You are on page 1of 197

z=f(x,y)

x
z
y
MATH 234
THIRD SEMESTER
CALCULUS
Spring 2014
1
2
Math 234 3rd Semester Calculus
Lecture notes version 0.9(Spring 2014)
Tis is a self contained set of lectuie notes foi Math 234. Te notes weie wiiuen by
Siguid Angenent, some pioblems weie taken fiom Guichaids open calculus text which
is available at http://www.whitman.edu/mathematics/multivariable/src/
Te L
A
T
E
X les, as well as the Pv1uoN and lNxscvisvc les that weie used to pio-
duce the notes befoie you can be obtained fiom the following web site
http://www.math.wisc.edu/~angenent/Free-Lecture-Notes
Tey aie meant to be fieely available foi non-commeicial use, in the sense that fiee
sofwaie is fiee. Moie piecisely
Copyiight (c) 2009 Siguid B. Angenent. Peimission is gianted to copy, distiibute and/oi modify this
document undei the teims of the GNU Fiee Documentation License, Veision 1.2 oi any latei
veision published by the Fiee Sofwaie Foundation, with no lnvaiiant Sections, no Fiont-Covei
Texts, and no Back-Covei Texts. A copy of the license is included in the section entitled GNU Fiee
Documentation License.
Contents
Chapter 1. Vector Geometry in Three dimensional space 5
1. Three dimensional space 5
2. Geometric description of vectors 5
3. Arithmetic of vectors 6
4. Vector algebra 7
5. Component representation of vectors 8
6. The dot product 9
7. The cross product 10
8. The triple product 12
9. Determinants 13
10. Determinants, the triple product, and the cross product 13
11. Defining equations for lines and planes 14
12. Problems 16
Chapter 2. Parametric curves and vector functions 19
1. Vector functions 19
2. Using vector functions to describe motion 19
3. Lines 20
4. Circular motion 20
5. The cycloid 21
6. The helix 21
7. The derivative of a vector function 22
8. The derivative as velocity vector 23
9. Acceleration 24
10. The dierentiation rules 25
11. Vector functions of constant length 26
12. Two examples 27
13. Arc length 28
14. Arc length derivative 29
15. Unit Tangent and Curvature 30
16. Osculating plane 31
17. Problems 31
Chapter 3. Functions of more than one variable 35
1. Functions of two variables and their graphs 35
2. Linear functions 38
3. adratic forms 39
4. Functions in polar coordinates r, 42
5. Methods of visualizing the graph of a function 44
Problems 46
Chapter 4. Derivatives 49
1. Interior points and continuous functions 49
2. Partial Derivatives 50
3. Problems 51
4. The linear approximation to a function 52
5. The tangent plane to a graph 55
3
4 CONTENTS
6. The Two Variable Chain Rule 58
7. Problems 61
8. Gradients 62
9. The chain rule and the gradient of a function of three variables 66
10. Implicit Functions 69
Problems 72
11. The Chain Rule with more Independent Variables;
Coordinate Transformations 73
12. Problems 75
13. Higher Partials and Clairauts Theorem 78
14. Finding a function from its derivatives 79
15. Problems 81
Chapter 5. Maxima and Minima 83
1. Local and Global extrema 83
2. Continuous functions on closed and bounded sets 83
3. Problems 85
4. Critical points 86
5. When there are more than two variables 89
6. Problems 91
7. A Minimization Problem: Linear Regression 92
8. Problems 93
9. The Second Derivative Test 94
10. Problems 99
11. Second derivative test for more than two variables 100
12. Optimization with constraints and the method of Lagrange multipliers 101
13. Problems 104
Chapter 6. Integrals 107
1. Ways of Integrating 107
2. Double Integrals 108
3. Problems 120
4. Triple integrals 121
5. Why compute a Triple Integral? 124
6. Integration in special coordinate systems 129
7. Problems 132
Chapter 7. Vector Calculus 137
1. Vector Fields 137
2. Examples of vector fields 137
3. Line integrals 140
4. Problems 142
5. Line integrals of vector fields 142
6. Another Fundamental Theorem of Calculus 148
7. Conservative vector fields 150
8. Problems 151
9. Flux integrals 151
10. Greens Theorem 155
11. Conservative vector fields and Clairauts theorem 157
12. Problems 159
13. Surfaces and Surface integrals 160
14. Examples 165
15. The divergence theorem and Stokes theorem 167
16.
#
dierentiating vector fields 168
17. Problems 171
CHAPTER 1
Vector Geometry in ree dimensional space
1. ree dimensional space
Te woild accoiding to oui ist and second semestei calculus couises is at except
foi a biief digiession about suifaces of ievolution, eveiything that we discussed in Math
221 and 222 took place in the (x, y)-plane. All cuives weie cuives in the plane and all
functions had giaphs that weie cuives in the plane. Tis semestei we leave two dimen-
sions behind and entei the thiee dimensional woild. ln oidei to undeistand the objects
we will be dealing with, such as cuives that aie fiee to loop aiound in space, oi functions
whose giaphs aie themselves two dimensional cuived suifaces, we will ist ieview some
thiee dimensional geometiy. ln paiticulai, we will ieview the use of vectois in thiee
dimensional geometiy.
2. Geometric description of vectors
2.1. Points and their coordinates. We aie used to desciibing the location of any
point in the plane by choosing two peipendiculai cooidinate axes (the x and y axes),
and specifying the coiiesponding (x, y)-cooidinates of any given point. ln the same way
we can desciibe wheie points aie in thiee dimensional space by choosing thiee mutually
peipendiculai axes, which we call the x, y, and z axes. To say wheie some given point P
is, we tiavel fiom the oiigin to P, ist along the x axis, then paiallel to the y-axis, and
nally paiallel to the z-axis. Te distances we had to go in the x, y, and z diiections aie
the x, y, and z cooidinates of oui point P.
y
-a
x
is
z
-
a
x
i
s
x
-
a
x
i
s
Figure 1. To determine the location of points in three dimensional space (such as the center of the
blue sphere in this drawing), we should choose three coordinate axes, and specify three numbers:
the x, y, and z coordinates of the point.

1. VECTOR GEOMETRY lN THREE DlMENSlONAL SPACE


2.2. Vectors. While points and theii cooidinates aie used to desciibed locations in
space, vectois aie used to desciibe displacements, i.e. how to go fiom one point to an-
othei. Such a displacement has a size (how fai we have to go), and a diiection (which way
do we go). Vectois also get used in non-geometiic situations to desciibe objects that have
size and diiection, e.g. velocities and foices in physics aie typical examples of vectoi-like
objects.
Informal denition of vectors. We will think of a vectoi as an aiiow connecting two
points. lf the points aie Aand B then we call the vectoi
#
AB. lf we tianslate a vectoi
#
AB
without turning it then we say that the iesulting vectoi
#
CD is the same vectoi as the
oiiginal vectoi
#
AB. A moie piecise way of saying that we should be able to move
#
AB
without tuining, is to insist that the line segments AB and CD should be paiallel, and
have the same length and oiientation.
A
B
C
D
Figure 2. This figure contains four points (A, B, C, D), two line segments (AB and CD), but only
one vector since
#
AB and
#
CD represent the same vector:
#
AB =
#
CD.
We say that the aiiows
#
AB and
#
PQ both represent the same vector. Since both
#
AB and
#
PQ aie the same vectoi we will ofen want to use a notation foi vectois that
does not emphasize any paiticulai choice of initial- and endpoint. Te notation we will
use in this couise is
#
a =
#
AB =
#
PQ,
i.e., a single leuei with an aiiow on top will always stand foi a vectoi in this couise.
to add
two vectors
move one vector
until its initial point
is the end point of
the other
and combine them.
B
P
Q
B
P
Q
C
B
C
B
C
A A A A
#
a
#
a
#
a
#
a
#
b
#
b
#
b
#
b
#
a +
#
b
Figure 3. Adding vectors
3. Arithmetic of vectors
To add two vectois
#
AB and
#
PQ we ist tianslate the vectoi
#
PQ so that its initial
point becomes B, let the iesult of this tianslation be the vectoi
#
BC. Ten, by denition,
4. VECTOR ALGEBRA
the sum of
#
AB and
#
PQ is
#
AC in a foimula,
#
AB +
#
PQ =
#
AB +
#
BC =
#
AC.
An equivalent way of adding two vectois
#
AB and
#
PQis to move the vectois aiound until
they have the same initial point. Two vectois with a common initial point foim two sides
of a paiallelogiam (see Figuie 4) and the sum of the two vectois is the diagonal of that
paiallelogiam.
A
B
CC
D
A
B
CC
D
A
B
CC
D
A
B
D
#
AB +
#
AD =?
Figure 4. Using a parallelogramto add vectors. To find
#
AB+
#
ADwe move the vector
#
ADso
that its initial point is at B, i.e. the endpoint of
#
AB. This gives us a parallelogram ABCD, where
#
AD =
#
BC. Therefore
#
AB +
#
AD =
#
AB +
#
BC =
#
AC
One can also multiply vectois with numbeis. To multiply a vectoi
#
a with a positive
ieal numbei t > 0, we multiply the length of the vectoi by a factoi t, without changing
the diiection of the vectoi.
#
a
2
#
a

#
a
#
a
#
b

#
a

#
b
#
a
#
b
#
b
#
a
Figure 5. Multiplying and subtracting vectors
4. Vector algebra
Te addition and multiplication of vectois and numbeis satisfy a numbei of alge-
biaic piopeities that should look familiai, as they aie veiy similai to the usual algebiaic
piopeities foi adding and multiplying numbeis. Heie they aie
#
a +
#
b =
#
b +
#
a commutative law
(
#
a +
#
b ) +
#
c =
#
a + (
#
b +
#
c ) t (s
#
a) = (ts)
#
a associative laws
t (
#
a +
#
b ) = t
#
a +t
#
b (t +s)
#
a = t
#
a +s
#
a distiibutive laws
8 1. VECTOR GEOMETRY lN THREE DlMENSlONAL SPACE
5. Component representation of vectors
5.1. Components of a vector in two dimensional space. Teie is a way to iepiesent
a vectoi by specifying a list of numbeis instead of by giving a geometiic desciiption of the
vectoi. To do this foi vectois in the plane, we must choose two peipendiculai cooidinate
axes (the x and y axes). We dene
#
e
1
= vectoi with length 1, in the diiection of the x axis
#
e
2
= vectoi with length 1, in the diiection of the y axis
Ten any othei vectoi can be wiiuen as the sum of a multiple of
#
e
1
and anothei multiple
of
#
e
2

(1)
#
a = a
1
#
e
1
+a
2
#
e
2
.
See Figuie . Te numbeis a
1
and a
2
aie called the components of the vector
#
a. lf we
know the components a
1
and a
2
of a vectoi, and if we know the two vectois
#
e
1
and
#
e
2
,
then we can ieconstiuct the vectoi
#
a by using the foimula (1).
#
e
1
#
e
2
#
a
#
a
#
a
a
1
#
e
1
a
2
#
e
2
Figure 6. Describing a vector in terms of its components.
lnstead of using the notation (1), one veiy ofen wiites
(2)
#
a =
_
a
1
a
2
_
, oi
#
a =
_
a
1
a
2
_
, oi
#
a = a
1
, a
2
.
Tis notation says that
#
a is the vectoi whose components aie a
1
and a
2
. Since the two
vectois
#
e
1
and
#
e
2
depend on oui choice of cooidinate axes, we can only use the compo-
nent notation if it is cleai to eveiyone how we chose the cooidinate axes.
Te ist way of wiiting the vectoi, in which the components a
1
and a
2
aie listed in a
column enclosed in eithei paientheses oi squaie biackets, is the standaid way of wiiting
column vectois, and is used in lineai algebia couises (math 320, 340, 341, etc.), as well
as by most computational sofwaie (Matlab
TM
, Octave, etc.). Te othei way of wiiting the
components, i.e. as a
1
, a
2
, also gets used, especially when one has to type the equations
iathei than wiite them by hand.
5.2. Components of a vector in three dimensional space. Te pieceding also ap-
plies to vectois in thiee dimensional space instead of choosing two cooidinate axes we
choose thiee axes, and call them the x, y, and z axes (oi, the x
1
, x
2
, and x
3
axes). Ten
we dene
#
,
#
, and
#
k (oi
#
e
1
,
#
e
2
, and
#
e
3
) to be vectois of length one in the diiection of
. THE DOT PRODUCT 9
the thiee cooidinate axes. A vectoi
#
a in space can then be wiiuen as a combination of
the thiee vectois
#
,
#
, and
#
k, namely,
#
a = a
1
#
+a
2
#
+a
3
#
k, oi
#
a =
_
_
a
1
a
2
a
3
_
_
.
Te
#
e
1
,
#
e
2
,
#
e
3
notation is moie systematic, but the
#
,
#
,
#
k notation, which was intio-
a
2
#
e
2 a
1
#
e
1
a
3
#
e
3
#
e
2
#
e
3
#
e
1
The vector
#
a =
_
_
a
1
a
2
a
3
_
_
is
#
a = a
1
#
e
1
+a
2
#
e
2
+a
3
#
e
3
_
_
a
1
a
2
a
3
_
_
Figure 7. Components of a vector in three dimensional space
Josiah Willard Gibbs
18391903
https://en.wikipedia.org/
wiki/Josiah_Willard_Gibbs
duced into vectoi geometiy and vectoi calculus by J.W.Gibbs, is also veiy common.
5.3. Length of a vector whose components are given. We will wiite

#
a
foi the length of a vectoi
#
a. lf the vectoi is given in components,
#
a = a
1
#
e
1
+a
2
#
e
2
, oi
#
a = a
1
#
e
1
+a
2
#
e
2
+a
3
#
e
3
,
then the length of the vectoi is deteimined by Pythagoias law (see Figuies and )
(3)
#
a =

a
2
1
+a
2
2
, oi
#
a =

a
2
1
+a
2
2
+a
2
3
.
6. e dot product
Teie aie two dieient desciiptions of the dot pioduct of two vectois one geometiic,
and the othei in teims of the components of the vectois.
6.1. Geometric description of the dot product. lf
#
a and
#
b aie two given vectois,
then, by denition,

#
a
#
b
The dot product between
two vectors.
(4)
#
a
#
b =
#
a
#
b cos ,
wheie is the angle between the two vectois
#
a and
#
b .
10 1. VECTOR GEOMETRY lN THREE DlMENSlONAL SPACE
6.2. e dot product in terms of vector components. lf we choose an oithonoimal
set of vectois
#
e
1
,
#
e
2
,
#
e
3
, and wiite
#
a = a
1
#
e
1
+a
2
#
e
2
+a
3
#
e
3
=
_
_
a
1
a
2
a
3
_
_
,
#
b = b
1
#
e
1
+b
2
#
e
2
+b
3
#
e
3
=
_
_
b
1
b
2
b
3
_
_
,
then
()
#
a
#
b = a
1
b
1
+a
2
b
2
+a
3
b
3
.
Te fact that (4) and () always give the same iesult is not obvious (the foimulas look veiy
dieient), and iequiies a pioof. A veiy common pioof ielies on the law of cosines (it was
given in math 222 see also Pioblem 12.1)
6.3. Algebraic properties of the dot product. Te dot pioduct has the following
algebiaic piopeities, which we will use veiy ofen thioughout this couise
#
a
#
b =
#
b
#
a commutative
s(
#
a
#
b ) = (s
#
a)
#
b associative
(
#
a +
#
b )
#
c =
#
a
#
c +
#
b
#
c . distiibutive
We will not piove these piopeities heie. Pioofs can be given if one staits eithei fiom
the algebiaic desciiption of the dot-pioduct (), oi fiom the geometiic desciiption (4) (al-
though the distiibutive piopeity is moie dicult to piove fiom the geometiic desciiption
than fiom the algebiaic desciiption.)
Te sign of the dot pioduct tells us if the angle between two vectois is acute, obtuse,
oi if the vectois aie peipendiculai
#
a
#
b
#
a
#
b = 0 (a)
#
a
#
b > 0 <

2
(b)
#
a
#
b < 0 >

2
. (c)
7. e cross product
As with the dot pioduct, the cioss pioduct of two vectois also has a geometiic de-
sciiption, and a desciiption in teims of components.
7.1. Geometric description of the cross product. Let
#
a and
#
b be two vectois in
thiee dimensional space, then theii cross product is the vectoi
#
a
#
b that satises

#
a
#
b is peipendiculai to
#
a, and also to
#
b
the length of
#
a
#
b is given by

#
a
#
b =
#
a
#
b sin ,
wheie is the angle between the vectois
#
a and
#
b ,
the thiee vectois
#
a,
#
b ,
#
a
#
b satisfy the right hand rule: if on youi iight hand
#
a is the index ngei and
#
b is the middle ngei, then youi thumb points in the
diiection of
#
a
#
b . See Figuie 8.
. THE CROSS PRODUCT 11
#
a
#
b
#
a
#
b
#
a
#
b
#
a
#
b
Figure 8. The cross product:
#
a
#
b is perpendicular to both
#
a and
#
b ; its direction follows from
the right-hand rule.
Te length of the cioss pioduct of two vectois has a geometiic inteipietation. Namely,
the quantity
#
a
#
b sin is exactly the aie of the paiallelogiam spanned by the vectois
#
a and
#
b .
height =
#
a sin
base =
#
b
#
a

Area=heightbase
#
b
7.2. Algebraic description of the cross product. lf
#
a and
#
b aie given by (4), i.e. by
#
a = a
1
#
e
1
+a
2
#
e
2
+a
3
#
e
3
=
_
a1
a2
a3
_
,
#
b = b
1
#
e
1
+b
2
#
e
2
+b
3
#
e
3
=
_
b1
b2
b3
_
,
then
#
a
#
b =
_
_
a
2
b
3
a
3
b
2
a
3
b
1
a
1
b
3
a
1
b
2
a
2
b
1
_
_
.
7.3. Algebraic properties of the cross product. Te cioss pioduct has the distiibu-
tive piopeity, namely,
() (
#
a +
#
b )
#
c =
#
a
#
c +
#
b
#
c ,
holds tiue foi any thiee vectois
#
a,
#
b ,
#
c .
Te cioss pioduct is not commutative
#
a
#
b and
#
b
#
a aie not the same thing.
lnstead, we have
(8)
#
a
#
b =
#
b
#
a.
Because of this piopeity the cioss pioduct is said to be anti-commutative.
12 1. VECTOR GEOMETRY lN THREE DlMENSlONAL SPACE
Te associative piopeity fails completely foi the cioss pioduct foi most vectois
#
a,
#
b ,
#
c one has
(9)

(
#
a
#
b )
#
c =
#
a(
#
b
#
c )

lf you need a vectoi that is peipendiculai to two given vectois, take theii cioss piod-
uct.
Te length of the cioss pioduct
#
a
#
b is the aiea of the paiallelogiam spanned by
those vectois.
8. e triple product
Just as two vectois in the plane foim a paiallelogiam, thiee vectois in space will
foim a shape called a paiallelepiped. By denition, a paiallelepiped is a solid body each
of whose faces is a paiallelogiam.

#
a
#
c
#
b
#
b
#
c
h
e
i
g
h
t

#
a
#
b
#
c
#
b
#
c
h
e
i
g
h
t
Figure 9. A parallelepiped spanned by three vectors
#
a,
#
b ,
#
c . Since the base of the paral-
lelepiped is a parallelogram with edges
#
b and
#
c , we have
Area of base =
#
b
#
c .
The height of the parallelepiped is
#
a cos , and therefore the volume is given by
Volume = height area of base =
#
a
#
b
#
c cos =
#
a
(
#
b
#
c
)
.
This derivation applies to the situation on the le, where the vector
#
a and the cross product
#
b
#
c
point in the same direction. If these vectors form an obtuse angle, as is the case on the right, then
cos < 0, and the height is
#
a cos . In that case one has
Volume = height area of base =
#
a
#
b
#
c cos =
#
a
(
#
b
#
c
)
.
lf we aie given thiee vectois
#
a,
#
b , and
#
c , then the volume of the paiallelepiped they
deteimine is given by the foimula
Volume equals Aiea of base times height
ln teims of the thiee vectois this is
(10) V =

#
a
_
#
b
#
c
_

.
A deiivation is sketched in Figuie 9. Te quantity
#
a (
#
b
#
c ) (without the absolute
values) is called the triple product of the thiee vectois
#
a,
#
b , and
#
c . Apait fiom its use
in computing the volume of a paiallelepiped, the tiiple pioduct appeais in many othei
10. DETERMlNANTS, THE TRlPLE PRODUCT, AND THE CROSS PRODUCT 13
contexts. At ist sight the expiession
#
a (
#
b
#
c ) suggests that the oidei in which the
vectois appeai is impoitant, but this tuins out not to be tiue. One has
#
a
_
#
b
#
c
_
=
#
b
_
#
c
#
a
_
=
#
c
_
#
a
#
b
_
foi any
#
a,
#
b ,
#
c .
9. Determinants
Foi any foui numbeis a, b, c, d, one denes the 2 2 deteiminant to be
(11)

a b
c d

= ad bc .
One can also dene 3 3 deteiminants. Namely, foi any nine numbeis a
1
, . . . , c
3
one
denes
(12)

a
1
b
1
c
1
a
2
b
2
c
2
a
3
b
3
c
3

= a
1
b
2
c
3
a
1
b
3
c
2
a
2
b
1
c
3
+a
2
b
3
c
1
+a
3
b
1
c
2
a
3
b
2
c
1
.
Tis can be wiiuen as

a
1
b
1
c
1
a
2
b
2
c
2
a
3
b
3
c
3

= a
1
_
b
2
c
3
b
3
c
2
_
a
2
_
b
1
c
3
b
3
c
1
_
+a
3
_
b
1
c
2
b
2
c
1
_
(13)
= a
1

b
2
c
2
b
3
c
3

a
2

b
1
c
1
b
3
c
3

+a
3

b
1
b
1
b
2
b
2

wheie each coecient in the ist iowis multiplied with the 22 deteimined that iemains
afei one deletes the iow and column containing the coecient.
lnstead of expanding along the ist iow one can also expand along the ist column
(14)

a
1
b
1
c
1
a
2
b
2
c
2
a
3
b
3
c
3

= a
1

b
2
c
2
b
3
c
3

b
1

a
2
c
2
a
3
c
3

+c
1

a
2
b
2
a
3
b
3

Many othei mnemonic devices exist to iemembei how to compute a 3 3 deteiminant.


A populai tiick is Saiius iule (see Figuie 10.)
One can also dene laigei deteiminants, i.e. 4 4, 5 5, etc, and geneially n n
deteiminants. Te theoiy, which is beyond the scope of this couise, is tieated in lineai
algebia couises such as Math 320, 340, oi 341.
10. Determinants, the triple product, and the cross product
lf the numbeis a
1
, . . . , c
3
in a deteiminant happen to be the components of thiee
vectois
#
a,
#
b ,
#
c , i.e. if
#
a =
_
_
a
1
a
2
a
3
_
_
,
#
b =
_
_
b
1
b
2
b
3
_
_
,
#
c =
_
_
c
1
c
2
c
3
_
_
,
then the coiiesponding deteiminant is exactly the tiiple pioduct
(1)

a
1
b
1
c
1
a
2
b
2
c
2
a
3
b
3
c
3

=
#
a
_
#
b
#
c
_
.
14 1. VECTOR GEOMETRY lN THREE DlMENSlONAL SPACE
a
1
a
2
a
3
a
1
a
2
+ + + - - -
b
1
b
2
b
3
b
1
b
2
c
1
c
2
c
3
c
1
c
2
a
1
b
2
c
3
a
2
b
3
c
1
a
3
b
1
c
2
a
3
b
2
c
1
a
1
b
3
c
2
a
2
b
1
c
3
Figure 10. Computing 3 3 determinants. There are several shortcuts to remember how
to compute a 3 3 determinant. Pictured here is Sarrus rule, which tells us to copy the first
two columns of the determinant to the right of the determinant, and read o the six terms in the
determinant by following the diagonals.
Related to this is the following piactical tiick foi computing the cioss pioduct of two
column vectois. Given two column vectois
#
b and
#
c one can wiite theii cioss pioduct as
_
_
b
1
b
2
b
3
_
_

_
_
c
1
c
2
c
3
_
_
=

#
e
1
b
1
c
1
#
e
2
b
2
c
2
#
e
3
b
3
c
3

b
2
c
2
b
3
c
3

#
e
1

b
1
c
1
b
3
c
3

#
e
2
+

b
1
c
1
b
2
c
2

#
e
3
.
Te 3 3 deteiminant in this equation is unusual in that some of its entiies aie vectois
instead of numbeis. Te intention of this notation is that one expand the deteiminant
along the ist column, as in (13) and then inteipiet the iesult as a vectoi.
11. Dening equations for lines and planes
11.1. Lines. Let be a line in the plane, and suppose we know one point A on the
line, and that we also have a vectoi
#
n that is peipendiculai to the line (and we exclude
#
n =
#
0.) Such a vectoi is called a normal vector to the line. Given any othei point X in
the plane we can foim the vectoi
#
AX and considei its dot-pioduct with the noimal. We
have
#
n
#
AX =
#
n
#
AX cos ,
wheie is the angle between the noimal vectoi
#
n and
#
AX.
Te combination
#
AX cos is, up to its sign, the distance fiom the line to the
point X lf X lies on the side of at which the noimal vectoi points then
#
n
#
AX > 0, if
X lies on the othei side then
#
n
#
AX < 0. We theiefoie have the following foimula foi
the distance between a point X and the line :
(1) d =
#
n
#
AX

#
n
When we use this equation to compute the distance fiom X to , it is good to iecall that
if
#
x = (
x1
x2
) and
#
a = (
a1
a2
) aie the position vectois of the points X and A, then
#
AX =
#
x
#
a =
_
x
1
a
1
x
2
a
2
_
.
11. DEFlNlNG EQUATlONS FOR LlNES AND PLANES 1
X
A

#
n
X
A

#
n

#
n
#
AX < 0
d =
#
AX cos( )
=
#
AX cos
#
n
#
AX > 0 d =
#
AX cos
Moieovei, the length of the noimal vectoi is
#
n =

n
2
1
+n
2
2
, so we can iewiite (1) as
d =
n
1
(x
1
a
1
) +n
2
(x
2
a
2
)

n
2
1
+n
2
2
.
Tis last foimula is moie impiessive than (1), but it is beuei to iemembei (1).
Te equation foi the distance fiom any point X to a given line is also impoitant
because it gives us the dening equation foi the line . Te dening equation is an
equation that tells us foi any given point X in the plane if that point is on the line oi not.
Since X is on exactly when the distance fiom to X vanishes, it follows fiom (1) that
X is on if and only if
(1)
#
n
#
AX = 0.
We can again iewiite this equation in a few dieient ways. lf we want to wiite it in teims
of the position vectois of A and X, then we get
#
n
_
#
x
#
a
_
= 0, i.e.
#
n
#
x =
#
n
#
a.
Wiiuen without vectois, but in teims of the cooidinates of the points A, X, and the
components of the noimal vectoi
#
n, we can wiite this last veision of oui equation as
n
1
x
1
+n
2
x
2
= n
1
a
1
+n
2
a
2
.
11.2. Planes. We can iepeat the deiivation of the distance fiom a point to a line in
the plane and deiive a foimula foi the distance fiom a point in thiee dimensional space
to a given plane. Te diawings aie haidei to make (at ist only, piactice makes peifect'),
but the iesulting foimulas aie the same.
Te distance fiom a point X to a plane P is given by equation (1), wheie
#
n is a
noimal vectoi to the plane (a vectoi that is peipendiculai to the plane), and A is some
point on the plane that we happen to know.
1 1. VECTOR GEOMETRY lN THREE DlMENSlONAL SPACE
A
X
#
n

d
d =
#
AX cos
#
n
#
AX =
#
n
#
AX cos
12. Problems
1. (a) Simplify the following
#
a =

