuse (c.287 BC c.212 BC ) Bhaskara II (1114 1185)
7/25/2019 MR-6
2/63
6.0 Calculus
2
Calculus is the mathematics of change.Two major branches: Differential calculusand Integral calculus, which arerelated by the Fundamental Theorem of Calculus.
Differential calculus determines varying rates of change. It is appliedto problems involving acceleration of moving objects (from a flywheelto the space shuttle), rates of growth and decay, optimal values, etc.
Integration is the "inverse" (or opposite) of differentiation. Itmeasures accumulations over periods of change. Integration can findvolumes and lengths of curves, measure forces and work, etc. Olderbranch: Archimedes (c. 287212 BC) worked on it.
Applications in science, economics, finance, engineering, etc.
7/25/2019 MR-6
3/63
6.0 Calculus: Early History
3
The foundations of calculus are generally attributed toNewtonandLeibniz, though Bhaskara II is believed to have also laid the basis of it.The Western roots go back to Wallis, Fermat, Descartes and Barrow.
Q: How close can two numbers be without being the same number?Or, equivalent question, by considering the difference of two numbers:How small can a number be without being zero?
Fermats and Newtons answer: The infinitessimal, a positive quantity,smaller than any non-zero real number.
With this concept differential calculus developed, by studying ratiosin which both numerator and denominator go to zero simultaneously.
7/25/2019 MR-6
4/63
4
6.1 Comparative Statics
Comparative statics: It is the study of different equilibrium statesassociated with different sets of values of parameters andexogenous variables.
Static equilibrium analysis: we start with y* =f(x)
Comparative static equilibrium analysis: y1* - y0* =f(x1) -f(x0)
(subscripts 0 and 1: initial & subsequent points in time)
Issues:
Quantitative & qualitative of change or
Magnitude & direction
The rate of change --i.e., the derivative (Y/ G)
7/25/2019 MR-6
5/63
5
We uses differential calculus to study what happens to anequilibrium in an economic model when something changes.Example: Macroeconomic Model
Given
Y = C + I0+ G
0
C = a+ b (Y-T)
T =d + tY
Solving for Y => Y* = (a bd + I0 + G0)/(1-(b(1-t)))
Question: What happens to Y* when an exogenous variable, say G0,changes in the model?
??;01
*0
*1*
0*1 =
=
GG
YYYY
6.1 Comparative Statics : Application
7/25/2019 MR-6
6/63
6
6.2 Rate of Change & the Derivative
Difference quotient:Let y =f(x)- Evaluate y=f(x) at two points: x0and x1: y0=f(x0) and y1=f(x1)- Define: x= x1- x0 => x1= x0+ x
y= y1- y0 => y=f(x0 + x) -f(x0)- Then, we define the difference quotientas:
Q: What happens tof(.)when x changes by a very small amount?
That is, we want to describe the small-scale behavior of a function
( ) ( )
x
xfxxf
x
y o
+=
0
7/25/2019 MR-6
7/63
7
x
yxf
=)('
Difference quotient:
To describe the small scale behavior of a function, we use thederivative.
Derivative(based on Newtons and Leibnizs approach):Lets take xas an infinitesimal (a positive number, but smaller thanany positive real number) in the difference quotient:
( ) ( )x
xfxxf
x
y o
+=
0
Another interpretation: Infinitesimals are locations which are notzero, but which have zero distance from zero.
6.2 Rate of Change & the Derivative
7/25/2019 MR-6
8/63
8
The derivative captures the small-scale behavior of a function i.e.,
what happens tof(.)when x changes by a very small amount.
But, the infinitesimal approach was not elegant. Definition andmanipulation of infinitesimals was not very precise.
DAlembert started to think about vanishing quantities. He sawthe tangent to a curve as a limit of secant lines. This was arevolutionary, though graphical, approach:
As the end point of the secant converges on the point of tangency, itbecomes identical to the tangent in the limit.
6.2 Rate of Change & the Derivative
7/25/2019 MR-6
9/63
9
As the lim of x0, then f(x) measures the tangent (rise/run) off(x) at the initial point A. Again, the secant becomes the tangent.