1
2
3

+ 3

0
1
3

#
b = 12
(
1
1/3
)
3
(
4
1
)
#
c = (1 +t)
(
1
1 t
)
t
(
1
t
)
#
d = t

1
0
0

+t
2

0
1
2

0
0
1

(b) Write the vectors from part (a) using


Gibbs notation, i.e. write them in terms of
#
,
#
,
#
k. (See 5).
2. If
#
a,
#
b ,
#
c are as in the previous prob-
lem, then which of the following expressions
mean anything? Compute those expressions
that are well defined.
(a)
#
a +
#
b (b)
#
b +
#
c (c)
#
a
(d)
#
b
2
(e)
#
b /
#
c (f)
#
a +
#
b
(g)
#
b
2
(h)
#
b /
#
c
3. Let
#
a =
(
1
2
2
)
and
#
b =
(
2
1
1
)
.
Compute:
(a) ||
#
a|| (b) 2
#
a (c) ||2
#
a||
2
(d)
#
a +
#
b (e) 3
#
a
#
b

4. Given: points A(2, 1) and B(1, 4).


Compute the vector
#
AB. Is
#
AB a position
vector?
5. Given: points A(2, 1), B(3, 2), C(4, 4)
and D(5, 2).
estion: Is ABCD a parallelogram?
6. Given: points A(0, 2, 1), B(0, 3, 2),
C(4, 1, 4) and D.
(a) If ABCD is a parallelogram, then what
are the coordinates of the point D?
(b) If ABDC is a parallelogram, then what
are the coordinates of the point D?
7. You are given three points in the plane:
A has coordinates (2, 3), B has coordinates
(1, 2) and C has coordinates (4, 1).
(a) Compute the vectors
#
AB,
#
BA,
#
AC,
#
CA,
#
BC and
#
CB.
(b) Find the points P, Q, Rand S whose po-
sition vectors are
#
AB,
#
BA,
#
AC, and
#
BC,
respectively. Make a precise drawing.
8. Explain how you can use the dot prod-
uct to find the angle between the vectors
#
a = 2
#
3
#
, and
#
b =
#
+
#
k.
12. PROBLEMS 1
A
B
C
D
E
F
G
H
Figure 11. Figure for problem 12.10
9. For which value(s) of the number s are
the vectors
#
a =
(
s
1 s
)
and
#
b =
(
2
3
)
perpendicular? For which values of s do they
make an acute angle?
10. Figure 11 shows a cube whose sides have
length 1.
Choose A to be the origin, and let the x,
y, and z axes be along the sides AB, AD,
and AE, respectively.
(a) Draw the vectors
#
e1,
#
e2, and
#
e3 in the
figure.
(b) Find a normal vector to the plane
through the points B, D, and E.
(c) Draw the plane through ACH (or at
least the portion of that plane that lies in-
side the cube). Find a normal to the plane
ACH.
(d) Find the angle between the two planes
BDE and ACH. (The angle between two
planes is the same as the angle between their
normal vectors, i.e. to find the angle between
two planes find a normal vector for each of
the planes and compute the angle between
these two vectors.)
(e) Find the angle between the two planes
BDE and HFC.
11. (a) Draw two vectors
#
a and
#
b for which
#
a has length 3,
#
b has length 5, and for
which
#
a
#
b = 12. How many solutions
are there?
(b) Can there be two vectors
#
a and
#
b whose
lengths are
#
a = 3 and
#
b = 5, and
whose inner product is
#
a
#
b = 25?
12. Compute
#
a = (
#

#
)
#
and
#
b =
#
(
#

#
).
What does your answer say about the asso-
ciative property for the cross product? (See
7.3.)
What about
#
c = (
#

#
)
#
k and
#
d =
#
(
#

#
k)?
13. Which of the following vector equations
are true for any pair of vectors
#
a and
#
b ? Ei-
ther give a proof (using the algebraic prop-
erties or the algebraic or geometric descrip-
tions).
(a) (
#
a +
#
b ) (
#
a
#
b ) =
#
a
2

#
b
2
?
(b) If
#
a
#
b then

#
a +
#
b
2
=
#
a
2
+
#
b
2
?
(c) If
#
a
#
b then

#
a
#
b
2
=
#
a
2

#
b
2
?
18 1. VECTOR GEOMETRY lN THREE DlMENSlONAL SPACE
14. True or False:
(a) If
#
a
#
b and also
#
b
#
c then
#
a
#
c
?
(b) If
#
a
#
b and also
#
a
#
c then
#
a
(
#
b +
#
c ) ?
(c) If
#
a
#
b and also
#
b
#
c then
#
b
(
#
a
#
c ) ?
(d) If
#
a
#
b +
#
c and also
#
a
#
b
#
c then
#
a
#
b ?
15. Simplify the following expressions
(a) (
#
a +
#
b )(
#
a +
#
b )
(b) (
#
a +
#
b +
#
c )(
#
a +
#
b +
#
c )
(c) (
#
a
#
b )(
#
a +
#
b )
(d) (
#
a +
#
b
#
c )(
#
a
#
b +
#
c )
(e) (
#
a +
#
b
#
c ) (
#
a
#
b +
#
c )
16. This problem is about cross division,
i.e. can you solve
#
a
#
b =
#
c for
#
b if you
know
#
a and
#
c ?
(a) Let
#
a =
#
e1
#
e3,
#
c =
#
e1 + 3
#
e2 + 2
#
e3.
Find a vector
#
b for which
#
a
#
b =
#
c , if
there is such a thing. (Hint: if
#
c =
#
a
#
b ,
then what do you know about
#
a
#
c ?)
(b) Let
#
a = 2
#
e1
#
e3, and
#
c =
#
e1 +3
#
e2 +
2
#
e3. Find a vector
#
b for which
#
a
#
b =
#
c ,
if such a thing exists.
17. The law of cosines says that in a triangle
ABC for which you know the sides AB
and AC, as well as the angle A, the length
of the opposing side BC is given by
(BC)
2
= (AB)
2
+ (AC)
2
2(AB)(AC) cos A.
Show how you can use the dot product to
(re)prove this law.
Hint: consider the vector equation
#
BC =
#
AC
#
AB. You will need both the
geometric description (4) of the dot product,
and the algebraic properties from 6.3.
CHAPTER 2
Parametric curves and vector functions
1. Vector functions
So fai in calculus we have only consideied functions y = f(x) wheie both the inde-
pendent vaiiable x and the dependent vaiiable y aie ieal numbeis.
A vector function is a function of one vaiiable whose values aie vectois instead of
numbeis. One way to specify a vectoi function is to say what its components aie
#
x(t) =
_
_
x(t)
y(t)
z(t)
_
_
= x(t)
#
e
1
+y(t)
#
e
2
+z(t)
#
e
3
.
2. Using vector functions to describe motion
One way to visualize a vectoi function
#
x(t) is to think of the vectoi
#
x(t) foi any
given value of t as the position vectoi of some point in space (oi the plane, if
#
x(t) is a two-
dimensional vectoi). ln othei woids, we iepiesent the vectoi
#
x(t) as an aiiow staiting
at the oiigin, and ending at some point X(t) whose cooidinates aie (x(t), y(t), z(t))
#
x(t) =
#
OX(t).
As t vaiies, the point X(t) moves aiound and tiaces out a cuive. Such a cuive is called a
parametrized curve, oi a parametric curve. Te quantity t is called the parameter.
We will now take a look at some examples of paiametiic cuives.
#
x(t)
O
X(t)
Figure 1. Aparametric curve: as the parameter t changes, the vector
#
x(t) will also move. Keep-
ing the initial point of the vector
#
x(t) at the origin O, the endpoint X(t) traces out a space curve.
19
20 2. PARAMETRlC CURVES AND VECTOR FUNCTlONS
3. Lines
Considei the paiametiic cuive given by
(18)
#
x(t) =
#
a +t
#
v
wheie
#
a and
#
v aie given constant vectois. As befoie we let X(t) be the point with
#
x(t) =
#
OX(t), i.e.
#
x(t) is the position vectoi of the point X(t), and as t changes, X(t)
tiaces out the paiametiic cuive.
To see what the paiametiic cuive looks like, we let A be the point with
#
OA =
#
a,
then, since
#
OX(t) =
#
OA+
#
AX(t),
it follows fiom (18) that
#
AX(t) = t
#
v. Now considei going fiom the oiigin O to the
point X(t) in two steps ist move fiom O to the point A, then go fiom A to X(t). Te
displacement in the second step is
#
AX(t) = t
#
v. Changing t will then make the point
X(t) slide along the line thiough the point A in the diiection of
#
v.
#
a
#
v
#
x(t) =
#
a +t
#
v
X(t)
Origin
A
t
#
v
Figure 2. Vector form of linear motion given by
#
x(t) =
#
a +t
#
v.
We say that
#
x(t) given by (18) desciibes motion with constant velocity, whose ve-
locity vectoi is
#
v.
4. Circular motion
Foi given constants R > 0 and we considei the vectoi function
(19)
#
x(t) = Rcos t
#
e
1
+Rsin t
#
e
2
=
_
Rcos t
Rsin t
_
.
Te coiiesponding point is X(t) =
_
Rcos t, Rsin t
_
. lt lies on the ciicle of iadius R
with centei at the oiigin, and the angle subtended by OX(t) and the positive x-axis is
exactly t.
lf > 0 then as t incieases, the angle t incieases and the point X(t) goes aiound
the ciicle in countei-clockwise diiection. lf < 0 then X(t) goes aiound in the clockwise
diiection.
Te numbei is the iate of inciease of the angle t, and is called the angular ve-
locity of the motion.
. THE HELlX 21
#
x(t)
t
X(t)
O
Figure 3. Circular motion with angular velocity .
5. e cycloid
Te cycloid is the cuive we get if we put a (bicycle) wheel on the giound, maik
the point on the tiie that touches the giound, and follow this point as we ioll the wheel
foiwaid. lf we call the point X, then it depends on the angle that the wheel has tuined
since X was on the giound. Figuie 4 piovides a deiivation of the vectoi function
#
x() =
#
OX() that desciibes the cycloid. Te iesult is
(20)
#
x() =
_
R Rsin
R Rcos
_
.
X
C
B
A O

O A A
C
C
X
X
Figure 4. The cycloid. A wheel of radius R rolls over the x-axis. Initially the wheel touches the
x-axis at the origin O. The cycloid is the curve traced out by a point X on the wheel.
Derivation of the cycloid motion. The arc AX and the line segment OA have the same
length. Since AX has length R, the x coordinates of the points A, B, and C are R. The right
triangle CXB has hypotenuse R, so the lengths of XB and CB are Rsin , and Rcos , respec-
tively. Therefore the coordinates of the point X are x = R Rsin , and y = R Rcos .
6. e helix
When we walk up a spiial staiicase we aie tiacing out a helix we aie going aiound
in ciicles, and moving upwaid at the same time. Te paiametiic cuive that does this (and
22 2. PARAMETRlC CURVES AND VECTOR FUNCTlONS
that has the z-axis as its cential axis) is given by
(21)
#
x() =
_
_
Rcos
Rsin
a
_
_
oi
#
x() = Rcos
#
e
1
+Rsin
#
e
2
+a
#
e
3
.
Heie R > 0 is the iadius of the helix, i.e. the iadius of the ciicle on the giound above
which the helix lies, the numbei a iepiesents the iate at which the helix goes up.
x
y
z

a
X
O
Y
A
Figure 5. The Helix. The point X traces out a helix: it sits at a height a above the point Y ,
while Y runs around on a circle of radius R; here = AOY
7. e derivative of a vector function
Foi a function y = f(x) of one vaiiable we had two ways of desciibing the deiivative
on one hand we had a geometiic desciiption of f

(x) as the slope of the tangent to the


giaph, and on the othei we could desciibe f

(x) in teims of a dieience quotient, i.e.


f

(x) = lim
x0
f(x + x) f(x)
x
.
Foi vectoi functions we can imitate both desciiptions. We begin with the foimal denition
in teims of limits and then pioceed to the geometiic desciiption, in which we inteipiet
the deiivative as the instantaneous velocity vectoi.
Denition. lf
#
x(t) is a vectoi function, then we set
(22)
#
x

(t)
def
= lim
t0
#
x(t + t)
#
x(t)
t
.
Foi (22) to make sense we would have to dene what the limit of a vectoi function is.
Tis can be done, but we will not go into the piecise denitions in this couise. Moie
8. THE DERlVATlVE AS VELOClTY VECTOR 23
impoitant foi oui use is that if the components of a vectoi function
#
x(t) aie given, then
the deiivative can be computed by just dieientiating those components
(23)
#
x

(t) =
_
_
x

(t)
y

(t)
z

(t)
_
_
, oi
#
x

(t) = x

(t)
#
e
1
+y

(t)
#
e
2
+z

(t)
#
e
3
.
As with oidinaiy functions of one vaiiable we will use Leibniz notation foi the deiivative
whenevei it seems convenient. Tus the following aie equivalent ways of expiessing the
same deiivative
#
a

(t) =
d
#
a(t)
dt
=
d
dt
#
a(t).
Example. Foi instance,
#
x() =
_
_
cos
0

_
_
= cos
#
e
1
+
#
e
3
denes a vectoi function. Heie we have called the independent vaiiable instead of t.
Te deiivative of this vectoi function is
d
#
x
d
=
d
d
_
_
cos
0

_
_
=
_
_
sin
0
1
_
_
= sin
#
e
1
+
#
e
3
.
8. e derivative as velocity vector
Suppose the motion of some point X(t) in space is desciibed by its position vectoi
function
#
x(t). Let us tiy to dene the instantaneous velocity of the point. Tis velocity
should have magnitude (how fast the point is moving) and also diiection (which way

x
v = dx/dt
x
(
t
)
x
(
t
+

t
)
X(t)
O
Figure 6. The vector function
#
x(t) traces out a curve in space. The vector
#
x(t) is the position
vector of a point X(t) on this curve. As we increase time from t to t + t, the point X(t) moves.
The displacement of the point X(t) is given by
#
x =
#
x(t + t)
#
x(t). The average velocity
vector during this displacement is displacement/time, i.e.
#
x/t.
If we let t 0, then the average velocity becomes the instantaneous velocity at time t:
#
v = lim
t0

#
x/t =
#
x

(t). This vector is tangent to the curve traced out by the vector
function
#
x(t). We call it a tangent vector.
24 2. PARAMETRlC CURVES AND VECTOR FUNCTlONS
is the point going`). Te velocity should theiefoie be a vectoi. To see which vectoi, we
go back to the notion that velocity is always displacement divided by time.
We considei two instances in time, say, time t and time t+t. Ten the position vec-
tois of the point X at these two dieient times aie
#
x(t) and
#
x(t+t). Te displacement
of the point X between these two times is then

#
x =
#
x(t + t)
#
x(t)
(see Figuie .) We say that the aveiage velocity ovei the time inteival fiom t to t +t is
the displacement divided by t, i.e.
#
v
aveiage
=
#
x(t + t)
#
x(t)
t
.
Note that the aveiage velocity is a vectoi. lf we wiite it out in components, we get a much
laigei foimula
#
v
aveiage
=
_
_
_
_
_
_
_
_
x(t + t) x(t)
t
y(t + t) y(t)
t
z(t + t) z(t)
t
_
_
_
_
_
_
_
_
.
One big advantage of using vectoi notation is that many foimulas simplify consideiably
when wiiuen in teims of vectois.
To get the instantaneous velocity, we do the same thing as in one vaiiable calculus
we take the limit as t 0 of the aveiage velocity ovei the time inteival fiomt to t+t.
Tus we get
(24)
#
v(t) = lim
t0
#
x(t + t)
#
x(t)
t
def
=
d
#
x
dt
.
ln teims of components this deiivative is
#
x

(t) =
d
#
x
dt
=
_
_
x

(t)
y

(t)
z

(t)
_
_
.
Tus the velocity vectoi of any given vectoi function
#
x(t) is the same as the deiivative
of this vectoi function.
9. Acceleration
Having found the velocity vectoi of a point X(t) whose position vectoi is a given
vectoi function
#
OX(t) =
#
x(t), we can also dene the accelerationvector of the moving
point. By denition, the acceleiation vectoi is the deiivative of the velocity vectoi, i.e.
(2)
#
a(t) =
d
#
v
dt
=
d
2 #
x
dt
2
=
_
_
x

(t)
y

(t)
z

(t)
_
_
.
Tis denition is entiiely analogous to the denition of acceleiation (a =
dv
dt
) fiom ist
semestei calculus. Te only dieience is that, heie, the position, velocity, and acceleiation
all have diiections in addition to magnitudes they aie vectois.
10. THE DlFFERENTlATlON RULES 2
Newtons famous law ielating foices and acceleiation continues to hold. lf a point
X(t) moves accoiding to some vectoi function
#
x(t), then some foice must be acting
on this point. Tis foice is a vectoi (it has magnitude and diiection), and, accoiding to
Newton, it is given by
(2)
#
F = m
#
a = m
d
#
v
dt
= m
d
2 #
x
dt
2
,
wheie m is the mass of the object at the point X(t) whose motion we aie consideiing. lt
is always assumed to be a positive numbei.
Note that accoiding to this law, the absence of foices, i.e.
#
F =
#
0, is the same as
d
#
v
dt
=
#
0, i.e. no foice acts on the point if and only if its velocity vectoi is constant. Heie
constant means constant magnitude and constant diiection.
10. e dierentiation rules
Just as with oidinaiy deiivatives, the deiivatives of vectoi functions satisfy ceitain
iules, such as the pioduct iule. Te puipose of these iules is not the same as in one vaiiable
calculus. Teie we used sum, pioduct, quotient and chain iules to compute deiivatives
of given functions without having to fall back on the denition of a deiivative all the
time. Foi vectoi functions we do not need such iules, because we can dieientiate them
by simply dieientiating each of theii components (see the above example). lnstead, the
dieientiation iules foi vectoi functions aie mostly used to gain insight and establish
geneial facts about vectoi functions, a numbei of which we will see shoitly.
10.1. e sum rule. Te analog of the sum iule (deiivative of the sum is the sum of
the deiivatives) looks exactly like the oidinaiy sum iule. lt says that foi any two vectoi
functions
#
a(t) and
#
b (t) one has
d
dt
_
#
a(t)
#
b (t)
_
=
d
#
a(t)
dt

d
#
b (t)
dt
.
10.2. e many product rules. Teie is no quotient iule foi vectoi functions, simply
because we have no way of dividing vectois. On the othei hand we have two ways of
multiplying vectois, and we can also multiply vectois and numbeis, so theie aie three
dieient pioduct iules. Foitunately they all look like the pioduct iule fiom ist semestei
calculus.
lf
#
a(t) and
#
b (t) aie vectoi functions, and if f(t) is a function, then
d
#
a(t)
#
b (t)
dt
=
d
#
a(t)
dt

#
b (t) +
#
a(t)
d
#
b (t)
dt
d
#
a(t)
#
b (t)
dt
=
d
#
a(t)
dt

#
b (t) +
#
a(t)
d
#
b (t)
dt
d f(t)
#
a(t)
dt
=
df(t)
dt
#
a(t) +f(t)
d
#
a(t)
dt
ln spite of the fact that these iules look iight, they could still be wiong, so to be suie
we would have to piove them. Te pioofs aie veiy stiaightfoiwaid. Heie is a shoit pioof
2 2. PARAMETRlC CURVES AND VECTOR FUNCTlONS
foi the pioduct iule involving the dot pioduct. To shoiten the foimulas we omit the (t)
fiom all functions
d
#
a
#
b
dt
=
d
dt
_
a
1
b
1
+a
2
b
2
_
=
da
1
b
1
dt
+
da
2
b
2
dt
=
da
1
dt
b
1
+a
1
db
1
dt
+
da
2
dt
b
2
+a
2
db
2
dt
ordinary product rule
=
da
1
dt
b
1
+
da
2
dt
b
2
+a
1
db
1
dt
+a
2
db
2
dt
switch terms around
=
d
#
a
dt

#
b +
#
a
d
#
b
dt
. recognize the dot-products
11. Vector functions of constant length
As an immediate application of the pioduct iule foi the dot-pioduct we piove the
following fact about vectoi functions whose length does not change, i.e. vectoi functions
#
a(t) that change theii diiection, but not theii length.
#
a(t)

#
a
#
a(t + t)
If a vector function
#
a(t) has
constant length, then, when the
parameter t undergoes a small
change t, the corresponding
small change
#
a in the vector
function will be almost perpendic-
ular to
#
a(t) itself.
eorem. Let
#
a(t) be a vector function. en a necessary and sucient condition for
the length
#
a(t) to be constant is that
#
a(t) and
#
a

(t) be perpendicular for all t.


Pvooi. Dieientiating both sides of the equation

#
a(t)
2
=
#
a(t)
#
a(t)
we get
(2)
d
dt

#
a(t)
2
=
#
a

(t)
#
a(t) +
#
a(t)
#
a

(t) = 2
#
a(t)
#
a

(t).
lf
#
a(t) has constant length, then
#
a(t)
2
is also constant, and thus
d
dt

#
a(t)
2
= 0.
Teiefoie, foi a vectoi function
#
a(t) whose length is constant,
#
a(t)
#
a

(t) = 0, i.e.
#
a(t)
#
a

(t).
Conveisely, if
#
a(t) is a vectoi function foi which
#
a(t)
#
a

(t) holds foi all t, then


#
a(t)
#
a

(t) = 0, and (2) implies that


d
dt

#
a(t)
2
= 0, i.e. that
#
a(t)
2
and hence
#
a(t)
aie constant.

12. TWO EXAMPLES 2


12. Two examples
12.1. Motion on a straight line. We ietuin to the motion given by (18), i.e.
(28)
#
x(t) =
#
a +t
#
v.
Te velocity and acceleiation aie easy to compute
d
#
x(t)
dt
=
#
v,
d
2 #
x(t)
dt
=
d
#
v
dt
=
#
0,
since
#
v is a constant vectoi in this case.
We see that if a point X(t) moves accoiding to the paiametiization (18), then its
velocity is constant, and its acceleiation is zeio. Accoiding to Newtons law, no foice is
exeited on an object undeigoing this motion.
12.2. Circular motion. Foi the point X(t) moving on a ciicle of iadius R with an-
gulai velocity we have (19), i.e.
#
x(t) = Rcos t
#
e
1
+Rsin t
#
e
2
so that the velocity and acceleiation aie easy to compute
#
v(t) =
#
x

(t) = Rsin t
#
e
1
+ Rcos t
#
e
2
,
#
a(t) =
#
v

(t) =
2
Rcos t
#
e
1

2
Rsin t
#
e
2
.
Note that the velocity vectoi
#
v(t) is peipendiculai to the position vectoi
#
x(t), as
piedicted in 11. Oui expiession foi the velocity vectoi
#
v(t) contains the familiai ie-
lation between angulai velocity and velocity the velocity v =
#
v(t) with which the
point X(t) is moving is
v(t) = Rsin t
#
e
1
+ Rcos t
#
e
2
(29)
=

2
R
2
sin
2
t +
2
R
2
cos
2
t
= R.
Hence the angulai velocity of an object undeigoing ciiculai motion is
(30) =
v
R
.
#
F
#
v(t)
t
R
X
Figure 7. If an object moves along a circle with constant angular velocity, then the force
#
F required
to make the object follow that motion is
#
F =
2 #
x. In particular it is parallel to the position
vector
#
x but in the opposite direction.
28 2. PARAMETRlC CURVES AND VECTOR FUNCTlONS
We also note that the acceleiation is a multiple of the position vectoi
#
a(t) =
2 #
x(t).
Accoiding to Newton the foice acting on the object at X(t) is
#
F = m
#
a = m
2 #
x, and
its magnitude is
(31) F =
#
F = m
2 #
x(t) = m
2
R,
because
#
x(t) = R at all times.
Using (30) we can ieplace the angulai velocity by the actual velocity, which leads
to the classical foimula foi the centiifugal foice
(32) F =
mv
2
R
.
13. Arc length
Foi any given vectoi function theie is a simple foimula foi the length of the cuive
it tiaces out. Te foimula is essentially the same as the foimula foi the length of a paia-
metiic cuive (oi, to a lessei extent, of the giaph of a function) that was desciibed in Math
221. Heie we iepeat the intuitive deiivation of the foimula, wiiuen in teims of vectois
this time.
Let
#
x(t) (a t b) be a vectoi function. To deteimine the length of the aic tiaced
out by X(t) as t vaiies fiom t = a to b, we divide the inteival a t b into many
veiy shoit subinteivals. Te coiiesponding points X(t) on the cuive split the cuive into
many shoit segments, each of which will be close to a line segment. We appioximate
the length of the cuive by adding the lengths of all these shoit segments. Finally we take
the limit in which the numbei of paitition points becomes innite and oui sum of lengths
of shoit segments becomes an integial. To see which integial we get, we need to nd an
expiession foi the length of a shoit segment between two adjacent paitition points on
the cuive.
Suppose we have two points on the cuive, with paiametei values t and t + t, ie-
spectively. Te points aie X(t) and X(t + t), and the distance between them is the
length of the vectoi
#
x fiom one point to the next. Tis vectoi is
x start
(t=a)
end
(t=b)
partition
piece
X(t)
X(t+t)
x =
#
x(t + t)
#
x(t) =
#
x(t + t)
#
x(t)
t
t
#
x

(t)t,
so that its length is
#
x

(t) t. Adding the lengths of the shoit segments togethei,


we nd that the length is appioximately

#
x

(t) t (wheie the summation is ovei all


shoit pieces of the cuive). Taking the limit we aiiive at this foimula foi the length of the
cuive tiaced out by
#
x(t), a t b
(33) Length =

b
t=a

#
x

(t) dt.
Tis integial looks simple, but that appeaiance tuins out to be deceptive as we nd
out when we wiite it in teims of the components of the vectoi function
#
x(t). Suppose
#
x(t) = x(t)
#
e
1
+y(t)
#
e
2
+z(t)
#
e
3
. Ten
#
x

(t) = x

(t)
#
e
1
+y

(t)
#
e
2
+z

(t)
#
e
3
,
so that

#
x

(t) =

(t)
2
+y

(t)
2
+z

(t)
2
.
14. ARC LENGTH DERlVATlVE 29
Teiefoie the length foimula (33) of the cuive is equivalent to
(34) Length =

b
t=a

(t)
2
+y

(t)
2
+z

(t)
2
dt.
Te squaie ioot makes this foimula a ieliable souice of veiy dicult integials. ln fact the
list of cuives whose length one can actually compute by doing the integial is iathei shoit
(see Pioblem ).
14. Arc length derivative
Let
#
x(t) be some vectoi function that desciibes the motion thiough space of some
point X(t), and let f(t) be some othei function. ln what follows it will help to think of
the paiametei t as time. Typical examples of functions f that we might want to considei
aie f(t) =
#
x(t) (the distance to the oiigin of the point X(t)) oi f(t) =
#
x

(t) (the
speed at which the point is moving.)
To desciibe the iate with which f(t) is changing we could compute its deiivative,
df
dt
which tells us what the iatio between the change f of f, and the change t in the
paiametei t is (at least appioximately, if t is small). lf we inteipiet t as time then
this deiivative tells us how fast f(t) changes pei second. But sometimes it is moie useful
to know how much f changes afei we have tiavelled a small distance along the cuive,
iathei than afei a shoit amount of time has passed. ln othei woids, foi two neaiby points
X(t) and X(t + t) on the cuive we would like to know the iatio
(3)
change in f
distance tiavelled
=
f(t + t) f(t)
distance fiom X(t) to X(t + t)
We can woik this out by obseiving that the distance fiomX(t) to X(t+t) is the length
of the vectoi fiom X(t) to X(t + t), i.e.
distance fiom X(t) to X(t + t) =
#
x(t + t)
#
x(t) .
Assuming t is small, we have

#
x(t + t)
#
x(t) =
_
_
_
_
#
x(t + t)
#
x(t)
t
_
_
_
_
t
_
_
#
x

(t)
_
_
t.
We substitute this in (3), and get
change in f
distance tiavelled

f(t + t) f(t)

#
x

(t)t
.
Now let t 0 the quantity on the lef becomes what is called the arc length deriv-
ative of the function f along the cuive vx(t), and which is commonly denoted by
df
ds
ln
the quantity on the iight we iecognize the deiivative of f with iespect to t (time), which
leads to
(3)
df
ds
=
1

#
x

(t)
df
dt
.
Heie
df
dt
= f

(t) is the usual deiivative of f with iespect to t.


lf we want to emphasize the distinction between these two deiivatives, then we can
call
df
dt
the time deiivative of f.
30 2. PARAMETRlC CURVES AND VECTOR FUNCTlONS
15. Unit Tangent and Curvature
15.1. Unit tangent. We have seen that we can nd a tangent vectoi to the cuive
tiaced out by some vectoi function
#
x(t), simply by dieientiating the vectoi function
#
x

(t) always piovides a tangent vectoi (if


#
x

(t) =
#
0). ln fact any multiple
#
x

(t)
A vector with length 1 is
called a unit vector
of this vectoi will also be a tangent vectoi (piovided = 0.) We can single out one
special tangent vectoi, by choosing > 0 so that
#
x

(t) has length 1. Since foi > 0


we have
#
x

(t) =
#
x

(t) the value of that will make


#
x

(t) a unit vectoi is


= 1/
#
x

(t).
Foi this ieason the vectoi
(3)
#
T(t) =
d
#
x
ds
=
#
x

(t)

#
x

(t)
is called the unit tangent vector to the cuive coiiesponding to the vectoi function
#
x(t).
15.2. Example. Foi oui constant velocity paiametiization (18) of a stiaight line fiom
3 we have
#
x(t) =
#
a +t
#
v,
so that
#
x

(t) =
#
v and hence
#
T =
#
v

#
v
.
We see that the unit tangent vectoi is constant.
15.3. Curvature and normal. lf the cuive desciibed by a vectoi function
#
x(t) is not
a stiaight line, then the tangent to the cuive will tuin as one moves along the cuive. Te
curvature vector
#
measuies how much the cuive is cuived. lt is dened to be the iate
of change of the unit tangent, but with iespect to aic length instead of with iespect to the
given paiametei t. Tus
(38)
#

def
=
d
#
T
ds
.
Accoiding to oui denition of deiivative with iespect to aic length the iight hand side
stands foi
(39)
d
#
T
ds
=
1

#
x

(t)
d
#
T
dt
.
To wiite this completely in teims of the oiiginal vectoi function
#
x(t) we use (3)
(40)
#
=
1

#
x

(t)
d
dt
_
1

#
x

(t)
d
#
x
dt
_
Tis foimula is not as shoit as the oiiginal denition (38), but it does show that the cuiva-
tuie vectoi comes about by dieientiating the vectoi function
#
x(t) twice (and dividing
by
#
x

(t) at the iight moments.)