This is the traditional motivation in calculus textbooks.
x
y
y
x0 x1 x2
A
C B
Graphical interpretation of the derivative:Slope = tangent of the function at x=x0:
Jean d'Alembert (17171783)
6.2 Rate of Change & the Derivative
7/25/2019 MR-6
10/63
10
( ) ( )dx
dy
x
xfxxf
x
yxf
xx=
=
=
00
00limlim)('
The graphical interpretation is subject to objections like those seenin Zenos paradoxes (say, Achilles and the Tortoise).
Based on the work of Cauchy and Weierstrass, we use limits toreplace infinitesimals. Limits have a precise definition and nice
properties. Then, we have an easier definition to work with:
Derivative(based on Weierstrasss approach):Lets take limits (x0) in the difference quotient we get the
derivative:
6.2 Rate of Change & the Derivative
7/25/2019 MR-6
11/63
11
VelocitySupposef(t) is the position of an object at time t, where a t b.
The average velocity, or average rate of change of fwith respect to t,of the object from time ato time bis:
Average velocity = change in position/change in time= [f(b) -f(a)] /(b-a)
If b= a +t => average velocity = [f(a + t) -f(a)]/t
Instantaneous velocityThe instantaneous velocity, or the instantaneous rate of change of fwith respect to t, at t = ais
lim t0[f(a +t) -f(a)]/t (provided the limitexists)
6.2 Rate of Change & the Derivative: Physics
Interpretation
7/25/2019 MR-6
12/63
12
Product of LaborSupposeQ=f(L) is a production function, where L is the only input,labor. Let L1= L0+L,then, we define:
Average product of labor = Change in production/change in labor
= [f(L1) -f(L0)] /(L1- L0)
Let L0= 0 andf(L0)=0. Define L1=L= 1 hour=> labor productivity =f(L1) /hour
Marginal product of laborThe marginal product of labor at L= L0is
lim L0[f(L0+L) -f(L0)]/L (if the limit exists)
6.2 Rate of Change & the Derivative: Economic
Interpretation
7/25/2019 MR-6
13/63
13
bx
yxf
bx
bxaxxba
x
y
x
yb
bbxay
x=
=
=
+++=
===
=+=
0lim)(')4
)()()3
in xchange
yinchange
run
rise)2
slope)1
6.2 Rate of Change & the Derivative: Linear
function
What are the secant slope and the derivative for a linear function?
7/25/2019 MR-6
14/63
14
( )
( ) ( )
( ) ( )
00
0
2
0
2
0
2
0
2
0
2
0
2
00
2
00
2
6lim)('
36434363
4343
43
43
43:Example
xx
yxf
xx
x
xxxxx
x
xxx
x
y
xxxxf
xxf
xy
x
=
=
+=
+++=
+=
+=+
=
=
6.2 Rate of Change & the Derivative: Power
function
7/25/2019 MR-6
15/63
15
Example (continuation):y= 3x2 4 (red)
Evaluate difference quotient at x=3, when x= 4y/x= 6x+ 3x => y/x= 30
y= 30x 67, secant through pts (3,f(3), 7, f(7)) (blue)
6.2 Rate of Change & the Derivative: Power
function
7/25/2019 MR-6
16/63
16
Now, evaluate derivative at x=3y= 3x2 4 (red)f (x=3)=18y = 18x 31, tangent at point (3,f(3)) (blue)
6.2 Rate of Change & the Derivative: Power
function
7/25/2019 MR-6
17/63
17
( ) ( )dx
dy
x
xfxxf
x
yxf
xx=
+=
=
00
00limlim)('
Standard Calculus:Based on Weierstrasss approach, using limits instead ofinfinitesimals. The infinitely small behavior of the function is foundby taking the limiting behavior for smaller and smaller numbers.Limits are the easiest way to provide rigorous foundations for
calculus, and for this reason they are the standard approach.
Nikolai Luzin (1883 1950, Russia) "They won't fool me: it's simply theratio of infinitesimals, nothing else."
6.2 Standard & Non Standard Calculus
7/25/2019 MR-6
18/63
18
( ) ( )
+=
x
xfxxfstxf 00)('
Non-Standard Calculus:Due to the work of Abraham Robinson (1918-1974,Germany/Israel), infinitesimals received a precise definition.Robinson used the hyperreals, *R, (an extension of the real numbers).that contains numbers greater than anything of the form
1 + 1 +... + 1.Such a number is infinite, and its inverse is infinitesimal. Then,
for an infinitesimal x. st(.) denotes the standard part function, whichassociates to every finite hyperreal the unique real infinitely close to it roughly speaking, the part that ignores the error term (i.e., the term
with x).