1. PROBLEMS 31
eorem. e curvature vector
#
is perpendicular to the tangent, i.e.
#

#
T.
Pvooi. We have to show that
#

#
T = 0. Fiom the second foim (39) of the denition
of
#
we see
#

#
T =
_
1

#
x

(t)
d
#
T
dt
_

#
T =
1

#
x

(t)
d
#
T
dt

#
T .
Remembei that
#
T(t) is always a unit vectoi, i.e.
#
T(t) has constant length by 11 this
implies that
d
#
T
dt

#
T(t) and thus
d
#
T
dt

#
T = 0, so we aie done.
Teie aie two concepts that aie deiived fiom the cuivatuie vectoi the curvature
is by denition the length of the cuivatuie vectoi
#
,
(41) =
#
=
_
_
_
_
_
d
#
T
ds
_
_
_
_
_
,
and the normal vector to the cuive is
(42)
#
N =
#

=
d
#
T
ds
_
_
_
d
#
T
ds
_
_
_
.
Te noimal vectoi is undened when
#
=
#
0, because it would iequiie division by zeio.
Since
#
is peipendiculai to
#
T, the noimal vectoi
#
N is also peipendiculai to
#
T (hence
its name).
(43)
d
#
T
ds
=
#
N
16. Osculating plane
At any point X(t) on a space cuive given by
#
x(t) one denes the osculating plane
to be the plane that contains the point X(t) and that is paiallel to both the tangent
#
T(t)
and noimal
#
N(t) of the cuive.
lf we want to wiite a dening equation foi the osculating plane as in 11.2 then
we need a vectoi peipendiculai to the osculating plane. Since this plane is dened to be
paiallel to both
#
T and
#
N, we can nd a noimal vectoi to the osculating plane by taking
the cioss pioduct of
#
T and
#
N. Tis vectoi is called the binormal to the cuive. ln a
foimula, it is dened to be
(44)
#
B =
#
T
#
N.
17. Problems
1. Let be the line given by
#
x(t) =

1
1
0

+t

1
2
1

.
(a) Find the unit tangent vector, the curva-
ture, and the tangent line to the line at the
point where t = 2.
(b) Find the unit tangent vector, the curva-
ture, and the tangent line to the line at any
point on the line.
2. What sign does have in Figure 7 ? How
would the figure change if we change the
32 2. PARAMETRlC CURVES AND VECTOR FUNCTlONS
sign of ? Does the force
#
F on the object
change if we change the sign of ?
3. Suppose a point P is rotating around a
line , keeping its distance to the line fixed
at r, and moving in a plane perpendicular to
the line. Suppose the point has angular ve-
locity : this means that during a time in-
terval of length t the angle swept out by the
line segment connecting P to is exactly t.
In a previous math or physics class it was
shown that the velocity of the point P is r,
where r is the distance from P to the line .
The angular velocity vector is defined to
be the vector
#
whose length is , and that
is parallel to the line . There are two such
vectors (
#
). By definition
#
points in the
direction in which a screw would move if it
were turning in the same direction as the
point P.
(a) Assuming the line passes through the
origin show from the drawing that the ve-
locity vector of the point P is
#
v is given by
#

#
x. You can do this in two steps, namely:
show that
#

#
x has the same direction as
#
v,
show that
#

#
x has the same length as
#
v.
(b) Show that the acceleration vector is
given by
#
a =
#
(
#

#
x). (hint: dont use
the drawing, but combine the definitions of
#
v and
#
a, in (24) and (2) and also the prod-
uct rule; finally, keep in mind that you have
just found that
#
v =
#

#
x.)
(c) If someone told you they had computed
the acceleration vector and found
#
a = (
#

#
)
#
x,
could they be right? Explain! What if they
told you they got
#
a =
#

#
x?
(d) True or False (explain your answers):
(a)
#
v
#
x? (b)
#
a
#
v? (c)
#
a
and
#
x are parallel?
(e) Include the acceleration vector
#
a in the
above drawing.
4. Consider the twisted cubic, i.e. the curve
given by
#
x(t) = t
#
e1 +t
2 #
e2 +t
3 #
e3.
(a) Find a parametrization for the tangent to
the curve at the point where t = 1. Where
does this point intersect the xy-plane?
(b) For any given t find the tangent line to
the curve at the point X(t), and find where
this curve intersects the xy-plane.
(c) If you call that intersection point P(t),
then which curve is traced out by the point
P(t) as t varies?
5. Compute the length of one full turn of the
helix by taking the parametrization given in
(21) and computing the length of the seg-
ment with 0 2.
Aer computing the length, consider
this: let P be the perimeter of the circle un-
derneath the helix, and let H be the height
achieved by one full turn of the helix. Show
that the length L of the helix satisfies L
2
=
P
2
+H
2
.
6. There is a multistory parking ramp where
the way out is a path in the shape of a he-
lix that is wound around the outside of the
building. As a car drives down this path
at night its headlights shine a spot on the
ground. Which curve is traced out by this
light spot as the car drives all the way down?
Origin
s = r = rt
#

#
x
#
v =
#

#
x

r
r
P
P
1. PROBLEMS 33
Make a good drawing. Assume for sim-
plicity that the center of the Parking ramp is
the z-axis.
7. Compute the tangent, curvature, normal
and binormal for the following curves
(a) The parabola:
#
x(t) =
(
t
2
t
)
. At which
point on the curve is the curvature the
largest?
(b) Neils parabola:
#
x(t) =
(
t
2
t
3
)
. At
which point on the curve is the curvature the
largest?
(c) The helix:
#
x() =
(
Rcos
Rsin
a
)
(see 6 for
an explanation of the constants Rand a). At
which point on the curve is the curvature the
largest?
(d) The graph of y = e
x
by using the
parametrization
#
x(t) =
(
t
e
t
)
. Where on
the graph is the curvature the largest?
CHAPTER 3
Functions of more than one variable
1. Functions of two variables and their graphs
1.1. Denition. A function of two vaiiables has two ingiedients a domain and a
rule. Te domain of the function is a collection of points in the xy-plane. Foi each point
(x, y) fiom the domain of the function, the iule should tell us how to nd the function
value f(x, y).
Just as with functions of one vaiiable, the iule that gives us the function value is
ofen specied by some foimula, e.g. f(x, y) = x + y. Te domain of a function is the
set of points at which we dene the function. Tis can in piinciple be any set of points
in the plane. Typically the domain will be a iectangle, oi a disc, oi it could be the entiie
xy-plane, possibly with some points and lines iemoved.
z
height:
z=f(x,y)
D
o
m
a
in
o
f
f
x
y
Figure 1. The graph of some function, and its domain (a rectangle in this example).
1.2. Graphs. By denition, the giaph of a function z = f(x, y) is the collection of
all points (x, y, z) in thiee dimensional space that satisfy the equation z = f(x, y).
Te giaph is usually a suiface that oats above (oi below) the domain of the function
(see Figuie 2).
3
3 3. FUNCTlONS OF MORE THAN ONE VARlABLE
1.3. Level sets. Te giaph of a function of two vaiiables is a suiface siuing in thiee
dimensional space, which can be dicult to diaw oi visualize. lnstead of looking at the
giaph we can also considei its level sets. lf c is any ieal numbei, then, by denition, the
level set at level c of the function is the set of all points (x, y) in the plane that satisfy
f(x, y) = c.
z
c
x
y
level set
at level c
level set at
level c
x
y
Figure 2. The graph of some function (top), and a construction of one of its level sets (boom).
Note that by definition the level set (at level c) is the curve in the xy-plane under the graph: it
is obtained by intersecting the graph of the function with a horizontal plane at height c, and then
projecting this curve of intersection onto the xy-plane.
Since the level set is the set of all solutions to the equation f(x, y) = c, one ofen uses
the notation f
1
(c) (f-inveise of c) foi the level set. We can summaiize the denition
in an equation
f
1
(c) =
_
(x, y) : f(x, y) = c
_
.

Note that the denition says that f


1
(c) is not a number, but a set of points!
1. FUNCTlONS OF TWO VARlABLES AND THElR GRAPHS 3
Level sets tend to be cuives in the xy-plane, although in geneial level sets can have
any shape (see Pioblem .13 foi an example.) Tey aie usually easiei to diaw than the
giaphs of the coiiesponding functions.
1.4. An example from the real world. Heie is a function of local inteiest. Te
domain of the function is the watei suiface of Lake Mendota (lets pietend this is a plane
domain), and the function, which we will call d instead of f, is given by d(x, y) = the
depth of the lake at location (x, y). Teie is no foimula foi this function, but the Wiscon-
sin Depaitment of Natuial Resouices has measuied the depth and piesented the iesults
in teims of the level sets of the function d.
Figure 3. The level curves of a function z = d(x, y). The domain of this function is the lake
surface, and d(x, y) is the depth in meters of Lake Mendota at (x, y). To see the graph of the
function we could try to drain the lake.
See http://limnology.wisc.edu/lake_information/mendota/mendota.html
1.5. Acomment about language and set-theoretic notation. We will ofen say con-
sidei a function z = f(x, y), but theie is a sense in which this is incoiiect. lt is conve-
nient to say considei a function z = f(x, y) since it not only names the function, but
it also gives the independent vaiiables x, y, and the dependent vaiiable z a name. Nev-
eitheless, the symbol in the equation z = f(x, y) that actually iepiesents the function is
f. Te coiiect way of intioducing the function' would be to say considei a function
f.
ln fact, in the notation that is used in modein mathematics one would wiite Considei
the function f : D R Heie f is the name of the function we aie intioducing, D is
'Saying considei the function z = f(x, y) to intioduce the function f is like saying Please meet my
biothei Joe, Bill, and Sue when you want to intioduce youi biothei Joe, who happens to be standing next to
Bill and Sue. To intioduce youi biothei, you would of couise say Please meet my biothei Joe. and to intioduce
the function you should ieally say Considei the function f.
38 3. FUNCTlONS OF MORE THAN ONE VARlABLE
the domain of that function (so D is a set of points in the plane), and R stands foi the set
of ieal numbeis, indicating that computing f always iesults in a ieal numbei.
1.6. Vector notation. lf
#
x is the position vectoi of the point (x, y) in the plane, i.e.
if
#
x = (
x
y
), then one sometimes wiites
f(x, y) = f(
#
x).
Physicists have a piefeience foi
#
r instead of
#
x (because they call the position vectoi the
iadius vectoi), and will wiite f(x, y) = f(
#
r ).
2. Linear functions
Te simplest function of one vaiiable aie those of the foim f(x) = ax + b. Teii
giaphs aie lines, and we called them lineai functions.
A lineai function of two vaiiables is a function f of the foim
(4) z = f(x, y) = ax +by +c,
wheie a, b, c aie constants.
x
y
z
Figure 4. The graph of a linear function z = ax +by +c.
Te giaph of a lineai function is always a plane. lndeed, the giaph consists of all
points (x, y, z) that satisfy the equation
ax by +z = c,
which we can wiite as
#
n
#
x =
#
n
#
p,
wheie
#
n =
_
_
a
b
1
_
_
, and
#
p =
_
_
0
0
c
_
_
.
3. QUADRATlC FORMS 39
3. adratic forms
Afei leaining about lineai functions in pie-calculus one usually goes on to quadiatic
functions. We will do the same foi functions of two vaiiables and study adratic Forms.
Just as in the one vaiiable case wheie quadiatic functions can have a maximum oi min-
imum, quadiatic foims piovide examples of functions of two vaiiables that can have a
maximum oi a minimum, oi, it tuins out, a thiid kind of min-max oi saddle shape.
Tey piovide the basic piole of what we will iun into when we look foi local minima
and maxima of functions of two vaiiables. ln paiticulai, the technique of classifying qua-
diatic foims by completing the squaie, which we will see in this section, is the key to the
second deiivative test foi functions of moie than one vaiiable.
3.1. Denition. Te geneial quadiatic foim in two vaiiables is
(4) f(x, y) = Ax
2
+Bxy +Cy
2
,
wheie A, B, and C aie constants. Depending on the values of these constants the giaphs
of the functions can have a numbei of dieient shapes.
ln addition to these quadiatic foims one can also considei the moie geneial class of
quadiatic functions,
f(x, y) = Ax
2
+Bxy +Cy
2
+Dx +Ey +F,
which also have teims of degiee 1 and 0. We will iestiict ouiselves to quadiatic foims
(foi now).
e prototypical examples. Teie aie seveial impoitant special cases that aie iepie-
sentative of what the giaphs of quadiatic foims can look like. Tese special cases aie
f(x, y) = x
2
+y
2
, and g(x, y) = x
2
y
2
, (4a)
h(x, y) = x
2
, and

h(x, y) = x
2
, (4b)
k(x, y) = xy (4c)
Teii giaphs aie discussed in Figuie .
3.2. Classifying quadratic forms the general procedure. All quadiatic foims have
giaphs that look like one of the examples shown above but how can we tell which it
is` ln othei woids, if Q(x, y) is a given quadiatic foim how can we tell if it is denite,
indenite, oi semidenite` Howdo we knowfoi which (x, y) the foimQ(x, y) is positive
oi negative` lt tuins out that we can always nd out by using the tiick of completing
the squaie.
Te geneial pioceduie foi a given quadiatic foimQ(x, y) = Ax
2
+Bxy +Cy
2
is as
follows
(1) lf A = 0, then we ieally have Q = Bxy +Cy
2
and we can factoi Q as
Q(x, y) = (Bx +Cy)y.
40 3. FUNCTlONS OF MORE THAN ONE VARlABLE
(2) Assume A = 0. We factoi out A, and complete the squaie foi the ist two
teims
Q(x, y) = A
_
x
2
+
B
A
xy +
C
A
y
2
_
= A
_
_
x +
B
2A
y
_
2

_
B
2A
y
_
2
+
C
A
y
2
_
= A
_
_
x +
B
2A
y
_
2
. .
u
2
+
4AC B
2
4A
2
y
2
. .
v
2
_
.
(3) lf 4AC B
2
> 0, then the expiession in biaces is positive, and we can wiite
Q(x, y) = A(u
2
+v
2
), wheie u = x +
B
2A
y, and v =

4AC B
2
2A
y.
Depending on the sign of A oui function is always positive oi always negative,
and we say the foim is positive denite oi negative denite.
The two forms f and g from (4a)
are called definite, since they cannot
change sign:
f(x, y) = x
2
+y
2
is the sum of two squares, and there-
fore is always positive, unless both x
and y vanish. Similarly, g(x, y) =
f(x, y) is always negative, except
at (x, y) = (0, 0).
The form h(x, y) = x
2
is called semi-
definite because it too cannot change
its sign. Clearly, h(x, y) = x
2
is
never negative, but for h(x, y) to be
positive, we need x = 0. So, the func-
tion h(x, y) is positive, except on the
line x = 0 (the y axis). The graph of
the function

h(x, y) = y
2
is simi-
lar, but upside down.
The form k(x, y) = xy is called in-
definite, because it can be both posi-
tive and negative: if x and y have the
same sign, then xy > 0, but if they
have opposite signs, then xy < 0.
Thus the graph of z = xy lies above
the xy-plane in the first and third
quadrants, and belowthe xy-plane in
the second and fourth quadrants.
xy > 0
xy > 0
xy < 0
xy < 0
x
y
Figure 5. Graphs of some representative quadratic forms.
3. QUADRATlC FORMS 41
(4) lf 4AC B
2
< 0, then we have
Q(x, y) = A(u
2
v
2
), wheie u = x +
B
2A
y, and v =

B
2
4AC
2A
y.
When this happens we can factoi the quadiatic foim, i.e. we have
Q(x, y) = A(u +v)(u v).
Te foim is indenite.
() in the only iemaining case we have 4AC B
2
= 0, so that
Q(x, y) = A
_
x +
B
2A
y
_
2
.
ln this case the foim is a peifect squaie (times A). Te foim is semi-denite.
To undeistand this pioceduie it is peihaps best to look at how it woiks in some examples.
3.3. Classifying quadratic forms two examples.
3.3.1. An indenite quadratic form. Considei the foimQ(x, y) = 3x
2
+9xy +6y
2
.
We iewiite this as follows
Q = 3x
2
+ 6xy + 9y
2
= 3
_
x
2
2xy 3y
2
_
= 3
_
x
2
2xy +y
2
. .
4y
2

complete the square


= 3
_
(x y)
2
4y
2

in this case we get the dierence of two


squares, so use a
2
b
2
= (a b)(a +b)
= 3(x y 2y)(x y + 2y)
= 3(x 3y)(x +y).
Tis shows that Q(x, y) > 0 when y >
1
3
x oi y < x, and Q(x, y) < 0 when x < y <
1
3
x.
y
x
Q(x,y)<0
Q(x,y)<0
+
+
+
+ +
+
+ +
+
+
+
+
+
+
+
+
+
+
+
+ +
+
+
+
Figure 6. The signs of the quadratic form in example 3.3.1.
42 3. FUNCTlONS OF MORE THAN ONE VARlABLE
3.3.2. Apositive denite quadratic form. To see a dieient example, considei the qua-
diatic foim Q(x, y) = 2x
2
4xy + 6y
2
. By completing the squaie we can wiite it as
Q(x, y) = 2
_
x
2
2xy + 3y
2
_
= 2
_
x
2
2xy +y
2
+ 2y
2
_
the square is complete
= 2
_
(x y)
2
+ 2y
2
_
= 2(x y)
2
+ 4y
2
.
We see that this paiticulai quadiatic foim is positive denite.
4. Functions in polar coordinates r,
Recall that instead of using Caitesian cooidinates (x, y) to specify the location points
in the plane, we can also use polai cooidinates. ln many cases it is much easiei to desciibe
a function using polai cooidinates than in Caitesian cooidinates.
To go back and foith between Caitesian and Polai Cooidinates we can use the fol-
lowing ielations
x = r cos (48a)
y = r sin (48b)
r =

x
2
+y
2
(48c)

= aictan
y
x

(48d)
Te equation foi is only valid foi x > 0, wheie

2
< <

2
. ln othei iegions of the
plane theie aie othei expiessions ielating to (x, y). See pioblem .8.

r
x
y
P

0
=
0
r
=
r
0
Figure 7. Polar coordinates are defined in the picture on the right (see also equations (48)). On
the le: the set of points at which has one given value 0 form a half line emanating from the
origin that makes an angle 0 with the positive x-axis. The set of points at which r has a given
value r0 form a circle centered at the origin, with radius r0.
Te simplest kinds of functions one can considei in polai cooidinates aie those that
only depend on one of those cooidinates, i.e. functions that only depend on the iadius r,
and functions that only depend on the polai angle . Lets look at some examples of such
functions.
4. FUNCTlONS lN POLAR COORDlNATES r, 43
x
y
z
z = r =

x
2
+y
2
r
z
z
=

(
r
)
=
r
Figure 8. Radially symmetric functions. The graph of z = r.
4.1. Radially symmetric functions. Te functions
f(x, y) = x
2
+y
2
, g(x, y) =

x
2
+y
2
, h(x, y) = ln
_
x
2
+y
2
_
,
all can be expiessed in teims of the iadius r only. Namely, using r
2
= x
2
+y
2
, we have
f(x, y) = r
2
, g(x, y) = r, h(x, y) = ln r
2
(= 2 ln r).
ln geneial, a function z = f(x, y) that can be wiiuen in teims of the iadius r only, i.e. a
function foi which theie is some function of one vaiiable with
f(x, y) = (r), i.e. f(x, y) =
_

x
2
+y
2
_
,
is called a radially symmetric function.
Since a iadially symmetiic function only depends on the iadius r, its level sets consist
of ciicles centeied at the oiigin (one exception the oiigin, r = 0 can also be a level set,
and this is obviously not a ciicle but a point.)
As an example, we considei the function g(x, y) =

x
2
+y
2
= r in moie detail.
Te function of one vaiiable heie is (r) = r. We can tiy to visualize the giaph of g
by ist looking at the positive x-axis only. Teie we have f(x, 0) =

x
2
= x. We get
the giaph of g by ievolving the giaph of z = x aiound the z-axis. See Figuie 8.
4.2. Functions of only. Heie aie two functions that happen to depend on the polai
angle only
f(x, y) = sin , h(x, y) = .
We can iewiite these functions in teims of x and y by using the ielations between Caite-
sian and Polai cooidinates (48). We get
f(x, y) = sin =
y
r
=
y

x
2
+y
2
foi f, and
h(x, y) = = aictan
y
x
foi h, at least in the iight half plane wheie x > 0.
A function that only depends on is constant on iays emanating fiom the oiigin
because the polai angle is constant on such iays. Te level sets of such a function
theiefoie consist of half-lines (iays) staiting at the oiigin. lts giaph consists of spokes
auached to the z-axis. Each spoke lies above a iay in the xy-plane with some polai angle
, and is auached to the z-axis at a height given by the function value. As we vaiy , the
44 3. FUNCTlONS OF MORE THAN ONE VARlABLE
spoke iotates aiound the veitical axis and moves up oi down, as dictated by the function.
Figuie 9 shows what happens foi f(x, y) = sin .