6.2 Standard & Non Standard Calculus
7/25/2019 MR-6
19/63
19
In the one dimensional case, we define the derivative as:
6.2 Differentials
( ) ( )x
xfxxfxf
x
+=
00
0lim)('
Equivalently, we can write
f(x0+x)=f(x0) +xf(x)
+ rx(x),where rx(x) represent the remainder, which is of smaller orderthanx. That is,
( ).0lim
0=
x
xrxx
The quantityf(x0+x) -f(x0) is composed of two terms:- xf(x), the part proportional to the change in x(x)- rx(x), an error, which gets smaller with x.
The expression df(x) = xf(x) is called the (first) differential off.
7/25/2019 MR-6
20/63
Figure 6.9 Differential Approximation and ActualChange of a Function
The differential df(x) = xf(x) is the linear part of the incrementf(x0+x) -f(x0). This is expressed by geometrically replacing thecurve at point x0by its tangent.
7/25/2019 MR-6
21/63
21
6.2 Differentials and Approximations
The differential is used to linearly approximate changes in f(x). Theerrori.e., the quality of the approximation- depends on thecurvature of isf(x)and, of course, the magnitude of x.
For very smallx, the approximation should be good, regardless off(x). But, when is xvery small?
An interesting case is when x=1. In this case, df(x) =f(x). Then,the first derivative approximates the change in the function peradditional unit of x.
In the production function example,f(L)measures the additionaloutput that can be produced with an additional unit of labor.
7/25/2019 MR-6
22/63
Figure 6.11 Differential Approximation forBeta with Different Functions
Note: Since x=1 in all cases, dy=f(x). As the function has morecurvature, the linear approximation becomes less precise.
7/25/2019 MR-6
23/63
23
Recall the solution to Y (income) in the Macroeconomic model:
We have a linear function in I (investment) and G (governmentspending) . Assume I is fixed, then we have y=f(G).
Comparative Static Question:
What happens to Y* when G increases by G? We approximate theanswer by:
0
)1(1
1)('where
);('*
>
=
tb
Gf
GfGY
)()1(1
1),( 00 GIbda
tbGIfY ++
==
6.2 Differentials and Approximations: Example
=> if G=$1, then dY=f(G).
7/25/2019 MR-6
24/63
24
6.2 Multivariate Calculus: Partial Differentiation
1...ni,limgeneral,In
)xw.r.t.derivativepartial(
),,,(),,,(limlim
),,,(
0
11
1
1
21211
01
0
21
11
=
+=
=
i
iix
nn
xx
n
fx
y
x
y
fx
y
x
xxxfxxxxf
x
y
xxxfy
i
It is straightforward to extend the concepts of derivative anddifferential to more than one variable. In this case,ydepends onseveral variables: x1, x2, , xn.
The derivative of yw.r.t. one of the variables while the othervariables are held constant- is calledpartial derivative.
7/25/2019 MR-6
25/63
25
6.2 Partial derivatives: Example
( )
( )
==
=
=
=
1
1
LAKQ:functionProduction
LAKLQMPL
LAKK
QMPK
Interpretation: As usual:MPL: Marginal product of labor
MPK: Marginal product of capital
We use them to linearly approximate the change in production Q inthe face of unitary changes, one at a time, of inputs. When L and Kchange simultaneously, we need to use the total derivative:
Q MPL L+ MPK K
Example:: Cobb-Douglas production function
7/25/2019 MR-6
26/63
26
6.3 Concept of Limit: Preliminaries
Definition:N-ballLet c be a point in Rnand rbe a positive number. The set of all points x Rn
whose distance is less than ris called an n-ballof radius rand center c. It is
usually denoted by B(c) or B(c,r). Thus:
B(c,r)={x: x Rn,x-c
7/25/2019 MR-6
27/63
27
6.3 Concept of Limit: Preliminaries
Definition: Boundary Point(d) if every n-ball B(x) contains at least one point of S and at least one point of
the complement of S, then xis called a boundary pointof S. The set of all
boundary points of S is called the boundary of S.
Note: Every non-isolated boundary point of a set SRis an accumulationpoint of S. An accumulation point is never an isolated point
Example: Lets look at the interval (0, 4).
- The boundary of (0, 4) is the set consisting of the two elements {0, 4}. - Theinterior of the set (0, 4) is the set (0, 4) -i.e., itself.
- No points of either set are isolated, and each point of the set is anaccumulation point. The same is true, incidentally for each of the sets (0, 4), [0,
4), (0, 4], and [0, 4].
7/25/2019 MR-6
28/63
28
6.3 Concept of Limit: Topology
Definition: Open and Closed setsA set S in Rnis said to be:
(a) openif all its points are interior points
(b) closedif it contains all its accumulation points
(c) boundedif there is a real number r>0 and a point cin Rn
such that S liesentirely within the n-ball B(c,r)
(d) compactif it is closed and bounded
Examples: Let A be an interval in R. For a
7/25/2019 MR-6
29/63
29
6.3 Concept of Limit: Topology
Properties:- Every union of open sets is again open.
- Every intersection of closed sets is again closed.
- Every finite intersection of open sets is again open
- Every finite union of closed sets is again closed.Note: Open sets in Rare generally easy. Closed sets can get complicated.
Cantor Middle Third Set
Start with the unit interval S0= [0, 1].Remove from S0the middle third. Set S1= S0\(1/3, 2/3)
Remove from S1the 2 middle thirds. Set S2=S1\{ (1/9, 2/9) U (7/9, 8/9)}.
Continue, where Sn+1=Sn\{ middle thirds of subintervals of Sn}.
Then, the Cantor set Cis defined as C= Sn
7/25/2019 MR-6
30/63
30
6.3 Concept of Limit: Preliminaries
The Cantor set C is an indication of the complicated structure of closed setsin the real line. C has the following properties:- C is compact (i.e. closed and bounded)
- C is perfect i.e., it is close and every point of C is an accumulation point of
C.
- C is uncountable (since every non-empty perfect set is uncountable).
- C has length zero, but contains uncountably many points.
- C does not contain any open set.
This set is used to construct counter-intuitive objects in real analysis or to
show lack of generalization of some results. For example, Riemann integration
does not generalize to all intervals.
7/25/2019 MR-6
31/63
31
6.3 Concept of Limit: Functions
Definition: Functions
Let S and T be two sets. If with each element xin S there is associated exactly
one elementyin T, denotedf(x), thenfis said to be afunctionfrom S to T. We
write
f: S T,
and say thatfis defined on S with values in T. The set S is called the domainoff; the set of all values offis called the rangeoff, and it is a subset of T. T is
called the targetor codomain.
The imageoffis defined asimag(f)= {tT : there is an sS withf(s) = t}.
If C is a subset of the range T, then thepreimage, or inverse image, of C under the
functionfis the set defined as
f
-1
(C) = {x
S :f(x)
C }
7/25/2019 MR-6
32/63
32
6.3 Concept of Limit: Functions
Example: Domain and image off :X Y
fis a function from domainXto codomain Y. The smaller oval inside Yis the
image off.
Peter G. Lejeune Dirichlet (1805 1859, Germany)
7/25/2019 MR-6
33/63
33
6.3 Concept of Limit: Functions
A functionf: S Rdefined on a set S with values in Ris called real-valued.A functionf: S Rm(m>1) whose values are points in Ris called a vectorfunction.
A vector function is bounded if there is a real number M such that
f (x) M for all xin S.
A functionffrom StoTcan be classified into three groups:
- One-to-oneif wheneverf(s) = f(w),then s=w. Also called injections.
- Ontoif for all tT there is an sS such thatf(s) = t. Also called surjections.
- Bijectionif it is one-to-one and onto -i.e., bijections are functions that are
injective and surjective.
Examples: A linear function is a bijection. A periodic function is not one-to-
one. Say,g(x) = cos(x)is neither one-to-one nor onto in R.
7/25/2019 MR-6
34/63
34
6.3 Concept of Limit: Inverse Functions
When f: S T is one-to-one on a set Cin S, there is a function fromf(C) back to C,which assigns to each tf(C) the unique point in Cwhich mapped to it. This map is called the inverse of fon Cand it writtenas:
f-1:f(C) C.
Examples:
- Letf: R T, sayf = 3x+2 => f-1: (y-2)/3
- The logarithm is the inverse of the exponential function.- The demand function q=D(p), under the usual assumptions, has as theinverse functionp=D-1(q),which is called the inverse demand function.