x
y
z=f()
ray
spoke
The graph of a function of only
consists of horizontal spokes
aached to the z-axis.
The graph of z = sin
(the x-axis is coming right at us.)
Figure 9
Te function z = has a simplei foimula in polai cooidinates but actually has a
moie complicated giaph. Let us tiy to visualize its giaph the spokes that make up the
giaph aie hoiizontal, auached to the z-axis, and aie at height . lf we inciease the angle
the spokes go up at a steady iate in a way that should iemind us of a helix (see
and Figuie ). Based on this desciiption its giaph should look like the suiface diawn in
Figuie 10. Te suiface is called the helicoid, and it is not the giaph of a function (it fails
the veitical line test.) We could have known this fiom the beginning , because when we
desciibed oui function as f(x, y) = , we should have immediately asked which ? Te
polai angle of any given point is only deteimined up to a multiple of 2. Te giaph
that we have diawn of the function z = ieects this. To make h(x, y) = into an
honest function we have to say which of the many possible angles we choose when we
aie given a point. One possible choice is to always iequiie the polai angle to lie between
0 and 2 (iadians). Moie piecisely, we can insist on
0 < 2.
lf we do this then theie is a unique angle foi each point (x, y) in the plane. Te giaph
of this function is shown on the iight in Figuie 10.
5. Methods of visualizing the graph of a function
5.1. Freezing a variable. lf a function is not familiai, then a good stiategy foi diaw-
ing its giaph is to freeze a variable. ln othei woids, to analyze a function z = f(x, y)
we pietend y is a constant then x is the only independent vaiiable, and we can tiy to
diaw the giaph of the function z = f(x, y), now thinking of this as a function of only
one vaiiable. Tis giaph is a cuive in the xz plane. We get one such cuive foi each choice
of y. Piecing these giaphs togethei then gives us the giaph of the two-vaiiable function
z = f(x, y).
We could apply the same pioceduie with the ioles of x and y switched i.e. foi each
xed x you tiy to giaph z = f(x, y) as a function of the vaiiable y only, afei which we
tiy to t all the giaphs we get foi dieient values of x togethei.
x
y
z
. METHODS OF VlSUALlZlNG THE GRAPH OF A FUNCTlON 4
x
y
x
y
Figure 10. The graph of z = is the helicoid. It is not the graph of a function, but one can extract
a function by choosing a branch of the function. One possible choice, drawn here on the right,
is to restrict the polar angle to the interval 0 < 2. There are many other possible choices.
5.2. Moving graphs. Teie is anothei way of visualizing a function z = f(x, y) of
two vaiiables in which we think of one of the independent vaiiables (e.g. y) as time. Te
nal pictuie is not one static image of a thiee dimensional suiface, but iathei a movie of
a giaph that is moving aiound in the xz plane.
lf we have a function z = f(x, y), then let us think of y as time, and let us ielabel
it as t, so that we aie looking at the function z = f(x, t). Now at each moment in time
t we can think of z = f(x, t) as a function of one vaiiable x whose giaph we can tiy to
diaw, iegaiding it as a still-image. Ten, as we let time t vaiy, puuing the still images in
a sequence, you get a movie of a giaph of a changing function of one vaiiable.
Foi instance, if the function is (once again) the saddle suiface function z = xy, then
we would be consideiing the function z = xt. At each moment t the giaph of z = xt is
t=1
z
x x x x x
z z z z
t=1 t=1/2 t=0 t=1/2
Figure 11. The saddle movie. Its about a line segment whose slope changes, even though it is
otherwise stuck to the origin.
4 3. FUNCTlONS OF MORE THAN ONE VARlABLE
a line with slope t. Puuing these giaphs togethei gives a movie which begins with a line
of iathei negative slope, duiing the movie the slope incieases, and in the middle of the
movie oui line has achieved hoiizontality, nally, the closing shot piesents us with a line
with a veiy positive slope. Figuie 11 shows some stills fiom the movie.
Tis inteipietation is not veiy dieient fiom the pioceduie of fieezing the y vaii-
able. Te only ieal dieience lies in what we do with all the sepaiate giaphs we get afei
we fieeze a vaiiable. ln one case we tiy to piece them togethei to make a biggei diaw-
ing of a thiee-dimensional object, in the othei we put them togethei to make a motion
pictuie.
Problems
In the problems in this stage of the course, you will be asked to sketch the graph of a function.
From math 221 you remember that this meant you had to find minima, maxima, inflection points,
and other features of the graph. In 234 you will learn to do the same for functions of two (and
more) variables, but for now you should try to use the method of freezing a variable or other
similar tricks to get an idea of what the graph of f looks like.
You can use a graphing program (such as Grapher.app on the Mac, GraphCalc on Windows,
or one of the many websites such as http://www.graphycalc.com/) to check your answer.
Note: very oen students try to fit
their drawings into a region the size
of a post-it. In this course, whenever
you make a drawing, especially if its
a three-dimensional drawing, make it
large! Use half a page for a drawing.
Make sure you have enough paper, try
to find lots of cheap scrap paper.
1. If we were to drain Lake Mendota, as sug-
gested in 1.4, would the lake boomgive us
the graph of d(x, y) or of d(x, y)? (where
d is the depth of the lake)?
2. What are the signs of the coeicients a,
b, and c for the linear function whose graph
is drawn in Figure 4?
3. About planes and their intersections with
the coordinate axes.
(a) Where does the plane z = 3x y + 6
intersect the three coordinate axes?
(b) Find the equation for the plane that in-
tersects the x-axis at x = 4, the y-axis at
y = 2, and the z-axis at z = 3.
(c) Find the equation for the plane that in-
tersects the x-axis at x = a, the y-axis at
y = b, and the z-axis at z = c. (Write the
equation as nice as possible.)
4. Find a formula for the distance to the ori-
gin of the graph of (4).
5. Classify the following quadratic forms as
definite, indefinite, or other, by completing
the square. Determine the zero set for each
of these quadratic forms.
(a) f(x, y) = x
2
+ 2y
2

(b) Q(x, y) = x
2
y
2

(c) g(x, y) = x
2
4xy + 3y
2

(d) Q(s, t) = 9s
2
36st + 81t
2

(e) M(, ) =
1
2

2
+
2
.
(f) Q(x, y) = xy +y
2

(g) Q(x, y) = x
2
+ 2xy
6. For which values of the constant k is the
quadratic form
Q(x, y) = x
2
+ 2kxy +y
2
positive definite?
7. Which functions of two variables z =
f(x, y) are defined by the following formu-
lae?
PROBLEMS 4
Find draw the domain of each function
(the largest domain on which the definition
would make sense).
Try to sketch their graphs.
Draw the level sets for each function.
(a) z = xy
(b) z x
2
= 0
(c) z
2
x = 0
(d) z x
2
y
2
= 0
(e) z
2
x
2
y
2
= 0
(f) xyz = 1
(g) xy/z
2
= 1
(h) x +y +z
2
= 0
(i) x +y +z
2
= 1
8. The following expressions are all equal to
the polar angle in some region of the xy-
plane. Explain why the expression gives ,
and identify in which region this holds.
(a) = aictan
y
x

(b) = + aictan
y
x

(c) = 2 + aictan
y
x

(d) =

2
aictan
x
y

(e) = aicsin
y

x
2
+y
2
.
9. The level set is always a curve not!
If d(x, y) is the depth function of Lake Men-
dota (see 1.4), then what are the level sets
d
1
(c) for c = 0, c = +24 and for c = 24
(meters)? What is the level set d
1
(400)
(meter)?
10. Describe and explain the relation be-
tween the graph of the function y = g(x) of
one variable, and the corresponding function
f(x, y) = g
(
x
2
+y
2
)
of two variables.
What do the level sets of f(x, y) look
like?
For instance, if g(x) = x, then f(x, y) =

x
2
+y
2
: what is the relation between the
graphs of g and f?
11. Find the largest domain on which the
following functions of two (or occasionally
three) variables can be defined:
(a) f(x, y) =

9 x
2
+

y
2
4
(b) f(x, y) = aicsin(x
2
+y
2
2)
(c) f(x, y) =

x

y
(d) f(x, y) =

xy
(e) f(x, y, z) = 1/

xyz
(f) f(x, y) =

16 x
2
4y
2

12. Here are two sets of level curves with lev-


els z = 0.2, 0.4, 0.6, 0.8, 1.0, 1.2, 1.4. One
is for a function whose graph is a cone (z =

x
2
+y
2
), the other is for a paraboloid
(z = x
2
+ y
2
). Which is which? Explain.

13.

Let Q be the square in the plane con-
sisting of all points (x, y) with |x| 1,
|y| 1. This problem is about the so-called
distance function to Q. This function is de-
fined as follows: f(x, y) is the distance from
the point (x, y) to the point in Q nearest to
(x, y).
(a) Which point in Q is nearest to (0,
1
2
)?
Which is closest to (0, 2)? Which is closest
to (3, 4)?
(b) Compute f(0,
1
2
), f(0, 2) and f(3, 4)).
(c) What is the zero set of f?
(d) Draw the level sets of f at levels 1,
1, 2, and 3. Describe the general level set
48 3. FUNCTlONS OF MORE THAN ONE VARlABLE
f(x, y) = c where c is an arbitrary number.

(e) Give a formula for f(x, y). (It turns out


to be too hard to capture the distance func-
tion in one formula. You will have to split
the plane into dierent regions and describe
f(x, y) by dierent formulas, according to
which region (x, y) belongs to.)
14. Describe the movie that goes with each
of the following functions.
(a) f(x, t) = xsin t
(b) f(x, t) = xsin 2t
(c) f(x, t) = t sin x
(d) f(x, t) = 2t sin x
(e) f(x, t) = t sin 2x
(f) f(x, t) = (x t)
2

(g) f(x, t) = (x sin t)


2

(h) f(x, t) = (x t
2
)
2
(i) f(x, t) =
t
2
1 +x
2
(j) f(x, t) =
1
(1 +x
2
)(1 +t
2
)

15. Describe the movie that goes with the
function
f(x, t) = aictan
x
t
,
for t > 0. The function is not defined at
t = 0, but can you describe the limit of this
function as t 0? (Hint: the sign of x mat-
ters).
16. If y = g(x) is any function of one vari-
able, then a function of the form f(x, t) =
g(xct) is oen called a traveling wave with
wave speed c and profile g. Let g be any
non constant function of your choice and de-
scribe the movie presented by the function
f(x, t) = g(x ct) (cant choose? Then try
Agnesis witch g(x) =
1
1+x
2
.)
The number c is called the wave speed.
If c > 0 is the motion to the le or to the
right? Explain.
17. If y = g(x) is any function of one vari-
able, then a function of the form
f(x, t) = cos(t)g(x)
is oen called a standing wave. Let g be any
non constant function of your choice and de-
scribe the movie presented by the function
f(x, t) = cos(t)g(x) (cant choose? Then
try Agnesis witch g(x) =
1
1+x
2
again, or
for this example, try g(x) = sin x.)
The number

2
is called the frequency
of the standing wave. The function g(x) is
called its profile. How long does it take be-
fore the standing wave returns to its original
position, i.e. what is the smallest T > 0 for
which f(x, T) = f(x, 0) for all x? Explain.

CHAPTER 4
Derivatives
1. Interior points and continuous functions
Befoie diving into the calculus of paitial deiivatives we need to discuss ceitain as-
sumptions that we shall always implicitly make about the functions in this couise. Te
ist conceins the domains of oui functions. Namely
(49) We only consider functions at interior points of their domain
Heie, by denition, a point (a, b) in the domain of a function is called an interior point if
the function is also dened at all points (x, y) that lie within some small disc centeied at
(a, b).
P
1
P
2
P
3
d
o
m
a
i
n
o
f
f
Q
Figure 1. Interior and boundary points in the domain of f: P1, P2, or P3 are interior points
in the domain. Each of these points is the center of a suiciently small disc that is still contained
in the domain. For points such as Q, that lie on the edge of the domain, any disc centered at Qwill
stick out of the domain, no maer how small the disc is chosen. If we talk about the derivative
of a function at some point in its domain, then, in this course, we will always assume that we are
not at an edge-point like Q.
Te othei standing assumption we make in this couise is that
(0) all functions we consider are continuous.
We have seen the concept of continuity foi functions of one vaiiable. Foi functions of
moie vaiiables continuity has a similai denition. ln this couise we will aim foi an
intuitive undeistanding of the concept, which can be foimulated as follows.
e function z = f(x, y) is continuous at some point (a, b) if the function
value f(x, y) at any point (x, y) is close to f(a, b) when (x, y) is close
to (a, b).
49
0 4. DERlVATlVES
Teie aie many othei ways of desciibing continuity, e.g. one can say that f is continuous
at (a, b) if
lim
(x,y)(a,b)
f(x, y) = f(a, b).
To make this piecise we would have to dene what lim
(x,y)(a,b)
. . . means.
A piecise denition of f is continuous at (a, b) invokes s and s
e function z = f(x, y) is continuous at some point (a, b) if for every
> 0 there is a > 0 such that for every point (x, y) that lies in the
disc of radius centered at (a, b) one has |f(x, y) f(a, b)| < .
ln this couise we will not use the denition much, but we will occasionally appeal to the
intuitive notion of continuity. Te pioblems show some examples of how a function of
two vaiiables can fail to be continuous (e.g. Pioblem 3.1).
Nowthat we have dispensed with these pieliminaiy issues, we can go on to the cential
topic in the ist half of the semestei paitial deiivatives and the chain iule.
2. Partial Derivatives
Te deiivative f

(x) of a function of one vaiiable, y = f(x), measuies a iate of change


if we inciease x by a small amount x then y = f(x) also incieases by a small amount
y. Te iatio between these two changes is the deiivative f

(x)
y
x
.
Foi a function z = f(x, y) of two vaiiables theie is a similai concept if we change
x and/oi y by a small amount then z will also change by a small amount, and theie aie
foimulas ielating the changes x, y and z. Because theie aie many dieient ways in
which we can change x and y theie aie a few dieient foimulas. We will encountei the
following veisions of the deiivative of f(x, y)
Change only one of the vaiiables but not the othei this leads to the so-called partial
derivatives.
Simultaneously vaiy both x and y the iesulting change tuins out to be the sum
of the changes we would get if we weie to vaiy only x oi only y, iespectively. Tis will
follow fiom the ain rule, and the iesulting foimula is called the total derivative.
We begin with the paitial deiivatives.
2.1. Denition of Partial Derivatives. If z = f(x, y) is a function of two variables
then the partial derivatives of f with respect to x and with respect to y are
(1)
f
x
(x, y) = lim
x0
f(x + x, y) f(x, y)
x
and
(2)
f
y
(x, y) = lim
y0
f(x, y + y) f(x, y)
y
Te following moie convenient notation is used veiy ofen (because its so much shoitei)
(3) f
x
(x, y) =
f
x
(x, y), f
y
(x, y) =
f
y
(x, y).
When we aie in a huiiy we can also diop the (x, y) fiom oui notation foi deiivatives
and just wiite f
x
and f
y
.
3. PROBLEMS 1
y
x
f
y
is the rate of change of f in the vertical direction
f
x
is the rate of change of f in the horizontal direction
When we define the partial derivatives at some point
(x, y), we assume that the function is defined on some
suiciently small disc centered at that point (x, y).
Figure 2. The partial derivatives of a function at some point (x, y) measure how fast the func-
tion f(x, y) changes if we move the point either horizontally (the x direction) or vertically (the y
direction).
2.2. Partial derivatives of functions of three or more variables. lf a function de-
pends on thiee oi moie vaiiables then one can dene its paitial deiivatives in the same
way as foi functions of two vaiiables. Foi instance, if w = f(x, y, z) is a function of thiee
vaiiables, then its paitial deiivative with iespect to x is dened to be
f
x
= lim
x0
f(x + x, y, z) f(x, y, z)
x
.
Te deiivatives of f with iespect to y and z have veiy similai denitions.
2.3. Examples. Computing paitial deiivatives is not haidei than computing oidi-
naiy deiivatives. To nd the paitial deiivative of a function with iespect to x we just
pietend all othei vaiiables aie constants and dieientiate. Oi, in othei woids, we could
think of the paitial deiivative of f(x, y) with iespect to x as the oidinaiy deiivative of
the function f in which we have fiozen the vaiiable y at some paiticulai value.
Foi instance, the paitial deiivatives of the function f(x, y, z) = x
2
sin y +z of three
vaiiables x, y, and z, aie
f
x
= 2xsin y, f
y
= x
2
cos y and f
z
= 1.
3. Problems
1. For each of the following functions sketch
the graph (use a graphing program, if nec-
essary) and decide if you think the function
has a limit as (x, y) approaches (0, 0).
(a) f(x, y) =
xy
x
2
+y
2
(b) g(x, y) =
1
x
2
+y
2
(c) h(x, y) =
x
x
2
+y
2
.
(d) p(x, y) =
x

x
2
+y
2
.
(e) q(x, y) =
x
2

x
2
+y
2
.
2. Find the partial derivatives of the follow-
ing functions:
(a) f(x, y) = x
2
y
3
x
3
y
2
.
(b) f(x, y) = cos(x
2
y) +y
3
.
(c) f(x, y) =
xy
x
2
+y
.
(d) f(x, t) = (x +t)
4
.
(e) f(x, t) = (x t)
4
.
(f) f(x, t) = sin t cos
2x
L
.
2 4. DERlVATlVES
(g) f(x, y) = e
x
2
+y
2
.
(h) f(x, y) = xy ln(xy).
(i) f(x, y) =

1 x
2
y
2
.
(j) f(x, y, z) =

x
2
+y
2
+z
2
(k) f(u, v) = e
u+v
(l) f(x, y) = xtan(y).
(m) f(x, y) =
1
xy
.
3. Let r be the radius in polar coordinates,
as defined in 4 of Chapter III.
(a) Compute the partial derivatives of r.
(b) Showthat the partial derivatives of r can
be wrien as
r
x
=
x
r
,
r
y
=
y
r
.
4. Let be the polar angle function, defined
in 4.2 of Chapter III.
(a) In the le half plane the function is de-
fined by
(x, y) = aictan
y
x
.
Use this expression to find its partial deriva-
tives,

x
and

y
.
(b) Check that the angle function also satis-
fies
xsin = y cos
at all points in the plane. Use implicit dier-
entiation to find the partial derivatives

x
and

y
.
5. Let f(x, y) = the distance from (x, y) to
the origin. Find a formula for f, and com-
pute
fx, fy, and

f
2
x
+f
2
y
.
(Hint: compare this problem with problem
3.3.)
6. Suppose f(t) and g(t) are single variable
dierentiable functions. Find z/x and
z/y for each of the following two variable
functions.
(a) z = f(x)g(y)
(b) z = f(xy)
(c) z = f(x/y)
7. Let f be the distance to the square Q
function from problem 5.13. Find the partial
derivatives fx and fy of f. (You will need
your answer to problem 5.13, in particular
the description of f as a piecewise defined
function.)
4. e linear approximation to a function
4.1. e Chain Rule and friends. When we compute the paitial deiivative of a func-
tion with iespect to a vaiiable x we pietend all othei vaiiables aie constants, and just dif-
feientiate with iespect to x, just as we would in ist semestei calculus. Teie is theiefoie
no need to state a pioduct iule oi quotient iule, because these aie exactly the same as foi
functions of one vaiiable. Te chain iule on the othei hand is dieient theie is a chain
iule foi functions of seveial vaiiables, but it has moie teims than the chain iule fiom
one-vaiiable calculus. Teie aie seveial ielated topics that t togethei in a discussion of
the chain iule, namely Linear Approximation, Tangent Planes to a Graph, and e
Total Derivative. We will go thiough these one at a time in the next few sections.
4.2. e linear approximation formula. Te key to the chain iule is the lineai ap-
pioximation foimula. Tis foimula tells us appioximately how much a function z =
f(x, y) of two vaiiables changes if both vaiiables aie subjected to a small change.
Moie piecisely, if we have a function z = f(x, y), and we know its value f(x
0
, y
0
)
at some point (x
0
, y
0
), then how much does the function value change if x is incieased
fiom x
0
to x
0
+ x, and if y is similaily incieased fiom y
0
to y
0
+ y`
4. THE LlNEAR APPROXlMATlON TO A FUNCTlON 3
x
0
x
0
+ x
y
0
y
0
+ y
We can change (x
0
, y
0
) to (x
0
+x, y
0
+y) in two
steps:
first keep y fixed and increase x by x,
then keep x fixed and increase y by y
( x, y
0
)
(x
0
+ x, y)
To express the change in function values in terms of
derivatives, we can use the Mean Value Theorem. We
get two intermediate points:
one at x = x for the increase in f when x changes,
and
one at y = y for the increase in f when y changes.
Figure 3. Computation of the linear approximation (4)
Te basic idea in the computation of the change in f(x, y) is to go fiom (x
0
, y
0
) to
(x
0
+ x, y
0
+ y) in two steps
f = f(x
0
+ x, y
0
+ y) f(x
0
, y
0
) (4)
= f(x
0
+ x, y
0
+ y) f(x
0
+ x, y
0
)
. .
only y changes
+f(x
0
+ x, y
0
) f(x
0
, y
0
)
. .
only x changes
We have wiiuen the total change in f as the sum of two changes, one of them caused by
the change in x, and the othei due to the change in y. See Figuie 3.
ln the second dieience only x changes while y iemains the same, so we can use the
one vaiiable Mean Value Teoiem to conclude that theie is some numbei x between x
0
and x
0
+ x with
f(x
0
+ x, y
0
) f(x
0
, y
0
)
x
= f
x
( x, y
0
),
i.e.
() f(x
0
+ x, y
0
) f(x
0
, y
0
) = f
x
( x, y
0
) x.
Likewise, in the dieience in (4) wheie only y changes we can use the Mean Value
Teoiem to conclude that theie is some y between y
0
and y
0
+ y such that
f(x
0
+ x, y
0
+ y) f(x
0
+ x, y
0
)
y
= f
y
(x
0
+ x, y),
and hence
() f(x
0
+ x, y
0
+ y) f(x
0
+ x, y
0
) = f
x
(x
0
+ x, y) x.
lf we now combine () and () with (4) then we get
f = f
x
( x, y
0
) x +f
y
(x
0
+ x, y) y.
Tis equation is exactly tiue, i.e. we have not made any appioximations, and we have not
ignoied any kind of eiioi teims. Howevei, the equation does contain the numbeis x
and y, which aie piovided by the Mean Value Teoiem, and of which we theiefoie do not
4 4. DERlVATlVES
know anything besides the fact that x lies between x
0
and x
0
+x, and y lies between y
0
and y
0
+ y. We can get iid of this unceitainty by seuling foi an appioximation foi f
instead of the exact expiession we have just found. To do this we assume that x and y
aie small. Ten, since x lies between x
0
and x
0
+ x, we know that x x
0
. We also
know that y
0
+y y
0
, so, if the function f
x
is continuous, then it seems ieasonable to
assume that
() f
x
( x, y
0
+ y) f
x
(x
0
, y
0
).
Similaily, we will assume that
(8) f
y
(x
0
, y) f
y
(x
0
, y
0
).
Substituting this in (4) we nd
(9) f f
x
(x
0
, y
0
)x +f
y
(x
0
, y
0
)y
Keeping in mind that f = f(x
0
+ x, y
0
+ y) f(x
0
, y
0
), we conclude
(0) f(x
0
+ x, y
0
+ y) f(x
0
, y
0
) +f
x
(x
0
, y
0
)x +f
y
(x
0
, y
0
)y
Te lineai appioximation foimula (0) is ofen wiiuen using Leibniz-style notation foi
the deiivatives, wheie one wiites
f
x
foi f
x
, and
f
y
foi f
y
. ln this notation the appioxi-
mation foimula takes these foims
f(x
0
+ x, y
0
+ y) f(x
0
, y
0
) +
f
x
(x
0
, y
0
) x +
f
y
(x
0
, y
0
) y,
oi, shoitei,
(1) f
f
x
x +
f
y
y.
Te appioximation (0) can also be wiiuen without x and y by a change of nota-
tion. To do this we intioduce
(2) x = x
0
+ x and y = y
0
+ y,
and inteipiet (0) as a foimula that tells us appioximately what the function value at
(x, y) is, piovided (x, y) is close enought to (x
0
, y
0
). Wiiuen in teims of x and y, (0)
says
(3) f(x, y) f(x
0
, y
0
) +f
x
(x
0
, y
0
) (x x
0
) +f
y
(x
0
, y
0
) (y y
0
).
4.3. Linear approximation innitesimal version. We expect the appioximation
in (1) to impiove as we deciease x and y (and we will tiy to make this statement
moie piecise in the next section, 4.4). We could then say, as is commonly done, that
theie is an exact equation when x and y aie innitely small, and wiite this equation
as
(4) df =
f
x
dx +
f
y
dy.
Te meaning of this equation is that innitesimally small changes in x and y, of magni-
tudes dx and dy, iespectively, lead to an innitesimally small change in f of magnitude
df, and that df, dx, and dy aie ielated by (4). Even though it is veiy dicult to make
sense of the innitely small quantities dx, dy, df, in (4), this notation is widely used,
because the make-belief it entails allows one to ignoie the moie awkwaid eiioi teims
that we will now discuss.
. THE TANGENT PLANE TO A GRAPH
4.4. e linear approximation formula with error term. ln oui computation of the
change f of the function we appioximated f
x
( x, y
0
) by f
x
(x
0
, y
0
), and f
y
(x
0
+x, y)
by f
y
(x
0
, y
0
). As a iesult oui lineai appioximation foimula (0) is not an exact equation,
but only says that one thing is appioximately equal to anothei.
We can make this a bit moie piecise by including eiioi teims, i.e. by saying that theie
aie small numbeis e
x
and e
y
such that
f
x
( x, y
0
) = f
x
(x
0
, y
0
) +e
x
, and f
y
(x
0
+ x, y) = f
y
(x
0
, y
0
) +e
y
.
Heie e
x
and e
y
depend on x and y, and as both x and y go to zeio, the eiiois e
x
and e
y
will also go to zeio.
Puuing this in (4) we get the lineai appioximation foimula with eiioi teims
() f(x
0
+ x, y
0
+ y) = f(x
0
, y
0
) +f
x
(x
0
, y
0
)x +f
y
(x
0
, y
0
)y
. .
lineai appioximation
+e
x
x +e
y
y
. .
eiioi
in which e
x
and e
y
depend on x, y, and satisfy
lim
x,y0
e
x
= lim
x,y0
e
y
= 0.
lf we ignoie the eiioi teim then we iecovei the lineai appioximation foimula (0). Oui
moie piecise lineai appioximation foimula () tells us that the eiioi in (0) (dieience
between lef and iight hand sides) is given by e
x
x +e
y
y, and that this eiioi is small
compaied to x and y. We could wiite this as
Eiioi in the appioximation = e
x
x +e
y
y = o(x) +o(y).
5. e tangent plane to a graph
5.1. e tangent plane. Foi a function z = f(x, y) and a point (x
0
, y
0
) the lineai
appioximation (3) gives us an appioximation foi the function f at any othei point (x, y)
neai (x
0
, y
0
). lt says
z f(x
0
, y
0
) +f
x
(x
0
, y
0
)(x x
0
) +f
y
(x
0
, y
0
)(y y
0
).
lf we ieplace by equality, then we get a new function of (x, y)
() z = f(x
0
, y
0
) +f
x
(x
0
, y
0
)(x x
0
) +f
y
(x
0
, y
0
)(y y
0
).
Keeping in mind that f(x
0
, y
0
), f
x
(x
0
, y
0
), and f
y
(x
0
, y
0
) aie constants, while only (x, y)
aie vaiiables heie, we see that this is the equation foi a plane which we call the tangent
plane to the giaph of f at the point (x
0
, y
0
, f(x
0
, y
0
)).
5.2. Example: tangent plane to the saddle surface at the origin. Find the equation
for the tangent plane to the saddle surface z = xy at the origin.
Solution: Te saddle suiface is the giaph of the function f(x, y) = xy whose paitial
deiivatives aie f
x
(x, y) = y and f
y
(x, y) = x. To nd the tangent plane at x
0
= 0,
y
0
= 0, we compute the paitial deiivatives,
f
x
(x, y) =
xy
x
= y, so at (x
0
, y
0
) = (0, 0) we have f
x
(0, 0) = 0,
and
f
y
(x, y) =
xy
y
= x, so at (x
0
, y
0
) = (0, 0) we have f
y
(0, 0) = 0,
4. DERlVATlVES
y+

y
y
x
+

x x
fy
y
f
x
x

y
Linear approximation
to the graph of z=f(x, y)

x
y+

y
y
x
+

x x
fy
y
f
x
x

x
fx
x+fy
y
Figure 4. Top: The graph of the linear approximation of f (graph of f itself is not shown see
the boom figure). If we increase x by x, then f will increase by approximately fxx, and if we
increase y by y, then f increases by approximately fyy. If we increase x and y by x and y
at the same time, then f increases by roughly fxx + fyy. The vertical doed line behind the
parallelogram represents this increase in f.
Boom: The graph of a function, and of its tangent plane at some point (x0, y0, z0). The
tangent plane is the graph of the linear approximation to f.
Moieovei, we also have f(x
0
, y
0
) = f(0, 0) = 0, so that the equation foi the tangent
plane is
z = 0 + 0 (x 0) + 0 (y 0) = 0,
i.e.,
z = 0.
Te tangent plane at the oiigin is just the xy-plane.
5.3. Example: another tangent plane to the saddle surface. Find the equation for
the tangent plane to the saddle surface z = xy at the point (2, 1, 2). Where does this plane
intersect the coordinate axes?
Solution: Tis is almost the same pioblem as befoie. Te only dieience is that we aie
tiying to nd the tangent plane at a point othei than the oiigin. To get the tangent plane
at the point (x
0
, y
0
) = (2, 1) we compute the deiivatives
f
x
(x, y) = y f
x
(2, 1) = 1,
. THE TANGENT PLANE TO A GRAPH
Figure 5. The graph of z = xy and the tangent plane at the origin.
and
f
y
(x, y) = x f
y
(2, 1) = 2.
Te equation foi the tangent plane is theiefoie
z = x
0
y
0
+y
0
(x x
0
) +x
0
(y y
0
) ()
= 2 + 1 (x 2) + 2 (y 1)
= 2 +x + 2y
Te inteisections with the x, y and z axes aie, iespectively, (2, 0, 0), (0, 1, 0), and (0, 0, 2).
5.4. Example: tangent plane to a sphere. e point (x
0
, y
0
, z
0
) lies on the upper half
of the sphere with radius 4 centered at the origin. Find an equation for the tangent plane to
the sphere at that point, if x
0
= 1 and y
0
= 3.
Solution: Te equation foi the spheie
is x
2
+ y
2
+ z
2
= 4
2
= 16, so the uppei
half is the giaph of the function
f(x, y) =