7/25/2019 MR-6
35/63
35
6.3 Concept of Limit: Composition Functions
Let f: S T andg: V W be two functions. Suppose that T is asubset of V.Then, the composition of f with g is defined as the function:(g f)(x)= g(f(x)) for all x in S.
That is,function compositionis the application of one function to theresults of another. The functionsfandg can be composedby computingthe output ofgwhen it has an argument off(x) instead of x. Intuitively,if zis a functiong(y)andyis a functionf(x), then zis a function h(x).
Example: Definef(x) = x5andg(x)= exp(x). Then, (g f)(x) = exp(x5)
7/25/2019 MR-6
36/63
36
6.3 Concept of Limit: Sequences
Definition: Sequence
A sequenceof real numbers is a functionf: N R.That is, a sequence can be written asf(1), f(2), f(3), ..... Usually, we will denote
such a sequence by the symbol {aj} where aj= f(j).
Example: The sequence 1, 1/2, 1/3, 1/4, 1/5, ...is written as {1/j}.
Definition: Convergence
A sequence {aj} of real (or complex) numbers is said to convergeto a real (or
complex) number cif for every >0, there is an integerN>0 such that ifj> N,then
| aj- c | < .The number cis called the limit of the sequence {aj} and we write ajc.
If a sequence {aj} does not converge, then we say that it diverges.
7/25/2019 MR-6
37/63
37
6.3 Concept of Limit: Sequences
Example: The sequence {1/j} converges to zero.
We need to show that no matter which >0 we choose, the sequence willeventually become smaller than this number. Take any >0. Then, there existsa positive integerNsuch that 1/N Nwe have:
| 1/j - 0 | = | 1/j | < 1/N < , wheneverj > N.This is precisely the definition of the sequence {1/j}converging to zero.
Note: Easy proof. A proper choice ofNis the key.
If {aj} is a convergent sequence, {aj} is bounded, and the limit is unique.
Example: The sequence of Fibonacci numbers is unbounded. Then, the
sequence cannot converge, since a convergent sequence must be bounded.
7/25/2019 MR-6
38/63
38
6.3 Concept of Limit: Sequences
Algebra of Convergent Sequences:Let {aj} be a convergent sequence. Then, the sequence is bounded, and the
limit is unique.
Suppose {aj} and {bj} are converging to aand b, respectively. Then
- Their sum converges to a + b, and the sequences can be added term by term.
- Their product converges to a * b, and the sequences can be multiplied termby term.
- Their quotient converges to a/b, provide that b0, and the sequences can bedivided term by term (if the denominators are not zero).
- If an bnfor all n, then a b. (It does not work for strict inequalities).
We know how to work with convergent sequences, we would like to have aneasy criteria to determine whether a sequence converges.
7/25/2019 MR-6
39/63
39
6.3 Concept of Limit: Sequences
Definition: Monotonicity
A sequence {aj} is calledmonotone increasing if aj + 1 ajfor allj.
A sequence {aj} is calledmonotone decreasing if aj aj + 1for allj.
Proposition:Monotone Sequences
- If {aj} is a monotone increasing sequence that is bounded above, then the
sequence must converge.
- If {aj} is a monotone decreasing sequence that is bounded bellow, then the
sequence must converge.
Example:
- {j/(j+1)} is monotone increasing & bounded above by 1. It must converge.
- {1/j} is monotone decreasing & bounded bellow by 0. It must converge.
7/25/2019 MR-6
40/63
40
Often, it is hard to determine the actual limit of a sequence. We want to
have a definition which only includes the known elements of a particularsequence and does not rely on the unknown limit.
Definition: Cauchy Sequence
Let {aj} be a sequence of real (or complex) numbers. We say that {aj} isCauchy if for each >0 there is an integerN>0 such that ifj, k > Nthen
|aj-ak|< . Now, we know what it means for the elements of a sequence to get closer
together, and to stay close together.
Theorem: Completeness Theorem in R.
Let {aj} be a Cauchy sequence in R. Then, {aj} is bounded.
Let {aj} be a sequence in R. {aj} is Cauchy iff it converges to some limit a.
6.3 Concept of Limit: Cauchy Sequence
7/25/2019 MR-6
41/63
41
By considering Cauchy sequences instead of convergent sequences we do
not need to refer to the unknown limit of a sequence (in effect, both conceptsare the same).
Q: Not all sequences converge. How do we deal with these situation?