16 x
2
y
2
.
Te z cooidinate of the given point is
theiefoie z
0
=

16 1
2
3
2
=

6. Te
paitial deiivatives of f at (x
0
, y
0
) = (1, 3)
aie
f
x
=
x
0

16 x
2
0
y
2
0
=
1

6
,
f
y
=
y
0

16 x
2
0
y
2
0
=
3

6
.
Te equation foi the tangent plane is then
z =

6
1

6
(x 1)
3

6
(y 3)
=
16

6

x

6

3y

6
.
8 4. DERlVATlVES
6. e Two Variable Chain Rule
6.1. e ain rule. Given two functions x = x(t), y = y(t) of one vaiiable, and a
function z = f(x, y) of two vaiiables, then what is the deiivative of the function
g(t) = f(x(t), y(t))?
We can nd a geneial foimula foi g

(t) by using the lineai appioximation ( 4) in the


following way.
To nd g

(t
0
) foi some t
0
, we must compute
g(t
0
+ t) g(t
0
)
t
and let t 0.
lf t incieases by an amount t fiom t
0
to t
0
+ t, then x and y will also change. We
wiite x and y foi the changes in x and y, i.e.
x = x(t
0
+ t) x
0
, y = y(t
0
+ t) y
0
,
wheie x
0
= x(t
0
) and y
0
= y(t
0
). Te iesulting change in g is thus
g = g(t
0
+ t) g(t
0
)
= f
_
x(t
0
+ t), y(t
0
+ t)
_
f
_
x(t
0
), y(t
0
)
_
= f(x
0
+ x, y
0
+ y) f(x
0
, y
0
).
By the lineai appioximation foimula () one then has
f
t
= f
x
(x
0
, y
0
)
x
t
+f
y
(x
0
, y
0
)
y
t
+e
x
x
t
+e
y
x
t
As we let t 0 the quotients x/t and y/t conveige to x

(t
0
) and y

(t
0
), while
the eiiois e
x
and e
y
conveige to zeio, so we get the two-variable ain rule:
(8)
df(x(t), y(t))
dt
= f
x
(x
0
, y
0
) x

(t
0
) +f
y
(x
0
, y
0
) y

(t
0
).
Te chain iule is ofen also wiiuen as
(9)
df
dt
=
f
x
dx
dt
+
f
y
dy
dt
.
Tis foim becomes easy to iemembei if we inteipiet the ist teim as the change in f
caused by the change in x and the second teim as the change in f caused by the change
in y.
ln the way (9) is wiiuen a numbei of details aie swept undei the iug the two deiiva-
tives
dx
dt
and
dy
dt
aie oidinaiy (Math 221) deiivatives of the two functions x(t) and y(t),
the two paitial deiivatives
f
x
and
f
y
aie the paitial deiivatives of f in whi one has
substituted x(t) and y(t). A moie coiiect way of wiiting the equation would be
(0)
df(x(t), y(t))
dt
=
f
x
(x(t), y(t)) x

(t) +
f
y
(x(t), y(t)) y

(t).
Many people nd (9) easiei on the eyes, so that is what we will usually wiite.
. THE TWO VARlABLE CHAlN RULE 9
6.2. e dierence between d and . Compaie (9) with the lineai appioximation
foimula (4) with innitesimal small quantities. Equation (9) is just (4) in which one
has divided both sides by dt. ln contiast to equation (4) which contains the stiange
innitely small quantities dx, dy, df, equation (9) contains the deiivatives
dx
dt
, etc.
which are well-dened.
Note that we have a bieakdown of Leibnizs notation if we ignoie the distinction
between d and , and just cancel dx and x, and also dy and y on the iight then we
end up with
df
dt
=
f

dx
dt
+
f

dy
dt
=
f
dt
+
f
dt
= 2
f
dt
,
which doesnt make a lot of sense. Te moial dont cancel dx against x'
6.3. An example. Suppose x(t) = cos t and y(t) = sin t, so that
#
x(t) = x(t)
#
e
1
+
y(t)
#
e
2
tiaces out the unit ciicle.
How fast does S(t) = 2x(t) + 3y(t) change along this motion?
ln othei woids, what can we say about
dS
dt
`
Te quantity S(t) is the composition of a function of two vaiiables with the functions
x(t) and y(t), i.e. it is the iesult of substituting x(t) and y(t) in the function f(x, y) =
2x + 3y.
Answer 1 without using the chain rule. We can simply compute S(t) = cos t +
sin t and dieientiate
(1)
dS
dt
=
d
dt
_
2 cos t + 3 sin t
_
= 2 sin t + 3 cos t.
Note that we did not use oui new two-vaiiable chain iule heie. Tis answei shows that
the point of the two-vaiiable chain iule is not to compute
d
dt
f(x(t), y(t)) in situations
wheie we have foimulas foi the functions f(x, y), x(t), and y(t). ln such a situation we
can always substitute x(t) and y(t) in the function f(x, y) afei which we get a function
S(t) = f(x(t), y(t)) of one vaiiable. We leained how to dieientiate those in oui ist
calculus couise.
Answer 2 using the chain rule. Te quantity we want to dieientiate is
S(t) = f
_
x(t), y(t)
_
,
wheie
f(x, y) = 2x + 3y, and x(t) = cos t, y(t) = sin t.
Te chain iule tells us that
(2)
dS
dt
=
f
x
dx
dt
+
f
y
dy
dt
.
Heie the ist teim stands foi the change in S that is caused by the change in x. To
compute it we ist nd
f
x
=
{2x + 3y}
x
= 2,
so that
f
x
dx
dt
= 2
dx
dt
.
Similaily, the second teim in (2) iepiesents the change in S(t) due to the fact that y is
changing
f
y
=
{2x + 3y}
y
= 3
f
y
dy
dt
= 3
dy
dt
.
0 4. DERlVATlVES
To get the iate of change of S we add both the x and y contiibutions to this iate of change,
which leads us to
(3)
dS
dt
= 2
dx
dt
+ 3
dy
dt
.
So fai we have not used what we know about x(t) and y(t). Tis expiession we have just
deiived foi dS/dt is tiue no mauei which x(t), y(t) we aie given. ln oui case we have
x(t) = cos t
dx
dt
= sin t,
y(t) = sin t
dy
dt
= + cos t.
Substitute this in (3)
dS
dt
= 2 sin t + 3 cos t,
as befoie.
e moral: ln this example the answei using the chain iule was longei, much moie
veibose, and peihaps moie complicated than the stiaightfoiwaid computation that led to
oui ist answei (1). lndeed, if the deiivative of S is all we want then oui ist computa-
tion is the most ecient way of geuing dS/dt. Howevei, the computation using the chain
iule did give us some useful inteimediate iesults, such as the geneial expiession (3) foi
dS/dt. Tis expiession iemains valid if we change the path (x(t), y(t)) and can theie-
foie be useful in situations wheie, foi example, we aie allowed to choose the path and we
would like to choose a path foi which dS/dt has some piesciibed value (e.g. suppose we
want to keep S constant, how do we choose the path`)
6.4. Another example. Suppose the tempeiatuie at the point (x, y) in the plane is
given by T(x, y), and suppose that an ant is walking along the paiametiized cuive
x(t) = Rcos t, y(t) = Rsin t.
Tus the ant is walking on a ciicle with iadius R, and with angulai velocity .
How fast is the temperature of the ant changing?
i.e. compute
dT
dt
.
Heie we aie not given an explicit foimula foi the function T(x, y), so we cannot substitute
x(t) and y(t) in T and dieientiate using only oui ist semestei calculus skills. Te
appioach in Answei 1 of oui pievious example does not apply heie, we must use the
chain iule.
ln .1 we have seen seveial equivalent ways of wiiting the chain iule. Let us look at
two of these and considei the meaning of the teims that aiise.
Te shoit foim (9) of the chain iule tells us that
dT
dt
=
T
x
dx
dt
+
T
y
dy
dt
.
Te T on the lef stands foi T(x(t), y(t)), which we can inteipiet as the tempeiatuie at
the point (x(t), y(t)). Tat point is the location of the ant at time t, so the T on the lef
is the tempeiatuie the ant feels at time t. Tis is a function of t. ln mathematical teims
it is the iesult of substituting (composing) the functions x(t) and y(t) in the function
T = T(x, y).
Te two Ts on the iight appeai in paitial deiivatives. Heie
T
x
stands foi the paitial
deiivative of the function T = T(x, y) with iespect to the vaiiable x. One can compute
this without knowing the ants path (x(t), y(t)). Similaily,
T
y
is the paitial deiivative of
. PROBLEMS 1
70
68
66
64
T=62F
60
58
56
54
48
52
50
72
74
Figure 6. Ant walking in a region of varying temperature.
T with iespect to y. Te paitial deiivatives
T
x
and
T
y
themselves aie again functions
of x and y. Aer computing these paitials they aie meant to be evaluated at the point
(x(t), y(t)).
Tis leads us to the moie veibose veision (0) of the chain iule, which tells us
dT(x(t), y(t))
dt
=
T
x
(x(t), y(t)) x

(t) +
T
y
(x(t), y(t)) y

(t).
At this point the only additional infoimation we have is about the ants motion, namely,
x(t) = Rcos t and y(t) = sin t. We can compute the deiivatives of x(t) and y(t),
which gives us the velocity of the ant in the x and y diiections
x

(t) = Rsin t, y

(t) = Rcos t.
lf we substitute eveiything we know in the chain iule we nd that the iate at which the
ants tempeiatuie changes is
dT
dt
=
T
x
(Rcos t, Rsin t) Rsin t +
T
y
(Rcos t, Rsin t) Rcos t.
To make the equation moie ieadable one can leave out the (Rcos t, Rsin t), which
iesults in
dT
dt
= Rsin t
T
x
+Rcos t
T
y
.
Te disadvantage of this shoitei veision is that the ieadei has to guie out wheie we
intended to evaluate the two paitial deiivatives
T
x
and
T
y
.
7. Problems
1. Find the linear approximation to f(x, y)
at the point (a, b) in the following cases:
(a) f(x, y) = xy
2
, (a, b) = (3, 1).
(b) f(x, y) = x/y
2
, (a, b) = (3, 1).
(c) f(x, y) = sin x + cos y, (a, b) = (, ).

(d) f(x, y) = xy/(x +y), (a, b) = (3, 1).


2. Find an equation for the plane tangent
to the graph of f(x, y) = sin(xy) at
(, 1/2, 1).
3. Find an equation for the plane tangent to
the graph of f(x, y) = x
2
+y
3
at (3, 1, 10).

2 4. DERlVATlVES
4. Find an equation for the plane tangent
to the graph of f(x, y) = xln(xy) at
(2, 1/2, 0).
5. (a) Find an equation for the plane tangent
to the surface defined by 2x
2
+3y
2
z
2
= 4
at (1, 1, 1). (Hint: first write the surface as
a graph z = f(x, y)).
(b) The same question at the point
(1, 1, +1).
6. (a) Suppose you have computed the
two partial derivatives of a function z =
f(x0, y0), and you found fx(x0, y0) = A
and fy(x0, y0) = B. Find a normal vec-
tor to the tangent plane of the graph of z =
f(x, y) at (x0, y0, z0).
(Hint: If you know the equation for a
plane, then how do you find a normal vec-
tor to this plane?)
(b) Find an equation in vector form for the
tangent plane to x
2
+ 4y
2
= 2z at (2, 1, 4).
Also find an equation for the normal line to
the graph at (2, 1, 4). (The normal line to the
graph of a function at some point P, is the
line through P that is perpendicular to the
tangent plane to the graph at P.)
7. Imagine a dierentiable function,
f(x, y). Make a good drawing of the func-
tion f and show how fx(a, b) and fy(a, b)
are the slopes of two lines which are tangent
to the graph at (a, b). Indicate clearly which
two lines you mean, and describe how they
are defined.
(Cant think of a nice graph? Take some-
thing like the boom drawing in Figure 4.)

8. Let f be as in problem 7.4. Use linear ap-


proximation to approximate f(1.98, 0.4) by
hand. Compare your answer with the actual
value of f(1.98, 0.4) (youll need a calcula-
tor).
9. (a) The tangent plane to the saddle sur-
face z = xy at the origin intersects the
graph of the saddle surface in two lines.
Which lines are they?
(b) Consider the tangent plane to the saddle
surface at x = 2, y = 1 that was computed
in 5.3. Let (x, y, z) be a point on the saddle
surface, and let (x, y, z) be the point on the
tangent plane with the same x and y coor-
dinates. What is the dierence in heights of
these two points?
(c) Show that the saddle surface and its tan-
gent plane intersect when x = 2 or y = 1.
10. (a) Find an equation for the tangent)
plane to the graph of f(x, y) = xy at the
point (a, b, ab). Here a and b are constants
which will appear in your answer.
(b) Showthat the intersection of the tangent
plane and the graph consists of two straight
lines.
8. Gradients
8.1. e gradient vector of a function. Te iight hand side in the chain iule (8)
can be wiiuen as a dot-pioduct of two vectois, namely
df
dt
= f
x
(x
0
, y
0
) x

(t
0
) +f
y
(x
0
, y
0
) y

(t
0
) (4)
=
_
f
x
(x
0
, y
0
)
f
y
(x
0
, y
0
)
_

_
x

(t
0
)
y

(t
0
)
_
Tis tuins out to be so useful that the vectoi containing the deiivatives of f has been
given a name. lt is called the gradient of f, and it is wiiuen as
()
#
f(x, y)
def
=
_
f
x
(x, y)
f
y
(x, y)
_
Te symbol
#
is pionounced nabla.
Te chain iule, wiiuen in vectoi foim, looks like this
()
df(
#
x(t))
dt
=
#
f(x(t))
#
x

(t)
8. GRADlENTS 3
#
f(P)
f = 0.0
-0.6
-0.3
0.3
0.6
A B
C
D
P
Figure 7. The gradient as direction of fastest increase: if we are at a point P, and we are allowed
to jump to any point at a given fixed distance fromP, and if we only know
#
f(P), then the linear
approximation formula tells us that
to maximize f we follow the gradient (choose A);
to minimize f we go in the direction opposite to
#
f(P) (choose D);
to keep f fixed we move perpendicular to the gradient (choose B or C).
Te lineai appioximation foimula (0) can also be iewiiuen moie compactly using the
giadient vectoi
() f(
#
x
0
+
#
x) f(
#
x
0
) +
#
f(
#
x
0
)
#
x.
8.2. e gradient as the direction of greatest increase for a function f. When
we apply the foimula
(8)
#
a
#
b =
#
a
#
b cos (
#
a,
#
b )
foi the dot pioduct to the vectoi foim () of the lineai appioximation equation, we nd
a veiy useful inteipietation of the giadient. lf we aie at a point with position vectoi
#
x
0
(P in guie ) and we aie allowed to make a small step
#
x in any diiection we like, but
of piesciibed length, then which way should we go if we want to inciease f as much as
possible` And wheie should we go if, instead, we want to deciease f as much as possible`
What if we want to keep f the same`
Fiom () we see that the change in f is (appioximately) given by
f
def
= f(
#
x +
#
x) f(
#
x)
()

#
f
#
x
(8)
=
#
f
#
x cos
wheie is the angle between the giadient
#
f and the vectoi
#
x which iepiesents the
step we take. ln this foimula the lengths
#
f and
#
x aie xed, and the angle is the
only thing we can change. Teiefoie the laigest change in f iesults if cos = +1, the
smallest when cos = 1, and no change will iesult if cos = 0. So we conclude
To inciease f as much as possible choose
#
x in the diiection of the giadient
#
f,
To deciease f as much as possible choose
#
x in the diiection opposite to the
giadient
#
f, i.e. in the diiection of
#
f,
To keep f constant choose
#
x peipendiculai to the giadient.
4 4. DERlVATlVES
Tis is sometimes summaiized by saying that the gradient
#
f points in the direction
of fastest increase for the function f.
8.3. e gradient is perpendicular to the level curve. Suppose that foi some func-
tion z = f(x, y) the level set at level C is a cuive, and suppose that we have a paiametiic
iepiesentation
#
x(t) =
_
x(t)
y(t)
_
of this cuive. Tis means that x(t) and y(t) satisfy
f(x(t), y(t)) = C.
By the chain iule we then get
0 =
df(
#
x(t))
dt
=
#
f(
#
x(t))
#
x

(t),
which tells us that the tangent vectoi
#
x

(t) to the level set is peipendiculai to the giadient


#
f(
#
x(t)) of the function. Teiefoie,
if
#
f(x
0
, y
0
) =
#
0, then
#
f(x
0
, y
0
) is a normal vector to the tangent
to the level curve of f at (x
0
, y
0
).
We now have the necessaiy ingiedients to wiite the equation foi the tangent, namely we
know a point (x
0
, y
0
) on the line, and we know a noimal vectoi to the line (the giadient).
Tus the equation foi the tangent is
#
f(
#
x
0
) (
#
x
#
x
0
) = 0,
oi, equivalently,
f
x
(x
0
, y
0
) (x x
0
) +
f
y
(x
0
, y
0
) (y y
0
) = 0.
8.4. e tangent to the parabola y = x
2
, again. Te veiy ist example anyone sees
in theii ist calculus couise must suiely be the computation of the tangent to the paiabola
y = x
2
at the point (x, y) = (1, 1). We know the answei it is a line with slope 2, thiough
the point (1, 1).
We can inteipiet the paiabola as the zeio set of the function of two vaiiables given by
f(x, y) = y x
2
, and theiefoie we should be able to nd the same tangent at (1, 1) by
computing the giadient of f. Te computation goes like this
f(x, y) = y x
2

#
f(x, y) =
_
f
x
f
y
_
=
_
2x
y
_
.
At (x, y) = (1, 1) we have
#
f(1, 1) =
_
2
1
_
.
Tis vectoi is peipendiculai to the tangent to its zeio set. lf we let
#
x
0
= (
1
1
) be the
position vectoi of oui point on the paiabola, then the equation foi the tangent to the
paiabola at this point is
#
n (
#
x
#
x
0
) = 0,
i.e.
_
2
1
_

_
x 1
y 1
_
= 0.
Simplifying this we get
2 (x 1) + 1 (y 1) = 0, and thus y = 2x 1.
Tis is the same line that we found in oui ist calculus couise.
8. GRADlENTS
8.5. Example: the tangent to the zero set of x
2
y
2
+y
3
. Considei the zeio set of
the function
f(x, y) = x
2
y
2
+y
3
.
Te iesulting cuive is not as familiai as the paiabola fiom the pievious example, and
diawing the cuive takes some eoit'.
We will not tiy to diaw the whole zeio set in this example, but instead we will see
what happens when we tiy to nd the tangent to the zeio set at two dieient points on
the zeio set, namely, at (0, 1) and at the oiigin.
e tangent at (0, 1). To nd the tangent at any point on the zeio set of f we use that
the noimal to the tangent is given by the giadient of f
#
f =
_
f
x
f
y
_
=
_
2x
2y + 3y
2
_
.
Te noimal to the tangent at the point (0, 1) is theiefoie
#
n =
#
f(0, 1) =
_
0
2 + 3
_
=
_
0
1
_
.
ln othei woids, the noimal to the tangent at (0, 1) is the veitical unit vectoi
#
e
2
, and
theiefoie the tangent is a hoiizontal line thiough (0, 1). lts equation is y = 1. We could
also nd this equation by woiking out the geneial equation
#
n (
#
x
#
x
0
) = 0 foi a line
with a given noimal and point. Heie we have
#
n =
#
f(0, 1) =
_
0
1
_
,
#
x
0
=
_
0
1
_
,
so the equation foi the tangent is
_
0
1
_

_
x 0
y 1
_
= 0,
which simplies to
y 1 = 0.
e tangent at the origin. When we iepeat the pievious calculation at (x
0
, y
0
) =
(0, 0) we iun into pioblems. Tese pioblems begin when we compute the giadient
#
f
at the oiigin
#
f(0, 0) =
_
2x
2y + 3y
2
_
x=0,y=0
=
_
0
0
_
.
Te giadient at the oiigin tuins out to be the zeio vectoi. Tis is pioblematic because
the zeio vectoi has no diiection, and thus is not peipendiculai to any paiticulai line. We
cannot nd the tangent at the origin!
To see what is going on one has to take a closei look at the cuive neai the oiigin see
guie 8. lt tuins out that neai the oiigin the zeio set of f consists of two smooth cuives
that cioss each othei. Te giadient has to be peipendiculai to both of these cuives, and
the only vectoi that achieves this is the zeio vectoi. Note also that theie is no single line
'One could stait by solving the equation foi x, which leads to x = y

1 y. Tis shows that y 1 on


the cuive. Giaphing x = y

1 y using oui 1st semestei calculus skills then gives us half the cuive, the othei
half is given by its ieection in the y-axis, i.e. x = y

1 y.
One way to see this is to solve x
2
y
2
+y
3
= 0 foi x, which gives x = y

1 y. Neai the oiigin y is


veiy small, so we can appioximate

1 y

1 = 1. Te zeio set neai the oiigin is theiefoie appioximately


desciibed by x = y, i.e. two ciossing lines.
4. DERlVATlVES
that is tangent to the zeio set at the oiigin. lf we had seen the diawing ahead of time then
we would not have expected to nd a tangent to the zeio set of f at the oiigin.
f(x, y) = 0
#
f(0, 1)
(a, b)
#
f(a, b)
Figure 8. The zero set of the function f(x, y) = x
2
y
2
+y
3
, and its gradient at various points
on this zero set. Since the gradient is always perpendicular to the level set of a function, a drawing
of the zero set tells us the direction of the gradient. However, the drawing does not say anything
about the length of the gradient.
9. e ain rule and the gradient of a function of three variables
9.1. e gradient, etc. So fai we have only looked at the giadient of a function of
two vaiiables. But foi a function of thiee vaiiables theie is a veiy similai denition, and
the facts we have discoveied have neaily identical counteipaits.
lf u = f(x, y, z) is a function of thiee vaiiables, then its giadient is dened to be the
vectoi
#
f(x, y, z) =
_
_
f
x
(x, y, z)
f
y
(x, y, z)
f
z
(x, y, z)
_
_
.
Te ain rule in this context says that, if x = x(t), y = y(t), and z = z(t) aie functions
of one vaiiable, then the deiivative of the function we get by substituting x(t), y(t), z(t)
in f is given by any of the following thiee equivalent foimulas
df(x(t), y(t), z(t))
dt
= f
x
(x(t), y(t), z(t)) x

(t) +f
y
(x(t), y(t), z(t)) y

(t) (9)
+f
z
(x(t), y(t), z(t)) z

(t)
=
f
x
dx
dt
+
f
y
dy
dt
+
f
y
dy
dt
=
#
f(
#
x(t))
#
x

(t), wheie
#
x(t) =
_
_
x(t)
y(t)
z(t)
_
_
.
Te lineai appioximation foimula foi the function f at some point (x
0
, y
0
, z
0
), which
gives us an appioximation of the amount by which f incieases if we go fiom (x
0
, y
0
, z
0
)
to (x, y, z) = (x
0
+ x, y
0
+ y, z
0
+ z), is as follows
f = f(x, y, z) f(x
0
, y
0
, z
0
) (80)

f
x
x +
f
y
y +
f
z
z,
9. THE CHAlN RULE AND THE GRADlENT OF A FUNCTlON OF THREE VARlABLES
in which the paitial deiivatives aie to be evaluated at (x
0
, y
0
, z
0
). Compaie this with the
two vaiiable veision (9). ln vectoi foim we have
(81) f = f(
#
x
0
+
#
x) f(
#
x
0
)
#
f(
#
x
0
)
#
x,
wheie
#
x
0
=
_
_
x
0
y
0
z
0
_
_
,
#
x =
_
_
x
y
x
_
_
.
Tis is the same foimula as in the two-vaiiable case, wheie we had (). Te discussion
about diiection of fastest inciease applies to the thiee vaiiable case without change.
Tus, if we aie at a point
#
x
0
, and we aie allowed to change oui position by a small
vectoi
#
x of a piesciibed length, then we should choose
#
x in the diiection of the
giadient
#
f(
#
x) if we want to inciease f as much as possible, we should choose
#
x in
the diiection of
#
f(
#
x) if we want to decrease f as much as possible, and we should
choose
#
x peipendiculai to
#
f(
#
x) if we want to keep f constant.
9.2. Tangent plane to a level set. lf t = f(x, y, z) is a function of thiee vaiiables
then it is haid to visualize its giaph, since this involves diawing four mutually peipendic-
ulai axes, something we, thiee dimensional cieatuies, cannot do. Howevei, we can tiy to
visualize the level sets of the function. Te level set at level C consists, by denition, of all
points in thiee dimensional space whose cooidinates satisfy the equation f(x, y, z) = C.
Foi instance, the unit spheie is given by the equation x
2
+ y
2
+ z
2
= 1, so it is the
level set at level 1 of the function f(x, y, z) = x
2
+y
2
+z
2
. Te spheie with iadius R is
the level set of the same function f at level R
2
.
Considei a function of thiee vaiiables, and let (x
0
, y
0
, z
0
) be some point on the level set
at level C (thus f(x
0
, y
0
, z
0
) = C.) Te equation foi the level set itself is f(x, y, z) = C,
and since (x
0
, y
0
, z
0
) satises this equation we can wiite the equation foi the level set as
f(x, y, z) f(x
0
, y
0
, z
0
) = 0.
Neai the point (x
0
, y
0
, z
0
) we can use the lineai appioximation of f to appioximate the
equation foi the level set of f. We have
f(x, y, z) f(x
0
, y
0
, z
0
)
f
x
(x x
0
) +
f
y
(y y
0
) +
f
z
(z z
0
),
wheie, as in (80), the paitial deiivatives aie to be computed at the given point (x
0
, y
0
, z
0
).
Tey aie, in paiticulai, constants (they depend on (x
0
, y
0
, z
0
) but not on (x, y, z).)
#
f
f(x, y, z) = C
8 4. DERlVATlVES
Tus we see that neai any paiticulai point on the level set of a function we can ap-
pioximate the equation foi the level set by
(82)
f
x
(x x
0
) +
f
y
(y y
0
) +
f
z
(z z
0
) = 0.
lf at least one of the paitial deiivatives at (x
0
, y
0
, z
0
) is non zeio, then this is the equation
of a plane. We call this plane the tangent plane to the level set.
ln vectoi foim the equation foi the tangent plane to a level set of f at a point with
position vectoi
#
x
0
can be wiiuen as
(83)
#
f(
#
x
0
) (
#
x
#
x
0
) = 0.
Fiom this equation we see that, just as in the case (8.3) of level cuives of a function of
two vaiiables, the gradient
#
f(
#
x
0
) is perpendicular to the tangent plane of the
level set of the function f at the point
#
x
0
.
9.3. Example: tangent plane to a sphere revisited. ln the example in .4 we found
the tangent plane to the spheie at the point (1, 3,