A: We change the sequence into a convergent one (extract subsequences) andwe modify our concept of limit (lim supand lim inf).
Definition: Subsequence.
Let {aj} be a sequence. When we extract from this sequence only certainelements and drop the remaining ones we obtain a new sequences consisting
of an infinite subset of the original sequence. That sequence is called a
subsequenceand denoted by {aj,k} (k=1, 2, ...,).
Note: We can think of a subsequence as a composition function.
6.3 Concept of Limit: Subsequences
7/25/2019 MR-6
42/63
42
Example: Take the sequence {(-1)j} , which does not converge. The sequence
is: {-1, 1, -1, 1,...}
Extract every even number in the sequence, we get: {-1, -1, -1, -1,...}
=> subsequence converges to -1.
Extract every odd number in the sequence, we get: {1, 1, 1, 1,...}
=> subsequence converges to 1.
Note: We can extract infinitely many subsequences from any given sequence
Proposition: Subsequences from Convergent Sequence
Let {aj} be a convergent sequence, then every subsequence of {aj} converges
to the same limit.
Let {aj} be a sequence such that every possible subsequence extracted from
{aj} converge to the same limit, then {aj} also converges to that limit.
6.3 Concept of Limit: Subsequences
7/25/2019 MR-6
43/63
43
Theorem: Bolzano-Weierstrass
Let {aj} be a sequence of real numbers that is bounded. Then, there exists a
subsequence {aj,k} that converges.
This is one on the most important results of basic real analysis, and
generalizes the above proposition. It explains why subsequences can beuseful, even if the original sequence does not converge.
Example: The sequence {sin(j)} does not converge, but since it is bounded,
we can extract a convergent subsequence.
Note: The Bolzano-Weierstrass theorem does guaranty the existence of thatsubsequence, but it does not say how to obtain it. It can be difficult. We will
extend the concept of limits to deal with divergent sequences.
6.3 Concept of Limit: Subsequences
7/25/2019 MR-6
44/63
44
Definition: Lim Sup and Lim Inf
Let {aj} be a sequence of real numbers. Define
Aj= inf{aj, aj + 1, aj + 2, ...}
and let c = lim (Aj). Then cis called the limit inferiorof the sequence {aj}.
Let {aj} be a sequence of real numbers. Define
Bj= sup{aj, aj + 1, aj + 2, ...}
and let d = lim (Bj). Then dis called the limit superiorof the sequence .
Summary::
- lim inf(aj) = lim(Aj), whereAj= inf{aj, aj + 1, aj + 2, ...}
- lim sup(aj) = lim(Bj), where Bj= sup{aj, aj + 1, aj + 2, ...}
These limits are often counter-intuitive, they have one very useful property:lim supand lim infalways exist (possibly infinite) for any sequence of real
numbers. Moreover, there is a subsequence converging to cand another to d.
6.3 Concept of Limit: Lim Sup and Lim Inf
7/25/2019 MR-6
45/63
45
Example 1: Consider {(-1)j}. We find the numbersAj= inf{aj, aj + 1, aj + 2, ...}
A1= inf{-1, 1, -1, 1, ...} = -1
A2= inf{1, -1, 1, -1, ... } = -1
etc. It is clear that lim inf{(-1)j}= -1. (also the infimum)
Similarly, lim sup{(-1)j}= 1. (also the supremum)
Example 2: Consider {1/j}. The sequence is {1, 1/2, 1/3, 1/4, ...}. Then, the
infimum is zero, while the supremum is 1. Lets get theAjand BjA1= inf{1, 1/2, 1/3, 1/4, ...}= 0 & B1=sup{1, 1/2, 1/3, 1/4, ...}= 1
A2= inf{1/2, 1/3, 1/4, 1/5, ...}= 0 & B2=sup{1/2, 1/3, 1/4, 1/5,.}=1/2
A3= inf(1/3, 1/4, 1/5, 1/6, ...} = 0 & B3=sup( 1/3, 1/4, 1/5, 1/6,.}= 1/3
etc. It is clear that lim inf{1/j} = 0. (also the infimum)
etc. It is clear that lim sup{1/j} = 0. (different from the supremum)
6.3 Concept of Limit: Lim Sup and Lim Inf
7/25/2019 MR-6
46/63
46
6.3 Concept of Limit: Definition
Definition: LimitLetf: S Rm. Let cbe an accumulation point of S. Suppose thereexists a point bin Rmwith the property that for every >0 there is a>0 such that
f (x) - b < for all xin S, xc, for which x-c < .Then, we say the limitoff (x) is b, as xtends to c, and we write
Note: This is the (,)-definitionof limit, introduced by Bolzano/Cauchyand perfected by Weierstrass.
bxfcx
=
)('lim
7/25/2019 MR-6
47/63
47
6.3 Concept of Limit: Right- & Left-hand Limit
The limit (f(x), x
c, direction) function attempts to compute thelimiting value off(x) as xapproaches cfrom left, c-(the left-hand limit) orright, c-+(the right-hand limit).
When the left-hand and the right-hand limits are equal, say to L, wesay the limit exists and equals L.
If q =f(v), what value does q approach as v N? Answer: LAs vN from either side, q L.
Then, both the left-side limit and the
right side-limit are equal.
Therefore, lim q= L.
q
L
vN
v
N N v
7/25/2019 MR-6
48/63
48
6.4 Evaluation of a Limit
To take a limit, substitute successively smaller values that tend toNfrom both the left and right sides sinceNmay not be in thedomain of the function.
If vis in both the numerator and denominator remove it fromeither depending on the function
Taking limits sometimes is not straightforward.
Example: Given q= (2v+ 5)/(v+ 1), find the limit of qas v+.Dividing the numerator by denominator:
2lim
1
32
1
52
=
++=
+
+=
+q
vv
vq
v
7/25/2019 MR-6
49/63
49
6.4 Limit Theorems
If q = av+b, => = aN+b
If q= g(v) = b, => = b
If q= v, => =N
If q= vk, => =Nk
)(lim)(lim)(lim 2121 qqqqNvNvNv
=
qNv
lim
qNv
lim
qNv
lim
qNv
lim
)(lim)(lim)(lim 2121 qqqqNvNvNv
=
)(lim/)(lim)/(lim 2121 qqqq NvNvNv =
Example: Find lim (1+v)/(2 + v) as v0
( )
( ) 2
1
2lim
1lim
0
0
2
1 =
+
+=
v
v
L
L
v
v
7/25/2019 MR-6
50/63
50
6.4 LHospitals Rules
Iffandgare differentiable in a neighborhood of x=c, andf(c)=g(c)=0,then
provided the limits exist.
Note: The same result holds for one-sided limits.
Iffandgare differentiable and limxf(x) = limxg(x) = , then
provided the last limit exists.
In other situations L'Hospital's rules may also apply, but often aproblem can be rewritten so that one of these two cases will apply.
)('
)('lim
)(
)(lim
xg
xf
xg
xf
cxcx =
)('
)('lim
)(
)(lim
xg
xf
xg
xf
xx =
7/25/2019 MR-6
51/63
51
6.4 Limit Jokes
qNv
lim
7/25/2019 MR-6
52/63
52
Definition: Continuous function
When a function q=f(v) possesses a limit as vtends to the pointNin the domain and
When this limit is also equal tof(N) --i.e., the value of thefunction at v=N--, then the function is continuous inN.
Note: Continuity preserves limits
Requirements for continuity
N must be in the domain of the function
f, q = f(v)f has a limit as vN
limit equalsf(N) in value
( )
)()(lim
exits)(lim
defined
Nfvf
vf
Nf
Nv
Nv
=
6.5 Continuityand Differentiability of a Function
7/25/2019 MR-6
53/63
53
Alternative definition: Continuous functionLetf: SR be a real valued function on a set S in Rn. Let cbe apoint in S. We say thatfis continuousat cif for every >0, thereexists a >0 such that
f(c+u)f(c)
7/25/2019 MR-6
54/63
54
If cis an accumulation (limit) point of S, the definition of continuityimplies that
limu0f(c +u) = f(c).
Intuition is tricky: Geometry seems to show that iffis continuous
at c, it must be continuous near c. This is wrong!
Example: The Dirichlet function
Letf: RR defined by
f(x) = x if xis rational= 0 if xis irrational
is continuous at x=0, but at no other point.
6.5 Continuityand Differentiability of a Function
7/25/2019 MR-6
55/63
55
6.5 Continuityand Differentiability of a Function
Almost all the basic functions in mathematical econ models areassumed to be continuous.
For example, a production function is continuous if a small changein inputs yields a small change in output. (A reasonable assumption.)