6), wheie the spheie had iadius 4, and


was centeied at the oiigin. Teie we iepiesented the top half of the spheie as the giaph
of a function. We will now iedo this calculation by iepiesenting the spheie as the level
set of some othei function.
By Pythagoias the distance d fiom a point (x, y, z) to the oiigin satises
d
2
= x
2
+y
2
+z
2
.
Te spheie with iadius 4 and centei at the oiigin theiefoie consists of all points (x, y, z)
that satisfy
x
2
+y
2
+z
2
= 4
2
= 16.
ln othei woids, it is the level set at level C = 16 of the function
f(x, y, z) = x
2
+y
2
+z
2
.
To nd an equation foi the tangent plane thiough the point (1, 3,

6) we need two ingie-


dients a point on the plane and a noimal vectoi to the plane. (See Chaptei l, 11.2.) We
alieady have a point on the plane, namely oui point (1, 3,

6), and the noimal is given


by the giadient of the function f whose level set is the spheie. Tis giadient is easy to
compute. Since f(x, y, z) = x
2
+y
2
+z
2
, we have
f
x
= 2x,
f
y
= 2y,
f
z
= 2z,
and thus
#
f(1, 3,

6) =
_
_
2x
2y
2z
_
_
(x,y,z)=(1,3,

6)
=
_
_
2
6
2

6
_
_
.
Te equation foi the tangent plane is
#
n (
#
x
#
x
0
) = 0, wheie the noimal
#
n to the
tangent plane is the giadient
#
f evaluated at oui given point
#
x
0
. So, the tangent plane
is given by
#
f(1, 3,

6) (
#
x
#
x
0
) = 0,
which we can wiite as
_
_
2
6
2

6
_
_

_
_
x 1
y 3
z

6
_
_
= 0,
10. lMPLlClT FUNCTlONS 9
i.e.
2(x 1) + 6(y 3) + 2

6(z

6) = 0.
Afei some cleaning up we get
x + 3y +

6z = 16.
Tis is the same answei we got in .4.
9.4. Example. Find the linear approximation of F(x, y, z) = e
y
(x z)
2
and tan-
gent plane to its level set at x = 1, y = 2, z = 5
Solution: At the given values of x, y, z on has F(1, 2, 5) = e
2
(1 5)
2
= 16/e
2
. Te
paitial deiivatives of F aie
F
x
= 2(x z)e
y
, F
y
= e
y
(x z)
2
, F
z
= 2(x z)e
y
,
which at (x, y, z) = (1, 2, 5) ieduces to F
x
= 8/e
2
, F
y
= 16/e
2
and F
z
= +8/e
2
. lf
(x, y, z) is close to (1, 2, 5), then the lineai appioximation foimula tells us that
F(x, y, z) F(1, 2, 5)
8
e
2
(x 1)
16
e
2
(y 2) +
8
e
2
(z 5)
oi, in x notation,
By definition:
x = x 1
y = y 2
z = z 5
F(1 + x, 2 + y, 5 + z) F(1, 2, 5)
8
e
2
x
16
e
2
y +
8
e
2
z.
Te equation foi the tangent plane to the level set of F at the point (1, 2, 5) is theiefoie

8
e
2
(x 1)
16
e
2
(y 2) +
8
e
2
(z 5) = 0,
oi, afei cancelling e
2
s and 8s (x 1) +2(y 2) (z 5) = 0. Fuithei simplication
shows that the equation foi the tangent plane is
x + 2y z = 0.
10. Implicit Functions
ln ist semestei calculus we leained a pioceduie foi nding deiivatives of implicitly
dened functions. lf some function y = f(x) was not given by an explicit foimula, but
iathei by an implicit equation
(84) F(x, y) = 0
then theie was a way to nd the deiivative of y = f(x) fiomthe above equation only. But
theie was no foimula foi f

(x). Te ieason is that the foimula foi the deiivative f

(x)
involves the paitial deiivatives of F.
ln this section we ieviewimplicit dieientiation again. Te following theoiemis about
the zeio set of the function F. One usually thinks of the zeio set of a function of two
vaiiables as a cuive (an equation denes a cuive) but this is not always so. Te theoiem
below gives us a way to nd out if the zeio set is ieally a cuive, at least neai any given
point on the zeio set which we happen to know.
0 4. DERlVATlVES
B
A
C
D
y=f(x)
x=g(y)
F
(
x
,
y
)

=

0
Figure 9. The Implicit Function Theorem. The zero set of a function F(x, y) does not have
to be the graph of a function, but if at some point (A) on the zero set we have Fy = 0, then, near
that point A, the zero set is the graph of a function y = f(x). If Fx = 0 at some point (B), then
near B the zero set is also the graph of a function, provided we let x be a function of y: x = g(y).
Exceptional points: At some points, like C and D in this figure, the level set of F cannot be
represented as the graph of a function y = f(x), nor can it be represented as a graph of the type
x = g(y). At such points the Implicit Function Theorem implies that both Fx = 0 and Fy = 0.
10.1. e Implicit Function eorem. Let F(x, y) be a function dened on some
plane domain with continuous partial derivatives in that domain, and suppose that a point
(x
0
, y
0
) in the zero set of F is given.
If
F
y
(x
0
, y
0
) = 0 then there is a small rectangle centered at (x
0
, y
0
) such that within
this rectangle the zero set of F is the graph of a function y = f(x). e derivative of this
function is
(8) f

(x) =
dy
dx
=
F
x
(x, f(x))
F
y
(x, f(x))
.
If
F
x
(x
0
, y
0
) = 0 then there is a small rectangle centered at (x
0
, y
0
) such that within
this rectangle the zero set of F is the graph of a function x = g(y). e derivative of this
function is
(8) g

(y) =
dx
dy
=
F
y
(g(y), y)
F
x
(g(y), y)
.
A pioof may be given in class, time peimiuing.
Teie is no need to memoiize the foimulas (8) and (8). We can get them by using the
method of implicit dieientiation fiom math 221. Foi instance, suppose that the giaph of
the function y = f(x) gives you a piece of the zeio set of F. Tis means that
F(x, f(x)) = 0 foi all x.
10. lMPLlClT FUNCTlONS 1
Dieientiating both sides of this equation leads us via the chain iule,
dF(x, f(x))
dx
=
F
x
(x, f(x))
dx
dx
+
F
y
(x, f(x))
df(x)
dx
,
to
(8) 0 =
dF(x, f(x))
dx
= F
x
(x, f(x)) +F
y
(x, f(x))f

(x).
Solve this foi f

(x) and we get


f

(x) =
dy
dx
=
F
x
(x, f(x))
F
y
(x, f(x))
,
which is what the theoiem claims.
10.2. e Implicit Function eorem with more variables. Teie aie many vaiia-
tions and extensions of Teoiem10.1. Te simplest is to considei the level set of a function
of thiee iathei than two vaiiables. Suppose F is a function of thiee vaiiables, with con-
tinuous paitial deiivatives, and considei the set of points dened by the equation
F(x, y, z) = C.
Tis is the level set of F at level C.
lf
F
y
(x
0
, y
0
, z
0
) = 0,
then neai (x
0
, y
0
, z
0
) the level set of F is the giaph of a function y = g(x, z), meaning
that the function y = g(x, z) satises
G(x, g(x, z), z) = 0.
Hence we can nd the paitial deiivatives of this function by implicit dieientiation. Te
iesult is
(88)
y
x
= g
x
(x, z) =
F
x
(x, y, z)
F
y
(x, y, z)
,
y
z
= g
z
(x, z) =
F
z
(x, y, z)
F
y
(x, y, z)
,
wheie y = g(x, z).
10.3. Example e saddle surface again. Te saddle suiface is the giaph of the
function z = xy, which we can think of as the zeio set of the function
F(x, y, z) = z xy.
Te point (2, 3, 6) lies on the saddle suiface, and at this point the paitial deiivatives of F
aie
F
x
=
(z xy)
x
= y = 3, F
y
=
(z xy)
y
= x = 2, F
z
=
(z xy)
z
= 1.
Since F
x
(2, 3, 6) = y = 3 is non zeio, the lmplicit Function Teoiem tells us that neai
this point the zeio set of F is the giaph of a function x = g(y, z). Solving F = 0 foi x
we see that his function is in fact
x = g(y, z) =
z
y
.
Te paitial deiivatives of g aie easy to compute in this example, but even if we couldnt
nd them diiectly, the lmplicit Function Teoiem would tell us that
g
y
(3, 6) =
F
y
(2, 3, 6)
F
x
(2, 3, 6)
=
2
3
, g
z
(3, 6) =
F
z
(2, 3, 6)
F
x
(2, 3, 6)
=
1
3
.
2 4. DERlVATlVES
Problems
1. Compute the gradient of each function in
Problem 3.2 of 3.
2. Show that for any two dierentiable
functions f and g one has
#
(f g) =
#
f
#
g,
#
(fg) = f
#
g +g
#
f,
#

(
f
g
)
=
g
#
f f
#
g
g
2
.
In other words the sum-, product- and quo-
tient rules for dierentiation also apply to
the gradient.
3. (a) Draw the level sets of the function
f(x, y) = x
2
+ 4y
2
at levels 0, 4, 16.
(b) Find the points on the level set f(x, y) =
4 where the gradient is parallel to the vector
(
1
1
). What can you say about the tangent
line to the level set at those points? Draw
the gradient vectors, and the tangent lines
at the points you just found.
Hint: two non-zero vectors
#
v and
#
w
are parallel if there is a number s such that
#
v = s
#
w.
(c) Repeat the same two problems for the
function g(x, y) = 4xy
2
.
4. (a) Draw the zero set of the function
f(x, y, z) = x
2
+y
2
2z.
(b) Find all points on the zero set of the func-
tion f where the gradient is parallel to the
vector
#
v =
(
1
1
2
)
.
5. A bug is crawling on the surface of a hot
plate, the temperature of which at the point
x units to the right of the lower le corner
and y units up from the lower le corner is
given by T(x, y) = 100 x
2
3y
3
.
(a) If the bug is at the point (2, 1), in what di-
rection should it move to cool o the fastest?

(b) If the bug is at the point (1, 3), in what


direction should it move in order to maintain
its temperature?
6. The level sets of a function z = f(x, y)
are oen curves. Must they always be
curves? Could the zero set of a function
be a solid square (e.g. all points (x, y) with
0 x 1 and 0 y 1)?
7. The caption of Figure 8 says that one can
only see the direction, but not the length of
the gradient
#
f of a function, from just one
of its level sets. It is however possible to see
where the gradient is larger from a drawing
of several level sets. We can read this infor-
mation from the way in which level sets are
more bunched together in some regions than
in others.
f
=
0
.
3
f
=
0
.
2
f
=
0
.
1
f
=
0
.
0
f
=
-
0
.
1
The picture above shows some level sets
of a function. On the boom le the
level sets are further apart, on the top right
they are more bunched together. Where is
the gradient the larger, i.e. where is
#
f
larger: boom-le, or top-right?
8. Have a look at Figure 8. Assume the func-
tion dierentiable at the origin.
(a) What can you say about the gradient
#
f
at the origin?
(b) Where is the function positive and where
is it negative (assume that the whole zero set
is drawn).
9. This problem asks you to think about the
Implicit Function Theorem 10.1
Consider the unit circle C with equation
x
2
+y
2
= 1.
The unit circle C is a level set of the function
F(x, y) = x
2
+y
2
.
(a) Where on C is Fy = 0? Near which
points P on C can one represent C as a graph
of the form y = f(x)?
(b) Near which points P on C can one rep-
resent C as a graph of the form x = g(y)?
10. Here is the zero set of a function z =
f(x, y) (in bold). The function is only zero
11. THE CHAlN RULE WlTH MORE lNDEPENDENT VARlABLES, COORDlNATE TRANSFORMATlONS 3
on the bold curve, it is nonzero everywhere
else.
A
B
f(x, y) = 0
f(x, y)
=
-0.1

(a) One of the two other curves above is the


level set f(x, y) = 0.1. Which one is it, A
or B? As always, explain your answer.
(b) Draw a possible level set f(x, y) =
+0.1.
(c) Draw possible gradients on the zero set
(similar to Figure 8).
11. Here is the zero set of a dierentiable
function z = f(x, y).
A
B
f(x,y)=0
Explain why the Implicit Function Theorem
(10.1) implies that
#
f =
#
0 at the two
points A and B.
12. (a) Compute the gradient of the distance
to the square function f fromproblems 5.13
and 3.7.
(b) How much is |
#
f|?
(c) Make a drawing of the level sets of f, and
the gradient
#
f.
13. Let f(x, y) = ln(2 + 2x +e
y
).
(a) Compute the gradient of f at the point
(x0, y0) with position vector
#
x0 = (
1
0
).
(b) You are allowed to choose a point at a
distance 0.01 from the point (1, 0). Where
would you choose the new point if you want
f to be as large as possible? (Hint: review
the linear approximation formula and sub-
sequent discussion about the gradient as di-
rection of greatest increase in 8.2)
(c) Is your answer to the previous the exact
answer, or only an approximation? I.e., could
someone else find a point at distance 0.01
from(1, 0) at which f has a (slightly) higher
value than at the point you found?
(d) The level set C of f through the point
(1, 0) happens to be the graph of a function
y = g(x). Find that function.
(e) Find a normal vector to the tangent line
to C at the point (1, 0). Find an equation for
the tangent line to C at (1, 0).
(f) How much is g(1)? Find two dierent
ways to compute g

(1) based on the work


you have done so far.
14. Let (a, b, c) be a point on the sphere with
radius R centered at the origin. Find an
equation for the tangent plane to the sphere
at (a, b, c). Simplify your answer as much as
possible (a, b, c, and R will show up in your
answer of course.)
11. e Chain Rule with more Independent Variables;
Coordinate Transformations
Te chain iule we have seen so fai tells us how to dieientiate expiessions of the foim
f(x(t), y(t)). Such expiessions aie the iesult of substituting two functions x(t), y(t) of
one vaiiable t in one function of two vaiiables z = f(x, y). What do we do if the functions
x, y that get substituted in f(x, y) depend on not one, but two (oi moie) vaiiables` Te
answei is easy we do exactly the same.
Foi instance, suppose we want to substitute x = x(u, v) and y = y(u, v) in a function
z = f(x, y), iesulting in a function F(u, v) = f(x(u, v), y(u, v)), and suppose we want
nd the paitial deiivatives of F with iespect to u. To compute this we keep v xed and
iegaid u as the vaiiable then x(u, v) and y(u, v) aie functions of one vaiiable u and we
4 4. DERlVATlVES
apply the chain iule we alieady know. Tis leads to
F
u
=
f
x
x
u
+
f
y
y
u
Te only dieience with (9) is that we have wiiuen the deiivatives of x and y as paitial
deiivatives. We do this to indicate that in computing this deiivative we momentaiily
considei x as a function of u, but latei we may want to vaiy v again.
Te same consideiations lead to the paitial deiivative of F with iespect to v
F
v
=
f
x
x
v
+
f
y
y
v
.
11.1. An example without context. Suppose f is some function of two vaiiables
and we want to nd the paitial deiivatives of
g(u, v, w) = f(2uv, u
2
+w
2
).
By this we mean that g is the iesult of substituting x = 2uv and y = u
2
+w
2
in f. Note
that g is a function of thiee vaiiables, and f is a function of two vaiiables.
Te chain iule tells us that the deiivatives of g aie
g
u
=
f
x
x
u
+
f
y
y
u
= 2v
f
x
+ 2u
f
y
g
v
=
f
x
x
v
+
f
y
y
v
= 2u
f
x
g
w
=
f
x
x
w
+
f
y
y
w
= 2w
f
y
11.2. Example: a rotated coordinate system. We aie used to specifying the location
of points in the plane by giving theii x and y cooidinates, but sometimes it is beuei to
use dieient cooidinates. Foi instance, two people A and B could have chosen the same
oiigin, but theii axes could be iotated with iespect to each othei. See Figuie 10. lf As
cooidinates aie called x, y and Bs cooidinates aie X, Y then it should be possible to nd
As cooidinates of a point if we know what cooidinates B assigns to this point given
Figure 10. Aer choosing dierent x and y axes, A and B will assign dierent x, y coordinates to
the same point in the plane. Equations (89) give the relation between these two sets of coordinates.
12. PROBLEMS
X, Y what aie x, y` Te answei to this question is
(89)
_
x = X cos Y sin ,
y = X sin +Y cos .
Suppose both A and B aie measuiing the tempeiatuie T at vaiious points in the plane.
A piedicts the tempeiatuie at vaiious points in the plane he says that at the point with
cooidinates (x, y) the tempeiatuie will be T(x, y). ln fact he has also found the paitial
deiivatives
T
x
and
T
y
.
Equipped with the X, Y x, y conveision (89) B can now take As foimula foi the
tempeiatuie and expiess it in teims of hei own X, Y cooidinates. lf we wiite T
A
(x, y)
foi the tempeiatuie at the point whose A-cooidinates aie (x, y) and T
B
(X, Y ) foi the
tempeiatuie at the point whose B-cooidinates aie (X, Y ), then we have
T
B
(X,Y ) = T
A
(x, y)
= T
A
(X cos Y sin , X sin +Y cos ).
What is the ielation between the paitial deiivatives of the tempeiatuies as computed by
A and by B` Te chain iule gives the answei
T
B
X
=

X
_
T
A
_
X cos Y sin
. .
=x
, X sin +Y cos
. .
=y
_
_
=
T
A
x
cos +
T
A
y
sin .
11.3. Another example Polar coordinates. Suppose a quantity P is given in teims
of Caitesian cooidinates x and y P = f(x, y). How does P change if we vaiy the polai
cooidinates r and , i.e. what aie the paitial deiivatives of P with iespect to r and `
To answei this question we must wiite P as a function of r and . Recall that the
ielation between Caitesian cooidinates and polai cooidinates is
(90) x = r cos , y = r sin .
Teiefoie P = f(x, y) = f(r cos , r sin ) and we get
(91)
P
r
= cos
f
x
+ sin
f
y
,
P

= r sin
f
x
+r cos
f
y
Since the function f always gives us the value of the quantity P, these ielations aie
usually wiiuen in this way
(92)
P
r
= cos
P
x
+ sin
P
y
,
P

= r sin
P
x
+r cos
P
y
Using the ielation (90) between polai and Caitesian cooidinates we can wiite these equa-
tions in yet anothei way
(93)
P
r
=
x
r
P
x
+
y
r
P
y
,
P

= y
P
x
+x
P
y
12. Problems
1. Use the chain rule to compute dz/dt for
z = sin(x
2
+y
2
), x = t
2
+ 3, y = t
3
.
2. Use the chain rule to compute dz/dt for
z = x
2
y, x = sin(t), y = t
2
+ 1.
One way of aiiiving at these ielations is to use vectois as in the ist vectoi woik sheet of this semestei.
4. DERlVATlVES
3. Use the chain rule to compute z/s and
z/t for z = x
2
y, x = sin(st), y = t
2
+s
2
.

4. Use the chain rule to compute z/s and


z/t for z = x
2
y
2
, x = st, y = t
2
s
2
.
5. (a) Let x = x(u, v), y = y(u, v) be the
following set of functions of u, v:
x = u
2
v
2
, y = 2uv.
If g(u, v) = f(x(u, v), y(u, v)) then
compute gu(1, 0), gu(1, 1), gv(1, 0), and
gv(1, 1), if you are given these values of the
partial derivatives of f:
x y fx(x, y) fy(x, y)
0 0 A B
1 0 C D
0 1 E F
1 1 G H
2 0 I J
0 2 K L
(b) Repeat the above problem if x and y are
given by x = u, y = v/u.
(c) Repeat part (a) of this problem if x and y
are given by x = u +v, y = u v.
6. Let x, y, X, Y, TA, and TB be as in the ex-
ample in 11.2. In that section we computed
T
B
X
.
(a) Compute
TB
Y
.
(b) Show that
(
TA
x
)
2
+
(
TA
y
)
2
=
(
TB
X
)
2
+
(
TB
Y
)
2
.
In other words, A and B may measure dier-
ent partial derivatives, but the temperature
gradients they find have the same length.

#
TA =
#
TB.
7. (About polar coordinates). In 11.3 we
sawhowwe can use the chain rule to find
f
r
and
f

if we know the function f in terms


of Cartesian coordinates (x, y). In this prob-
lem we turn the question around: suppose
we are given a function in polar coordinates,
how do we compute its gradient.
Recall that polar and Cartesian coordi-
nates are related by
r =

x
2
+y
2
and = aictan
y
x
,
at least in the region where x > 0. (See
Chapter III, 4.)
(a) Compute
r
x
,
r
y
,

x
,

y
. Try to simplify
your answer as much as possible, by reusing
the variables r and . For instance, the sim-
plest way to write
r
x
is as
r
x
=
x
r
.
(b) Suppose a quantity P is given in terms of
Polar coordinates by P = f(r, ). Express
P
x
and
P
y
in terms of
f
r
and
f

.
More precisely, compute
P
x
def
=

{
f(r(x, y), (x, y)
}
x
and
P
y
def
=

{
f(r(x, y), (x, y)
}
y
(c) Show that

#
P
2
=
(
f
r
)
2
+
1
r
2
(
f

)
2
.
8. For some function f we are told that at
the point with Cartesian coordinates (4, 3)
one has
f
r
= 3,
f

= 6.
Compute the gradient
#
f at (2, 1).
9. In physics an electric field is described
by its potential function, = (x, y) (in
this problem we assume the world is two-
dimensional; the potential is measured in
Volts). Minus the gradient of the potential
function is called the electric field:
#
E =
#
.
The electric potential of a point charge in
the plane is given in Polar coordinates by
= C ln r, for some constant C (the
physicists will tell you that C depends on the
charge that was placed at the origin; for us
it is just some number, and we will in fact
assume that C = 1.)
(a) Compute the electric field
#
Ecorrespond-
ing to the potential = ln r.
(b) Compute
#
E (this quantity measures
the strength of the electric field, but not
12. PROBLEMS
its direction.) Where is the electric field
stronger?
(c) Make a drawing of the level curves of the
potential , and the electric field
#
E.
(d) In the three dimensional world the elec-
tric potential generated by a charged parti-
cle at the origin is not given by C ln r, but
instead by the so-called Coulomb potential
=
C
r
, where r =

x
2
+y
2
+z
2
.
Compute the corresponding electric field
#
E =
#
.
10. The ideal gas law, given by PV = nRT,
relates the Pressure, Volume, and Tempera-
ture of n moles of gas. (R is the ideal gas
constant). Thus, we can view pressure, vol-
ume, and temperature as variables, each one
dependent on the other two.
(In this problem pressure is measured in
Pascals, temperature in degrees Kelvin, and
volume in Liters.)
Each of the following three questions can
be answered by applying the chain rule to
dierentiate z(t) = f(x(t), y(t)) for suit-
able quantities x, y, and z. In each case state
which variables play the role of x, y, z, and
what the function f is.
(a) If pressure of a gas is increasing at a rate
of 0.2Pa/min and temperature is increasing
at a rate of 1

K/min, how fast is the volume


changing?
(b) If the volume of a gas is decreasing at
a rate of 0.3L/min and temperatuere is in-
creasing at a rate of 0.5

K/min, how fast is


the pressure changing?
(c) If the pressure of a gas is decreasing at
a rate of 0.4Pa/min and the volume is in-
creasing at a rate of 3L/min, how fast is the
temperature changing?
11. The ideal gas law says PV = nRT,
where P, V, T are variables, and n, R are
constants. Verify the following identity:
P
V
V
T
T
P
= 1
12. The previous exercise was a special case
of the following fact, which you are asked to
verify here:
Assume that F(x, y, z) is a function of
3 variables, and suppose that the relation
F(x, y, z) = 0 defines each of the vari-
ables in terms of the other two, namely x =
f(y, z), y = g(x, z) and z = h(x, y), then
x
y
y
z
z
x
= 1.
Hint: this is a problem about implicit dier-
entiation.
13. Four cartographers are using dierent co-
ordinates to describe the same landscape.
Each of them describes the landscape by
specifying a the height of a point in the land-
scape as a function of its position above a
horizontal plane.
Cartographer A uses Cartesian coordi-
nates (x, y) in the plane, B uses Cartesian
coordinates (X, Y ) in the plane. The coordi-
nates (X, Y ) are rotated by 45

with respect
to (x, y) (see 11.2).
Cartographer C works with A but uses
polar coordinates (r, ) (r is the distance to
the origin, is the angle with As x-axis).
Cartographer D works with B and uses
polar coordinates (r, ) (r is the distance to
the origin, is the angle with Bs X-axis).
Here is a picture of the landscape that A,
B, C, and D are looking at:
(a) If B has found that the height is given
by the function f(X, Y ) = 2XY /(X
2
+
Y
2
), then what function does A find for the
height?
(b) What height function does C find?
(c) What height function does D find?
14. Brian and Ally are using dierent Carte-
sian coordinate systems in the plane: (x, y)
for Ally, (X, Y ) for Brian. They have the
same origin, but Brians coordinates are ro-
tated by an angle of = aictan
4
3
( 53

,
although that is only an approximation. You
8 4. DERlVATlVES
can give exact answers in this problem, and
you dont need a calculator.)
(a) What is the relation between (x, y) and
(X, Y )?
(b) If Ally has found that TA(x, y) =
32+0.1y, then what formula TB(X, Y ) will
Brian use to describe the temperature?
(c) On a dierent occasion Ally found that
the temperature had changed. Now Ally
measures the temperature and finds that at
the point with x = 1, y = 1 one has
TA(1, 1) = 35, and also
T
A
x
= 0.05 and
T
A
y
= 0.8. Which coordinates does Brian
assign to this point, which temperature TB,
and which derivatives
T
B
X
and
T
B
Y
does
Brian compute at this point?
[Hint: before you compute anything, find
sin and cos ; also drawa right triangle one
of whose acute angles is .]
13. Higher Partials and Clairauts eorem
13.1. Higher partial derivatives. By denition
(94)

2
f
x
2
=

_
f
x
_
x

2
f
xy
=

_
f
y
_
x

2
f
yx
=

_
f
x
_
y

2
f
y
2
=

_
f
y
_
y
ln subsciipt notation one wiites these highei paitial deiivatives as follows
f
xx
(x, y) =

2
f
x
2
f
xy
(x, y) =

2
f
yx
f
yx
(x, y) =

2
f
xy
f
yy
(x, y) =

2
f
y
2
.
Note the reversal in xy order in the mixed partial derivatives!
13.2. Example. lf f(x, y) = x
2
y + cos xy then f
x
= 2xy y sin xy, and hence
f
xx
=
(2xy y sin xy)
x
= 2y y
2
cos xy,
f
xy
=
(2xy y sin xy)
y
= 2x sin xy xy cos xy.
Te othei paitial deiivatives follow fiom f
y
= x
2
xsin xy, and they aie
f
yx
= 2x sin xy xy cos xy, f
yy
= x
2
cos xy.
Eveiy time we take a deiivative, we can choose whethei we dieientiate with iespect
to x oi y. Dieientiating once we have two possibilities, dieientiating twice we have
2 2 = 4 possibilities, etc. Tat is why we found foui paitial deiivatives of second oidei
in the above example. But if we look caiefully, we also see that f
xy
and f
yx
aie the same.
Tis is no coincidence.
13.3. Clairauts eorem mixed partials are equal. If for a given function f of
two variables the mixed partial derivative f
xy
(x, y) exists for all (x, y) in a neighborhood
of a point (a, b), and if this derivative is continuous at (a, b), then the other mixed partial
derivative f
yx
(a, b) also exists, and f
xy
(a, b) = f
yx
(a, b).
So we noimally dont have to woiiy about the oidei in which we take paitial deiiva-
tives.
14. FlNDlNG A FUNCTlON FROM lTS DERlVATlVES 9
13.4. Proof of Clairauts theorem. With some algebia we can show that the deni-
tion of paitial deiivatives implies
(9)