If a function fails to be continuous at a point c, then the function iscalled discontinuousat c, (c is called apoint of discontinuity).
Examples: Non-continuous functions:
7/25/2019 MR-6
56/63
56
We sayf: RnRis differentiableat x* iff(x*) exists. Not every continuous function has a derivative at every point. For
example: f(x)=|x|.
|x|is not differentiable at x=0, the left and the right-handed
limits are different (-1 and +1). Then, there is no unambiguoustangent line defined at x=0.
We need the functionf(.) to be smoothi.e., no kinks.
Iff is differentiable at x*, thenfis continuous at x*. (Converse
is, of course, not true.)
As with continuous functions, differentiable functions can beadded, multiplied, divided, and composed with each other to yield
again differentiable functions.
6.5 Continuityand Differentiability of a Function
7/25/2019 MR-6
57/63
57
A functionf: Rn
Ris continuously differentiableon an open set Uof Rn
if and only if for each x,df/dxiexists for all xin Uand is continuousin x.
This rational function is not defined at v= 2, even though the limit
exists as v2. It is discontinuous and thus does not havecontinuous derivatives --i.e., it is not continuous differentiable.
Thisimagecannotcurrentlybedisplayed.
4
442
23
+=
v
vvvq
This continuous function is not differentiable at x=3 and, thus, doesnot have continuous derivatives (it is not continuously differentiable):
>
=
)3(,1
)3(,5
xwherex
xwherexy
6.5 Continuityand Differentiability of a Function
7/25/2019 MR-6
58/63
not continuous:
f(t) = -1 if t= 0f(t) = t2otherwise
continuousbut notdifferentiable:
f(t) = |t|
continuous& differentiable:
f(t) = t2
t
f(t)
6.5 Continuityand Differentiability of a Function
x
7/25/2019 MR-6
59/63
59
For a function to be continuous differentiable
All points in the domain off defined
The limit is taken on the difference quotient at x=x0 as x0 fromboth directions. The continuity condition is necessary, not sufficient.
The differentiability condition (smoothness) is both necessary andsufficient for whetherfis differentiable.
Theorem: Rolles Theorem
Iffis continuous on [a, b] and differentiable on (a, b), andf(a)=f(b)=0,then there exists a number xin (a, b)such thatf'(x) = 0.
Note: An extension of Rolle's theorem that removes the conditions onf(a)andf(b)is the Mean-Value Theorem. These theorems form the
basis for the familiar test for local extrema of a function.
6.5 Continuityand Differentiability of a Function
7/25/2019 MR-6
60/63
60
Here we want to list some functions that illustrate more or lesssubtle points for continuous and differentiable functions.
Dirichlet function: A function that is not continuous at any point in R
Countable discontinuities: A function that is continuous at the irrational
numbers and discontinuous at the rational numbers.
C1function: A function that is differentiable, but the derivative is not
continuous.
6.6 Continuityand Differentiability: Examples
=
=irrationalisif,0
rationalis/if,/1
x
qpxqy
=
=
0if,0
0if),/1sin(2
x
xxxy
7/25/2019 MR-6
61/63
61
Cnfunction: A function that is n-times differentiable, but not (n+1)-times differentiable
Cinffunction: A function that is not zero, infinitely oftendifferentiable, but the n-th derivative at zero is always zero.
Weierstrass function: A function that is continuous everywhere and
nowhere differentiable in R.
Cantor function: A continuous, non-constant, differentiable function
whose derivative is zero everywhere except on a set of length zero
6.6 Continuityand Differentiability: Examples
3 13 += nxy
=
=
0if,0
0if),/1exp(2
x
xxy
|)4sin(|)4/3(0
xy n
n
n
=
=
7/25/2019 MR-6
62/63
62
6.7 The Limit and the Quotient Ratio
Let qy/x and vx such that q =f(v) and
qx
y
vx 00
limlim
=
Q: What value does variable qapproach as vapproaches 0?
A: If the function is differentiable, we move from the quotient ratio tothe derivative.
Note: This definition may not work well when x is a vector, say x (y,z).Measuring y is not a problem (the difference between twofunctions), but measuring x (y, z). is not clear.
We need the concept of distance for a vector, a norm.
7/25/2019 MR-6
63/63
6.8 Resolution of a Controversy: Butter Biscuits
or Fruit Chewy Cookies?