2
f
xy
=
lim
x0
lim
y0
f(x + x, y + y) f(x, y + y) f(x + x, y) +f(x, y)
xy
while
(9)

2
f
yx
=
lim
y0
lim
x0
f(x + x, y + y) f(x, y + y) f(x + x, y) +f(x, y)
xy
So its a mauei of showing that one can switch the two limits. We wont go into the
details heie, but the hypothesis that f
xy
is continuous implies that we aie indeed allowed
to switch the limits.
14. Finding a function from its derivatives
We now look at integiating the paitial deiivatives of a function, which looks out of
place heie (this being a chaptei on deiivatives and not on integials), but Claiiauts Teo-
iem actually tuins out to play a iole.
lf we have the deiivative f

(x) of some function of one vaiiable then we know how to


iecovei the function f(x) we integiate, i.e.
f(x) =

(x)dx +C.
Fuitheimoie, any (continuous) function can be the deiivative of a function, because, if
someone gives us a continuous function f(x), then
F(x)
def
=

x
a
f(t)dt
is a dieientiable function whose deiivative is F

(x) = f(x).
What about functions of moie than one vaiiable` Suppose we knowthe paitial deiiva-
tives
(9)
f
x
= P(x, y) and
f
y
= Q(x, y)
of a function of two vaiiables, can you then nd the function f(x, y)`
Te answei is yes, you can nd f by integiating, if it exists, but not eveiy paii of
functions P and Q aie the paitial deiivatives of some function.
Te following two examples aie typical of what can happen.
14.1. Example. Does theie exist a function f(x, y) of two vaiiables such that
f
x
= x
3
2xy, and
f
y
= 3y
2
both hold` Te answei is no, such a function cannot exist, and heie is the ieason if theie
weie such a function, then we could compute

2
f
yx
=
(x
3
2xy)
y
= 2x, and

2
f
xy
=
(3y
2
)
x
= 0.
80 4. DERlVATlVES
By Claiiauts Teoiemboth computations should give us the same answei, but they dont.
Teiefoie the function f whose paitials aie as above cannot exist.
14.2. Example. Does theie exist a function f(x, y) of two vaiiables whose deiiva-
tives aie
f
x
= x
3
2xy, and
f
y
= sin y x
2
?
Lets check Claiiauts condition

2
f
yx
=
(x
3
2xy)
y
= 2x, and

2
f
xy
=
(sin y x
2
)
x
= 2x.
Tis time both computations gave us the same answei, so Claiiauts theoiem does not
iule out the existence of the function f that we aie looking foi. We can tiy to compute
it by integiating both paitial deiivatives. Teie is a systematic way of doing this that
usually leads to the answei.
We ist integiate f
x
while tieating y as a constant
f(x, y) =

{x
3
2xy} dx =
1
4
x
4
x
2
y +C(y).
Te constant is only a constant in the sense that it does not depend on x. lt may depend
on y, and that is why we wiote it as C(y). To nd C(y) we dieientiate this iesult with
iespect to y
sin y x
2
= f
y
=

_
1
4
x
4
x
2
y +C(y)
_
y
= x
2
+C

(y).
So we see that C

(y) = sin y, and hence C(y) =


1

cos y + K, wheie K is a ieal


constant (K depends neithei on x noi on y).
We nd that the following function has the piesciibed paitial deiivatives
f(x, y) =
1
4
x
4
x
2
y
1

cos y +K
wheie K is constant, i.e. wheie K depends on neithei x noi y.
Te method used in this example always woiks, and we summaiize this fact in the
following theoiem.
14.3. eorem. Suppose P(x, y) and Q(x, y) are two functions that are dened on a
rectangular domain R = {(x, y) : a < x < b, c < y < d}, and suppose that they have
continuous partial derivatives on this domain.
If a function f(x, y) exists such that (9) holds on R, then
(98)
P
y
=
Q
x
must hold on R.
Conversely, if P and Q satisfy (98) then there is a function f dened on R that satises
(9).
To piove this theoiemwe need to undeistand integials of functions of seveial vaiiables,
and Gieens theoiem in paiticulai, so this will have to wait until the end of the semestei.
See Vll.11.
lt should be noted that the assumption above that the functions P and Q be dened
on a iectangle is impoitant the theoiem is no longei tiue if the domain of P and Q has
holes. See pioblem 1.1.
1. PROBLEMS 81
15. Problems
1. Find all first and second partial deriva-
tives of x
3
y
2
+y
5
.
2. Find all first and second partial deriva-
tives of 4x
3
+xy
2
+ 10.
3. Find all first and second partial deriva-
tives of xsin y.
4. Find all first and second partial deriva-
tives of sin(3x) cos(2y).
5. Find all first and second partial deriva-
tives of e
x+y
2
.
6. Find all first and second partial deriva-
tives of ln

x
3
+y
4
.
7. Find all first and second partial deriva-
tives of z with respect to x and y if x
2
+
4y
2
+ 16z
2
64 = 0. (Hint: solve for z or
use implicit dierentiation)
8. Find all first and second partial deriva-
tives of z with respect to x and y if xy +
yz + xz = 1. (Hint: solve for z or use im-
plicit dierentiation)
9. How many dierent second partial
derivatives does a function of two variables
have? What about a function of three vari-
ables? Howmany derivatives of third degree
does a function of two variables have?
10. Derive the formulas (9) and (9) fromthe
definition of partial derivatives (1) and (2).
11. The equation which describes the vibrat-
ing string (as in a guitar, piano, or violin
string) is
(99)

2
f
t
2
= c
2

2
f
x
2
where c > 0 is some constant. The equation
is called the wave equation. It is an example
of a partial dierential equation.
Note : this problem looks like a prob-
lem about dierential equations, but to an-
swer the following questions you really only
have to compute partial derivatives of cer-
tain functions, and solve some (easy) alge-
braic equations.
(a) For which values of the constant v is a
traveling wave with velocity v and profile
F(x) a solution of the wave equation (99)?
Does it maer which profile F is used here?
(For the terminology used here, revisit
problem 5.16 in Chapter III, 5.2.)
(b) Suppose the string is clamped down at
its ends, and that its length is L. For which
values of the constants A and is
f(x, t) = Asin(t) sin
x
L
a solution of the wave equation? (Assume
A = 0).
(c) Same question for
g(x, t) = Bsin(t) sin
2x
L
.
(d) Describe the movies that go with the so-
lutions you found in (b) and (c). Which of
the two graphs moves faster?
(e) Show that h(x, t) = f(x, t) + g(x, t) is
again a solution of the wave equation, where
f and g are as above. (Dont use the formu-
las for f and g: it is easier to prove a more
general fact, namely, if two functions f and
g satisfy (99), then so does their sumf +g.)
(f) Describe the movie that goes with the
function h(x, t) (it is probably beer to use
a graphing application like grapher.app on
Mac OS X, graphcalc.exe on Windows or
Linux).
12. Suppose P(x, y) = x
2
2xy
3
and
Q(x, y) = (xy)
2
. Does there exist a func-
tion f(x, y) such that P = fx and Q = fy?
13. Suppose P(x, y) = x
2
+ axy
3
and
Q(x, y) = (xy)
2
, where a is a constant. For
which a does there exist a function f(x, y)
such that P = fx and Q = fy?
14. Suppose P(x, y) = x
2
2xy
3
and
Q(x, y) = (xy)
2
. Does there exist a func-
tion f(x, y) such that P = fx and Q = fy?
82 4. DERlVATlVES
15. Suppose x = u + v, y = u v, and
suppose f(x, y) = g(u, v). Then compute
(a)

2
g
u
2

(b)

2
g
v
2

(c)

2
g
uv

(d)

2
g
u
2


2
g
v
2

(e)

2
g
u
2
+

2
g
v
2

16. [For discussion] Let
P(x, y) =
y
x
2
+y
2
, Q(x, y) =
x
x
2
+y
2
.
(a) What is the domain of P and Q?
(b) Show that
P =

x
, Q =

y
where is the angle variable from polar co-
ordinates.
(c) Show that P and Qsatisfy the condition
(98). (You dont have to compute the deriva-
tives to check this, although you could.)
(d) Is there a function f such that (97) holds?
CHAPTER
Maxima and Minima
ln ist semestei calculus we leained how to nd the maximal and minimal values
of a function y = f(x) of one vaiiable. Te basic method is as follows assuming the
independent vaiiable is iestiicted to some inteival a x b, we ist look foi inteiioi
maxima and minima. Tese always occui at critical oi stationary points of the function,
i.e. solutions x of f

(x) = 0. We then check the function values at the endpoints a and b


of the inteival, to see if they might be maxima oi minima.
To nd out which solutions of f

(x) = 0 aie actually local maxima oi minima we


can look at the sign of the deiivative f

(x) to see wheie the function is incieasing oi


decieasing, oi we can apply the second derivative test.
Tis chaptei we will see how to solve similai questions about functions of two oi moie
vaiiables.
1. Local and Global extrema
Let z = f(x, y) be the function whose maximal oi minimal values we aie looking
foi, and let D be the domain of this function. Tis domain could be the laigest possible
domain foi the given function (in case f is dened by a foimula), but it could also be some
smallei iegion which we ouiselves have chosen. Te question we aie consideiing is
What are the largest and smallest values that f(x, y) can have
if the point (x, y) belongs to the domain D?
1.1. Denition of global extrema. e function f has a global maximum or abso-
lute maximum at a point (a, b) in D if f(x, y) f(a, b) for all points (x, y) in D.
Similaily, the function f has a global minimumor absolute minimumat a point (a, b)
in D if f(x, y) f(a, b) for all points (x, y) in D.
1.2. Denition of local extrema. e function f has a local maximum at a point
(a, b) in D if there is a r > 0 such that f(x, y) f(a, b) for all points (x, y) in D which
also lie in a disc of radius r centered at (a, b).
Local minima aie dened analogously.
1.3. Interior extrema. Recall that a point (a, b) in a domain D is called interior if it
is not a boundaiy point, oi, moie piecisely, if theie is some small r > 0 such that the disc
with iadius r centeied at (a, b) is entiiely contained in D. We will apply this distinction
to the local and global maxima and minima that we nd an interior local minimum is
a local minimum that occuis at an inteiioi point of the domain D of the function.
2. Continuous functions on closed and bounded sets
Befoie we go into the details of how we can actually nd the maxima and minima, it is
good to know the following geneial fact. lt tells us wheie to expect maxima and minima.
83
84 . MAXlMA AND MlNlMA
Figure 1. The graph of f(x, y) = x
2
+ y
2
from example 2.2 on three dierent rectangles Q.
From le to right:
(i) 0 x 1, 0 y 1. Both max and min are aained at a corner point of the rectangle.
(ii) 0 x 1, 1 y 1. Two maxima, both are aained at corner points of the rectangle;
the minimum is aained at an edge point.
(iii) 1 x 1, 1 y 1, Four maxima, all aained at corner points of the rectangle; the
minimum is aained at an interior point.
Let z = f(x
1
, . . . , x
n
) be a continuous function dened on some closed and bounded
iegion Din R
n
. Heie closed means that Dcontains all its boundaiy points, and bounded
means that all points in D aie not fuithei away fiom the oiigin than some xed iadius R
(D does not stietch all the way to innity.)
We will also assume that f is continuous on D.
2.1. eorem about Maxima and Minima of Continuous Functions. A continuous
function dened on a closed and bounded region D R
n
has both a maximum and mini-
mum within that region.
Te piecise denitions of the concepts (continuous, closed, bounded) and the pioof of
this theoiem all involve a faii numbei of s and s. Tis mateiial is tieated in couises
like Math 421, 21 (ieal analysis) oi 1 (point set topology) and ieally does not belong
heie in Math 234. Neveitheless it is impoitant to have some undeistanding of what is
meant in the above theoiem. Te following examples aie meant to claiify this.
2.2. Example e function f(x, y) = x
2
+ y
2
. Tis function is continuous, and
the squaie Q = {(x, y) : 0 x 1, 0 y 1} is bounded, and it contains all
boundaiy points (the edges of the squaie). Teiefoie Teoiem 2.1 tells us that f auains
both its highest and lowest values somewheie in the squaie. Te theoiem does not say
wheie these max/min points aie, but in this example they aie easy to nd. Te function
f(x, y) = x
2
+ y
2
is at its smallest when both x = 0 and y = 0, i.e. at the bouom-lef
coinei of the squaie. And f(x, y) is at its laigest when x and y aie both as laige as they
can be, i.e. when x = 1 and y = 1. Tis happens at the top-iight coinei of the squaie.
Note that the boundaiy of the iectangle Qhas two dieient kinds of points it has foui
coinei points, and then all the othei points that lie on the edges.
lf we change the iectangle Q then the minimum can appeai at a coinei point, a point
on an edge, oi in an inteiioi point. See Figuie 1.
2.3. A shy example. Considei the function f(x, y) = x
2
x
3
y
2
. lts zeio set is
the cuive y
2
= x
2
x
3
, which is shaped like the leuei , oi like a sh see Figuie 2.
3. PROBLEMS 8
Te function is positive on the tail (D
1
) and also on the body (D
2
) of the sh, it vanishes
on the cuive that tiaces out the sh, and f is negative elsewheie.
We assume that both iegions D
1
and D
2
aie closed, which means that we assume that
they include theii boundaiy points. See Figuie 2 below.
Teoiem2.1 does not apply to the iegion D
1
because D
1
is not bounded (it contains the
whole negative x-axis). But the iegion D
2
is bounded, and oui function f is continuous,
so Teoiem 2.1 does apply to D
2
. Te theoiem tells us that the function f has a maximal
value and a minimal value somewheie in D
2
. ln the inteiioi of D
2
the function is stiictly
positive, and at the boundaiy points of D
2
we have f = 0. Teiefoie each boundaiy
point is a minimum point of f on D
2
. Te point(s) in D
2
wheie f auains its highest value
must be somewheie in the inteiioi of D
2
. ln the next section we will see how to nd it
(and how to check that in this case theie ieally only is one such point.)
x
y
D
1
D
2
x
2
-
y
2
-
x
3
=
0
interior points
boundary points
Figure 2. Le: The region where f(x, y) = x
2
x
3
y
2
is positive consists of two parts, one
bounded (D2), and the other unbounded (D1). Theorem 2.1 does not apply to the unbounded
region, but it does apply to the bounded region D2. In that region f must aain a maximum
and also a minimum. Since f = 0 on the boundary of the region D2, and f > 0 in the interior, f
achieves its lowest value in D2 everywhere on the boundary of D2 and its highest value somewhere
in the interior. Theorem 2.1 does not tell us how to find that interior point, and allows for the
possibility that there might be more interior maxima, as well as a few interior (local) minima.
Right: The graph of the function z = x
2
y
2
x
3
.
3. Problems
1. Suppose you want to find the maximal
value of f(x, y) = x
2
x
3
y
2
over all
possible (x, y) with x 0 (and no restric-
tion on y this region is called the right half
plane).
(a) Explain why you should always choose
y = 0 in order to maximize this particular
function f(x, y).
(b) Use your answer to part (a) to find the
point (x, y) that maximizes f(x, y) over the
right half plane.
(c) Does our function f(x, y) have a maxi-
mal value if (x, y) can be any point in the
plane? (hint: what is f(1000, 0)?)
8 . MAXlMA AND MlNlMA
2. Suppose that D is a bounded and closed
region in the plane (you should draw one:
any region will do as long as you include the
boundary points).
Where does the function f(x, y) = x
aain its maximum in the region that you
drew? Can f aain its maximum at an inte-
rior point of the region?
What about minima?
3. Draw the region
R =
{
(x, y) : y
2
4(x
3
x
4
)
}
.
Find the largest and smallest values that the
function f(x, y) = x can have on this re-
gion.
(Hint: where is 4(x
3
x
4
) = 4x
3
(1x)
positive? The region looks like an Onion).
4. Critical points
Foi functions y = f(x), a x b, of one vaiiable the standaid way of nding
minima (and maxima) is to look foi them in two dieient places eithei the minimum is
auained at one of the end points x = a oi x = b of the inteival, oi else the minimum is
auained at an inteiioi point. At an inteiioi minimum one has f

(x) = 0, so they can be


found by solving the equation f

(x) = 0. Te same appioach woiks foi functions of two


oi moie vaiiables. Te basic fact that tells us that this is so, is the following theoiem.
4.1. Denition (critical point). A critical point of a function z = f(x, y) of two vari-
ables is a point (a, b) at which
#
f(a, b) = 0, i.e. at which
f
x
(a, b) = 0 and f
y
(a, b) = 0.
At a ciitical point of a function the tangent plane to the giaph is hoiizontal.
4.2. eorem. Local extrema are critical points. If a function z = f(x, y) dened
on a domain D has a local minimum or local maximum at an interior point (a, b) then one
has
f
x
(a, b) = 0, and
f
y
(a, b) = 0.
Picture proof. (See Figuie 3.) lf f has a local maximum at an inteiioi point (a, b) then
f(x, y) f(a, b) foi all (x, y) close to (a, b). Tis means that a small piece of the giaph
of f neai its local maximum at (a, b, f(a, b)) lies below the plane z = f(a, b). Tis plane
must theiefoie be the tangent plane to the giaph of f. Being hoiizontal, its slopes aie
zeio, and these slopes aie exactly the paitial deiivatives of f at (a, b).
Frozen variable proof. Suppose f has a local maximum at an inteiioi point (a, b) of
the domain D. Ten we can fieeze the y-vaiiable at the value y = b and considei the
function of one vaiiable g(x) = f(x, b). Tis function has a maximum at x = a, so by
ist semestei calculus we know that g

(x) = 0. By denition g

(a) = f
x
(x, b), so we
conclude that f
x
(a, b) = 0.
By fieezing x instead of y we nd that f
y
(a, b) = 0 also must hold.
Te same aiguments apply in the case of a local minimum.
4.3. ree typical critical points. Lets nd the ciitical points of the following thiee
functions
f(x, y) = x
2
+y
2
, g(x, y) = x
2
y
2
, h(x, y) = x
2
y
2
.
4. CRlTlCAL POlNTS 8
fx
=
0 f
y=
0
x
y
Figure 3. Theorem 4.2: at a local maximum the tangent plane to the graph is horizontal. The
partial derivatives w.r.t. both x and y vanish, and in fact, the derivative along any path through
(a, b) vanishes. To see a picture of a local minimum turn the page upside down.
f(x, y) = x
2
+y
2
. Computing the paitial deiivatives we nd foi the ist function
f
x
= 2x,
f
y
= 2y.
lf (x, y) is a ciitical point of f then x and y must satisfy the equations f
x
(x, y) = 0 and
f
y
(x, y) = 0, in this case, 2x = 0 and 2y = 0. So we see that f has exactly one ciitical
point, namely the oiigin (x, y) = (0, 0).
Is this critical point perhaps a minimum or a maximum? Since squaies can nevei be
negative, f(x, y) = x
2
+ y
2
is always non-negative, and it is at its smallest when both
teims x
2
and y
2
vanish, i.e. when x = y = 0. So f(x, y) has a global minimum at the
oiigin.
h(x, y) = x
2
y
2
. Tis function is just f(x, y), and without looking at its
deiivatives we can tell that it has a global maximum at the oiigin (because f(x, y) has a
global minimum theie). Te deiivatives aie
h
x
= 2x,
h
y
= 2y
so that the oiigin is the only ciitical point of this function.
local max local min saddle point
Figure 4. The three most common kinds of critical point. See the examples in 4.3 and also the
second derivative test in 9.
88 . MAXlMA AND MlNlMA
g(x, y) = x
2
y
2
. Te deiivatives of g aie
g
x
= 2x,
g
y
= 2y,
so, once again, the oiigin is the only ciitical point. But, unlike the pievious two functions,
g has neithei a maximum noi a minimum at the oiigin. We can see this by ist looking
at what g does on the x-axis, and then what g does on the y-axis
On the x-axis we have g(x, 0) = +x
2
, so g has a minimum at the oiigin.
On the y-axis we have g(0, y) = y
2
, so g has a maximum at the oiigin.
So aibitiaiily close to the oiigin we can nd points (x, y) wheie g(x, y) is laigei than
g(0, 0), and we can nd othei points wheie g(x, y) is smallei than g(0, 0). Teiefoie g
does not have a local maximum oi a local minimum at the oiigin.
Figuie 4 shows the thiee cases we have just discussed.
4.4. Critical points in the shy example. What are the critical points of the function
f(x, y) = x
2
x
3
y
2
from 2.3?
We compute the paitial deiivatives of the function
f
x
= 2x 3x
2
= (2 3x)x,
f
y
= 2y.
Te equation f
y
= 0 implies that y = 0, while f
x
= 0 implies x = 0 oi x =
2
3
. Teiefoie
f has two ciitical points one at the oiigin (0, 0), and the othei at (
2
3
, 0).
x
y
D
1
D
2
x 2
-
y 2
-
x 3
=
0
interior points
boundary points
ln this example we could have alieady piedicted fiom the shape of the zeio set of f
that f has at least two ciitical points we dont need to compute the deiivatives of f foi
that. Namely, the zeio set of f is a cuive that ciosses itself at the oiigin, so the lmplicit
Function Teoiem10.1 (chaptei 2) cannot hold at the oiigin, and hence f
x
= f
y
= 0 theie.
And in 2.3 we aigued that the function f must have a local maximum somewheie in
the iegion D
2
(Figuie 2), so f must have at least two ciitical points. On the othei hand,
by computing the ciitical points we have found that theie is only one local maximum in
the iegion D
2
.
4.5. Another example nd the critical points of f(x, y) = x x
3
xy
2
.
Solution: Te deiivatives of oui function aie
f
x
= 1 3x
2
y
2
,
f
y
= 2xy.
Te ciitical points aie theiefoie the solutions of the equations
1 3x
2
y
2
= 0, 2xy = 0.
Tis is a system of two equations, with two unknowns (that always happens when we
look foi ciitical points, since we aie looking foi solutions of f
x
(x, y) = 0, f
y
(x, y) = 0.)
Te second equation, 2xy = 0, implies that eithei x = 0 oi y = 0 (oi both). We have
to tieat these two cases sepaiately
e case x = 0. lf x = 0 then we only have the ist equation lef,
which tells us 1y
2
= 0, i.e. y = 1. We nd two ciitical points with
x = 0, namely, (0, 1) and (0, 1).
e other case, x = 0. lf x = 0, then the second equation (2xy =
0) implies y = 0. Substitute this in the ist equation and we nd
1 3x
2
= 0, i.e. x =
1
3

3, so that we have two ciitical points with


x = 0, namely, (
1
3

3, 0) and (
1
3

3, 0).
. WHEN THERE ARE MORE THAN TWO VARlABLES 89

+
+
+
+
+
+

A
B
D C

+
+
+
+
+
+
+
+
+
+
+
+

Figure 5. The zero set and signs of the function f(x, y) = x x


3
xy
2
.
Te conclusion is that this function has foui ciitical points, two on the x-axis, and two
on the y-axis. Without looking into this in any fuithei detail we cannot tell if any of these
points aie local maxima oi minima. ln geneial the second deiivative test (to be explained
in 9) will piovide this infoimation. Foi this example a look at the zeio set of f also helps
us guie out what kind of ciitical points we have found. Since f factois as
f(x, y) = x (1 x
2
y
2
),
we see that its zeio set consists of the line x = 0 and the unit ciicle x
2
+ y
2
= 1. ln the
above pictuie f > 0 in the giey iegion, and f < 0 in the white aiea. Considei the iight
half of the unit disc. Te function is positive in the inteiioi, and zeio on the boundaiy
of this iegion. Just as in the shy example of 2.3, we have anothei case wheie the
maximum of the function must be auained at one oi moie inteiioi points of the iight half
of the unit disc. Accoiding to oui computation f only has one ciitical point in the iight
half ciicle, and theiefoie this point must be a local maximum of the function. Conclusion
D = (
1
3

3, 0) is a local maximum.
ln the same spiiit you can aigue that f has a local minimum at C.
Te othei two points A, B aie neithei local maxima noi minima, since arbitrarily close
to A or B theie aie both points (x, y) with f(x, y) positive, and points with f(x, y) neg-
ative. Te points A and B tuin out to be saddle points (see 9 on the second deiivative
test.)
5. When there are more than two variables
Te whole discussion so fai has been about functions of two vaiiables. Foitunately,
not much changes when you have moie vaiiables. Te concepts local minimum and local
maximum aie dened in the same way, and it tuins out that any interior local maximum
or minimum must be a critical point of the function. Heie, by denition, a ciitical point of
a function w = f(x
1
, . . . , x
n
) of n vaiiables is a solution of the equations
_

_
f
x
1
(x
1
, , x
n
) = 0
f
x
2
(x
1
, , x
n
) = 0
.
.
.
f
x
n
(x
1
, , x
n
) = 0.
90 . MAXlMA AND MlNlMA
Obseive that theie aie n equations, and that theie aie also n unknowns (x
1
, , x
n
) so
that we should in principle be able to solve these equations. ln piactice the system of
equations we get can be veiy easy, dicult, oi simply impossible to solve.
. PROBLEMS 91
6. Problems
1. Find all critical points of the following
functions. Try to classify them into lo-
cal/global maxima/minima, saddles, or other
kind of critical points. (Write clear solutions.
You will need your solutions later in problem
10.5.)
(a) f(x, y) = x
2
+ 4y
2
2x + 8y 1
(b) f(x, y) = x
2
y
2
+ 6x 10y + 2
(c) f(x, y) = x
2
+ 4xy +y
2
6y + 1
(d) f(x, y) =
x
2
xy + 2y
2
5x + 6y 9
(e) f(x, y) = y
2
18x
2
+x
4

(f) f(x, y) = y
4
4y
2
18x
2
+x
4

(g) f(x, y) = 9 + 4x y 2x
2
3y
2

(h) f(x, y) = xy(4 x 2y)


(i) f(x, y) = x(x y)(x 1)
(j) f(x, y) = (x y)(xy 4)
(k) f(x, y) = y
2
+ cos x
(l) f(x, y) = x
2
y
1
3
y
3

(m) f(x, y) = (x y
2
)(x 1)
(n) f(x, y) = (x y)(xy 4)
(o) f(x, y) = x
2

(p) f(x, y) = x
2
y
(q) f(x, y) =
(
1 x
2
y
2
)
2

(r) f(x, y) = x
2
y
2.
(a) Draw the zero set of the function
f(x, y) = sin(x) sin(y).
(b) Where is the function f positive? Find
as many critical points as you can without
computing fx or fy.
(c) Find all critical points of f(x, y). Which
are local minima or local maxima?
3. Find the critical points of the function
f(x, y, z) = x
2
+y
2
+z
2
2x + 4y 2.
4. Draw the zero set and find the critical
points of the functions
f(x, y, z) = x
2
+y
2
z
2
and
g(x, y, z) = x
2
y
2
z
2
5. If we have three points A, B, and C in
the plane, which point is closest to all three
of them? The answer depends on what we
mean by closest to all three points. The fol-
lowing problem gives us one interpretation
of this general question.
Consider the three points (1, 4), (5, 2),
and (3, 2) in the plane. The function
f(x, y, z) =
(x 1)
2
+ (y 4)
2
+
(x 5)
2
+ (y 2)
2
+
(x 3)
2
+ (y + 2)
2
is the sum of the squares of the distances
from point (x, y) to the three points.
(a) Assuming that there is a global mini-
mum, find x and y so that f(x, y) is mini-
mized.
(b) (For discussion) Does f(x, y) have a
global minimum? How can we be sure that
the point we found in part (a) is not actu-
ally a maximumor some other critical point?
(c) Given the three points (a, b), (c, d), and
(e, f), let f(x, y) be the sum of the squares
of the distances from point (x, y) to the
three points. Find x and y so that this quan-
tity is minimized.
6. Suppose that a function f(x, y) factors,
i.e. we can write it as the product of two
other dierentiable functions, f(x, y) =
g(x, y)h(x, y).
Prove: if a point (a, b) lies in the zero set
of g and also in the zero set of h, then (a, b)
is a critical point of f.
Hint: compute the partial derivatives of f by
applying the product rule to f = g h.
7. Find the critical points of the functions
(a) f(x, y, z) = x
2
+y
2
+z
2
2x+4y 2
(b) f(x, y, z) = x
4
+y
2
+z
2
2xz + 4y
(c) f(x, y, z) = xyze
xyz
(d) f(x, y, z) = x
2
+y
2
+z
2
2xyz
92 . MAXlMA AND MlNlMA
7. A Minimization Problem: Linear Regression
Suppose we aie measuiing two quantities x and y in some expeiiment, and suppose
that we expect that theie is a lineai ielation of the foim y = ax + b between x and y.
lf we have a set of data points (x
k
, y
k
) fiom oui expeiiment, then what do they tell us
about a and b` Whi oice of coecients a and b bests ts our data? Because
of expeiimental eiiois we would not expect oui data points to lie on a stiaight line, but
instead, we expect them to be clusteied aiound a stiaight line. We could plot the data
points, get a iulei, and diaw a stiaight line by hand that looks like the best match then
we could measuie a, b fiom oui diawing. A moie systematic appioach is to ist dene
what we mean by best match and then nd the line that best matches accoiding to oui
chosen ciiteiion.
A veiy common ciiteiion is the least-mean-squaie-t. To desciibe it, imagine we have
N data points, (x
1
, y
1
), , (x
N
, y
N
), and considei the line with coecients a and b.
Most data points (x
k
, y
k
) will then piobably not lie on the line y = ax +b, and one uses
E
k
=
1
2
_
ax
k
+b y
k
_
2
as a measuie foi the mismatch between the data point (x
k
, y
k
) and the line y = ax + b
(the factoi
1
2
makes foimulas latei on nicei). Adding all these eiiois we get the total
mean squaie eiioi
E = E
1
+ +E
N
.
lf we think of all the numbeis x
1
, . . . , x
N
, y
1
, . . . , y
N
as given constants (afei all, we
measuied them, so we shouldnt change them any moie'), then the total eiioi only de-
pends on the coecients a and b. lt is a measuie foi how well the line y = ax + b ts
oui data points, and the common method of linear regression consists in choosing the
coecients a and b so as to minimize this eiioi E.
y
=
a
x
+
b
(x
k
, y
k
)

ax
k
+b y
k

Figure 6. Which line best fits a set of data points?


Tis leads us to the pioblem of nding the ciitical points of the total eiioi E as a
function of a and b. We have to solve
E
a
= 0
E
b
= 0.
'Tis is called the Sushi Piinciple raw data is beer than cooked data.
8. PROBLEMS 93
Te total eiioi is the sum of the individual eiiois E
k
(a, b) so we get
E
a
=
E
1
a
+ +
E
N
a
,
E
b
=
E
1
b
+ +
E
N
b
.
Te individual eiiois have the following deiivatives
E
k
a
= x
k
_
ax
k
+b y
k
_
,
E
k
b
= ax
k
+b y
k
.
Adding all these deiivatives then leads to
E
a
=

x
k
_
ax
k
+b y
k
_
= (

x
2
k
)a + (

x
k
)b

x
k
y
k
and
E
b
=

_
ax
k
+b y
k
_
= (

x
k
)a +Nb

y
k
Heie

iepiesents summation ovei k = 1, , N, i.e.



x
k
y
k
= x
1
y
1
+ +x
N
y
N
,
etc.
lf (a, b) is a ciitical point then a and b must satisfy
(

x
2
k
)a + (

x
k
)b =

x
k
y
k
(

x
k
)a +Nb =

y
k
Tese aie two lineai equations foi the two unknowns a and b. Solving them leads to
a =
N

x
k
y
k


x
k

y
k
N

x
2
k

_
x
k
_
2
; b =

x
k

x
k
y
k
+

x
2
k

y
k
N

x
2
k

_
x
k
_
2
.
Tese aie the standaid foimulas foi the coecients a and b piovided by the method of
lineai iegiession. Most calculatois, and ceitainly all spieadsheets (like Excel) have these
foimulas piepiogiammed, so we only have to entei the data points (x
k
, y
k
) and push
the iight buuon to get a and b.
8. Problems
1. We are given N measurements x1, , xN
from some experiment, and, inspired by the
Linear Regression example, we decide to see
which number a best fits the data. We
define the error (or measure of misfit) for
each measurement to be
E
k
(a) =
1
2
(a x
k
)
2
and we look for the number a which mini-
mizes the total error
E(a) = E1(a) + +EN(a).
(a) Is this a problem about several variable
calculus, or about one variable calculus?
(b) Which number a do we find?
2. We have a series of data points (x
k
, y
k
),
and when we plot them we think we see a
convex curve rather than a straight line. In
fact it looks like a parabola, and so we set out
to find a quadratic function y = ax
2
+bx+c
that minimizes the error
E(a, b, c) = E1 + +EN,
with
E
k
(a, b, c) =
1
2
(
ax
2
k
+bx
k
+c y
k
)
2
.
(a) How many variables are there in this
problem?
(b) If (a, b, c) is a critical point of E(a, b, c)
then a, b, and c satisfy three linear equa-
tions. Find these equations (dont solve
them).
3. A measurement in a certain experiment
results in three numbers (x, y, z). The point
94 . MAXlMA AND MlNlMA
of the experiment is to see if there is a linear
relation of the form z = ax + by + c be-
tween the three measured quantities, and to
estimate the coeicients a, b, c.
Aer repeating the experiment N times
we have N data points (x
k
, y
k
, z
k
) (k =
1, . . . , N). We decide to choose a, b, c so as
to minimize the mean square error
E = E1 + +EN,
with
E
k
(a, b, c) =
1
2
(
ax
k
+by
k
+c z
k
)
2
.
Which (linear) equations will we get for a, b,
and c?
9. e Second Derivative Test
9.1. Review of the one-variable second derivative test and Taylors formula. Foi
a function y = f(x) of one vaiiable you can tell if a ciitical point a is a local maximum
oi minimum by looking at the sign of the second deiivative f

(a) of the function at that


point.
a b
f "(a)>0 f "(b)<0
lf f

(a) > 0 then the giaph of f is cuived upwaids and f has a local minimum at a,
if f

(a) < 0 then f has a local max. Tis section is about the analogous test foi ciitical
points of functions of two vaiiables.
One way to undeistand the second deiivative test is to look at the Tayloi expansion of
the function y = f(x). lf x = a is a ciitical point foi f, then
f(x) = f(a) +f

(a)(x a) +
1
2
f

(a)(x a)
2
+
Since a is a ciitical point of f we have f

(a) = 0, so that the Tayloi expansion ieduces to


(100) f(x) = f(a) +
1
2
f

(a)(x a)
2
+
lf we ignoie the iemaindei teim (the dots), then we nd that
f(x) f(a) +
1
2
f

(a)(x a)
2
.
Neai the ciitical point the giaph of y = f(x) is a appioximately a paiabola. lt is cuived
upwaids if f

(a) > 0, and downwaids if f

(a) < 0.
To apply the same ieasoning to a function of two (oi moie) vaiiables we need to know
the Tayloi expansion of such a function.
9.2. Taylors formula for a function of several variables. Te Tayloi expansion of
a function z = f(x, y) should give us an appioximation of f(a + x, b + y) in teims
involving poweis of x and y. Teie is a geneial foimula, but heie we only need the
second oidei teims, so well deiive those and stop theie.
Te tiick to nding the Tayloi expansion is to considei the function
(101) g(t) = f(a +tx, b +ty).
By denition
g(1) = f(a + x, b + y)
9. THE SECOND DERlVATlVE TEST 9
is the quantity we want to appioximate, and g(0) = f(a, b). Since g(t) is a function of
one vaiiable, we can apply Taylois foimula fiom Math 222 to it. We get
(102) g(t) = g(0) +g

(0)t +g

(0)
t
2
2!
+
Te dots contain the iemaindei teim, which we will ignoie. Now we set t = 1, and we
get
g(1) = g(0) +g

(0) +
1
2
g

(0) +
Te deiivatives of g can be computed with the chain iule
g

(t) =
df(a +tx, b +ty)
dt
(103)
= f
x
(a +tx, b +ty)
d(a +tx)
dt
+f(a +tx, b +ty)
d(b +ty)
dt
= f
x
(a +tx, b +ty)x +f
y
(a +tx, b +ty)y.
Te second deiivative is
(104) g

(t) = f
xx
(a +tx, b +ty)(x)
2
+ 2f
xy
(a +tx, b +ty)xy
+f
yy
(a +tx, b +ty)(y)
2
.
ln computing g

(t) we iun into teims involving f


xy
and teims with f
yx
. Because of
Claiiauts theoiem these aie the same, and combining them leads to the coecient 2 in
fiont of f
xy
above.
Seuing t = 0 in (103) and in (104) gives you expiessions foi g

(0) and g

(0), and by
substituting these in (102) we get the second order Taylor expansion of a function of
two variables:
(10) f(a + x, b + y) = f(a, b) +f
x
(a, b)x +f
y
(a, b)y
+
1
2
_
f
xx
(a, b)(x)
2
+ 2f
xy
(a, b)xy +f
yy
(a, b)(y)
2
_
+
Te ist thiee teims aie exactly the lineai appioximation (0) of the function that we
(a,b) x
y
(a+x,b+y)
Figure 7. x and y: Taylors formula lets us approximate a function z = f(x, y) at points
(x, y) = (a + x, b + y) close to (a, b). The expansion gives us f(x, y) = f(a + x, b + y)
as a function of x and y.
saw in Chaptei lll, 4.2. Te next teims in 10 aie
1
2
f
xx
(a, b)(x)
2
+f
xy
(a, b)xy +
1
2
f
yy
(a, b)(y)
2
.
9 . MAXlMA AND MlNlMA
Tese teims deteimine a quadiatic foim in the vaiiables x and y. Te quantities
1
2
f
xx
(a, b), etc. aie the coecients of the foim.
As always, the dots in the expansion (10) contain the iemaindei teim. By caiefully
including the one-vaiiable Lagiange iemaindei in the deiivation we can get a foimula foi
the iemaindei in (10). We will not do that, but it can be shown that the iemaindei is
o
_
(x)
2
+ (y)
2
_
, i.e. that it is small compaied to the othei teims in the expansion, at
least when x and y aie small.
9.3. Example compute the Taylor expansion of f(x, y) = sin 2xcos y at the
point (
1
6
,
1
6
). To nd the expansion we need to compute f, f
x
, f
y
, f
xx
, f
xy
, and f
yy
at
(
1
6
,
1
6
). Heie goes
f = sin 2xcos y =
3
4
f
x
= 2 cos 2xcos y =
1
2

3
f
y
= sin 2xsin y =
1
4

3
f
xx
= 4 sin 2xcos y = 3
f
xy
= 2 cos 2xsin y =
1
2
f
yy
= sin 2xcos y =
3
4
.
Substituting in the Tayloi expansion we get
f
_
1
6
+ x,
1
6
+ y
_
=
3
4
+
1
2

3 x
1
4

3 y +
1
2
_
3(x)
2
2
1
2
xy
3
4
(y)
2
_
+
=
3
4
+
1
2

3 x
1
4

3 y
3
2
(x)
2

1
2
xy
3
8
(y)
2
+
Note that the ist thiee teims in the expansion aie the lineai appioximation of the func-
tion
f
_
1
6
+ x,
1
6
+ y
_
=
3
4
+
1
2

3 x
1
4

3 y +
9.4. Another example the Taylor expansion of f(x, y) = x
3
+ y
3
3xy at the
point (1, 1). Te function f(x, y) = x
3
+y
3
3xy has the following deiivatives at (1, 1)
f = x
3
+y
3
3xy = 1
f
x
= 3x
2
3y = 0
f
y
= 3y
2
3x = 0
f
xx
= 6x = 6
f
xy
= 3 =3
f
yy
= 6y = 6
Te ist deiivatives vanish, so (1, 1) is a ciitical point of f. Te second oidei Tayloi
expansion of f at (1, 1) is
(10) f(1 + x, 1 + y) = 1 + 3(x)
2
3xy + 3(y)
2
+
Note that theie aie not ist oidei teims in this expansion because (1, 1) is a ciitical point
the coecients of the ist oidei teims aie both zeio.
To see what kind of ciitical point (1, 1) is, we have to analyze the second oidei, qua-
diatic, teims
(10) 3(x)
2
3xy + 3(y)
2
.
Tis expiession is a quadiatic foim in x and y, and by completing the squaie (see
Chaptei lll, 3) we nd that
3(x)
2
3xy + 3(y)
2
= 3
_
_
x
1
2
y
_
2
+
3
4
(y)
2
_
.
ln paiticulai, the quadiatic teims in the Tayloi expansion of f at the ciitical point aie
always positive, no mauei what x and y we choose (as long as they aie not both
zeio). lf we aie allowed to ignoie the iemaindei teim (the ), then this implies that
9. THE SECOND DERlVATlVE TEST 9
the function has a local minimum afei all, the Tayloi expansion (10) says that foi small
x and y the function value f(1 + x, 1 + y) is
f(1 + x, 1 + y) f(1, 1) + 3
_
x
1
2
y
_
2
+
9
4
(y)
2
.
Te second oidei teims aie all positive, so the Tayloi expansion tells us that
f(1 + x, 1 + y) f(1, 1),
at least foi small x and y. Te function theiefoie has a local minimum at (1, 1).
9.5. Example of a saddle point. Te same function f(x, y) = x
3
+ y
3
3xy has
anothei ciitical point, namely, the oiigin. By calculating the deiivatives at (0, 0) we nd
that the Tayloi expansion at the oiigin is
(108) f(x, y) = 3xy +
lgnoiing the iemaindei teims we see that neai the oiigin f(x, y) 3xy, which
suggests that f is negative when x and y aie both positive, oi when they aie both
negative, while f is positive when x and y have opposite signs.
Aibitiaiily close to the oiigin the function f theiefoie has both positive and negative
values, and theiefoie f has neithei a local maximum noi a local minimum at the oiigin.
ln fact the Tayloi expansion (108) suggests that the giaph of f should look like that of
the saddle function z = xy.
9.6. e two-variable second derivative test. Te last two examples essentially
show us how the second deiivative test foi functions of two vaiiables woiks. To explain
how it woiks in geneial, lets suppose a function f has a ciitical point at (a, b). Ten the
ist paitial deiivatives of f vanish at (a, b) and hence the Tayloi expansion has no ist
oidei teims. We get
(109) f(a + x, b + y) = f(a, b)+
1
2
_
f
xx
(a, b)(x)
2
+ 2f
xy
(a, b)xy +f
yy
(a, b)(y)
2
_
+
Tis is the two-vaiiable analog of equation (100). To see if (a, b) is a local maximum oi
minimum (oi something else), we have to see if the quadiatic teims in (109) aie always
negative, always positive, oi if they can have eithei sign, depending on the choice of x,
y.
Te piecise statement of the second deiivative test uses the teiminology intioduced in
Chaptei l, 3 and Figuie in that chaptei.
eorem (second derivative test). If (a, b) is a critical point of f(x, y), and if
Q(x, y) =
1
2
_
f
xx
(a, b)(x)
2
+ 2f
xy
(a, b)xy +f
yy
(a, b)(y)
2
_
is the quadratic part of the Taylor expansion of f at the critical point, then
If Q is positive denite then (a, b) is a local minimum of f,
If Q is negative denite then (a, b) is a local maximum of f,
If Q is indenite then (a, b) is a saddle point of f
If Q is semidenite the second derivative test is inconclusive.
98 . MAXlMA AND MlNlMA
When the foim Q is indenite, so that it can be factoied as
Q(x, y) = (kx +ly)(mx +ny),
then the level set of the function f containing the ciitical point (a, b) consists of two
cuives. One of these cuives is tangent to the line
kx +ly = 0, i.e. k(x a) +l(y b) = 0
while the othei is tangent to
mx +ny = 0, i.e. m(x a) +l(y b) = 0.
9.7. Example apply the second derivative test to the shy example. ln 2.3 and
4.4 we had found that the function f(x, y) = x
2
x
3
y
2
has two ciitical points,
one at the oiigin, and one at the point (
2
3
, 0). By caiefully looking at the zeio set of the
x
y
D
1
D
2
x
2
-
y
2
-
x
3
=
0
interior points
boundary points
function we discoveied that the oiigin is neithei a local maximum noi a local minimum,
and that the point (
2
3
, 0) is a local maximum. Te second deiivative test piovides a moie
systematic way of ieaching these conclusions. To apply the test we need to know the
second deiivatives of f at the ciitical points. Tey aie
(x, y) f
xx
(x, y) f
xy
(x, y) f
yy
(x, y)
(x, y) 2 6x 0 2
(0, 0) 2 0 2
(
2
3
, 0) 2 0 2
Teiefoie the second oidei Tayloi expansion of f at the oiigin is
f(x, y) = f(0, 0) +
1
2
_
2 (x)
2
+ 2 0 xy + (2)(y)
2
_
+
= (x)
2
(y)
2
+
= (x y)(x + y) +
Te quadiatic pait of the Tayloi expansion can be factoied, so this is the indenite case.
lt can be both positive and negative, depending on oui choice of x and y. Te second
deiivative test implies that the oiigin is a saddle point. lt also says that the zeio set of f
neai the oiigin consists of two cuives, whose tangents at the oiigin aie given by the two
equations
(110) x y = 0 and x + y = 0.
10. PROBLEMS 99
ln this case the point (a, b) is the oiigin, so x = x a = x and y = y b = y, and
the two tangents aie the lines y = x.
Te second oidei Tayloi expansion at the othei ciitical point (
2
3
, 0) is given by
(111) f(
2
3
+ x, y) = f(
2
3
, 0) (x)
2
(y)
2
+
Tis time we see that the second oidei teims of the Tayloi expansion aie negative denite.
Te second deiivative test theiefoie says that we have a local maximum at (
2
3
, 0).
10. Problems
1. [for discussion] Are x in 9.4 and 9.5
the same?
Are the x in the equations (110) and in
(111) of the second derivative test example
the same? Explain what they stand for.
2. Compute the second order Taylor expan-
sion of the following functions at the indi-
cated points:
[In this problem you are asked to find
Taylor expansions of functions at various
points. Since these points are not necessar-
ily critical points, the expansions you find
will generally have first and second oder
terms. In the expansions you will compute
when you use the second derivative test later
on, there will be no first order terms.]
(a) f(x, y) =
(
1 x +xy
)
2
at (0, 0)
(b) f(x, y) =
(
1 x +xy
)
2
at (1, 1)
(c) f(x, y) = e
xy
2
at (0, 0)
(d) f(x, y) = e
xy
2
at (1, 1)
(e) f(x, y) =
x
1 y
at (0, 0)
(f) f(x, y) =
x
1 +y
at (1, 0)
3. Factor, or complete the square in the fol-
lowing quadratic forms, drawtheir zero sets,
and determine if they are positive definite,
negative definite, indefinite or degenerate.
(a) Q(x, y) = x
2
+ 3xy +y
2
(b) Q(x, y) = x
2
+xy +y
2
(c) Q(x, y) = 2x
2
+ 3xy 4y
2
(d) Q(x, y) = 2x
2
+ 3xy 5y
2
(e) Q(x, y) = (x)
2
+ (y)
2
(f) Q(x, y) = (x)
2
3(y)
2
(g) Q(x, y) = xy
(h) Q(x, y) = xy 2(y)
2
4. If a is a constant, then for which values
of a is the form Q(x, y) = x
2
+ 2axy + y
2
positive/negative definite, indefinite, or de-
generate?
5. Find all critical points of the following
functions (you did many of these in problem
6.1). Apply the second derivative test to all
critical points you find.
(a) f(x, y) = x
2
+ 4y
2
2x + 8y 1
(b) f(x, y) = x
2
y
2
+ 6x 10y + 2
(c) f(x, y) = x
2
+ 4xy +y
2
6y + 1
(d) f(x, y) = x
2
xy +2y
2
5x +6y 9
(e) f(x, y) = y
2
18x
2
+x
4
(f) f(x, y) = y
4
4y
2
18x
2
+x
4
(g) f(x, y) = 9 + 4x y 2x
2
3y
2
(h) f(x, y) = xy(4 x 2y)
(i) f(x, y) = x(x y)(x 1)
(j) f(x, y) = (x y)(xy 4)
(k) f(x, y) = y
2
+ cos x
(l) f(x, y) = x
2
y
1
3
y
3
(m) f(x, y) = (x y
2
)(x 1)
(n) f(x, y) = (x y)(xy 4)
(o) f(x, y) = x
2
(p) f(x, y) = x
2
y
4
(q) f(x, y) = x
2
+y
4
(r) f(x, y) = x
2
y
6. (a) Draw the zero set of the function
f(x, y) = sin(x) sin(y). (b) Where is the
function f positive? Find as many critical
points as you can without computing fx or
fy.
(c) Find all critical points of f(x, y). Which
are local minima or local maxima?
100 . MAXlMA AND MlNlMA
7. Find all critical points of the following
functions, and apply the second derivative
test to the points you find.
(a) f(x, y) = x
2
+y
2

1
2
xy
2

(b) f(x, y) = x
2
+y
2
x
2
y
2
(c) f(x, y) = x + 2y xy
2

(d) f(x, y) = 8x
4
+y
4
xy
2
8. Suppose that f(x, y) = x
2
+ y
2
+ kxy.
Find and classify the critical points, and dis-
cuss how they change when k takes on dif-
ferent values.
9. Consider the function
f(x, y) = x
3
3xy
2
.
The graph of this function is known as the
Monkey Saddle.
(a) Show that (0, 0) is the only critical point
of f.
(b) Show that the second derivative test is
inconclusive for f.
(c) Draw the zero set of f, and indicate
where f > 0 and where f < 0.
(d) What kind of critical point is (0, 0)?
10. Consider the function
f(x, y) = x
3
x
2
y.
(a) Drawthe zeroset of f and indicate where
f(x, y) is positive, and where f(x, y) is neg-
ative.
(b) Find all the critical points of the function.
(c) Does the second derivative test apply to
any of the critical points of f?
(d) Use the sign-diagram you made in part
(a) to decide which critical points are local
maxima or minima.
11. Second derivative test for more than two variables
Te ideas that lead to the second deiivative test foi functions of two vaiiables also
woik when we have a function with moie vaiiables. Howevei, the second deiivative test
foi functions of moie than two vaiiables is beyond the scope of Math 234, and this shoit
section tiies to explain why.
11.1. e second order Taylor expansion. lf z = f(x
1
, x
2
, , x
n
) is a function of
n vaiiables, then its Tayloi expansion of oidei two at some point (a
1
, a
2
, , a
n
) tuins
out to be
f(a
1
+ x
1
, , a
n
+ x
n
) =
f(a
1
, , a
n
)+f
x1
x
1
+ +f
xn
x
n
+
1
2
_
f
x1x1
(x
1
)
2
+ +f
x1xn
x
1
x
n
+f
x2x1
x
2
x
1
+ +f
x2xn
x
2
x
n
.
.
.
+f
xnx1
x
n
x
1
+ +f
xnxn
(x
n
)
2
_
+
wheie the paitial deiivatives f
xi
and f
xixj
aie to be evaluated at the point (a
1
, , a
n
).
Te same tiick involving the function g(t) that was used in 9.2 to deiive the two-
vaiiable Tayloi expansion woiks without modication.
12. OPTlMlZATlON WlTH CONSTRAlNTS AND THE METHOD OF LAGRANGE MULTlPLlERS 101
lf (a
1
, , a
n
) is a ciitical point then f
x1
= f
x2
= = f
xn
= 0, so the lineai teims
aie absent, and the function is desciibed by the quadiatic teims of the Tayloi expansion
f(a
1
+ x
1
, , a
n
+ x
n
) =
f(a
1
, , a
n
) +
1
2
_
f
x1x1
(x
1
)
2
+ +f
x1xn
x
1
x
n
+f
x2x1
x
2
x
1
+ +f
x2xn
x
2
x
n
.
.
.
+f
xnx1
x
n
x
1
+ +f
xnxn
(x
n
)
2
_
+
Just as in the two-vaiiable case we could now tiy to see if the quadiatic teims aie positive
denite oi negative denite by completing squaies. Te pioceduie is howevei much moie
complicated, and best undeistood in teims of eigenvalues of matrices, a subject which is
explained in couises on lineai algebia oi matiix algebia (Math 320, 340, oi 341). Teiefoie,
we will only use the second deiivative test foi functions of two vaiiables in this couise.
12. Optimization with constraints and the method of Lagrange multipliers
ln many optimization pioblems we want to nd the maximal oi minimal value of a
function f(x, y) wheie (x, y) can be any point satisfying a ceitain constraint
(112) g(x, y) = C.
Tus the domain D of the function we want to minimize consists of all points (x, y) that
satisfy the equation g(x, y) = C it is a level set of g.
12.1. Solution by elimination or parametrization. One appioach to minimization
pioblems with a constiaint is to eliminate one vaiiable. lf we aie asked to nd the
minimal value that f(x, y) can have if (x, y) must satisfy the constiaint g(x, y) = C,
then we ist tiy to solve the constiaint equation foi one of the vaiiables, say, foi y
g(x, y) = C y = h(x).
Now the only (x, y) that we have to considei aie points of the foim (x, h(x)), so the
old minimization pioblem is equivalent to a new pioblem nd the minimal value of
F(x) = f(x, h(x)), wheie theie aie no constiaints on x. Tis new pioblem is a one
vaiiable pioblem of the kind we leained to solve in Math 221.
12.2. Example whi rectangle with perimeter 1 has the largest area? Tis is
anothei pioblem, like nding the tangent to the paiabola y = x
2
, that appeais in almost
eveiy ist semestei calculus couise. We iecall its solution.
lf the sides of the iectangle aie x and y, then its aiea is xy and its peiimetei is 2(x+y).
Hence the function we want to maximize is f(x, y) = xy and the constiaint is
g(x, y) = 2(x +y) = 1.
Solving the constiaint foi y tells you that y =
1
2
x, so we want to maximize the function x
y y
x
F(x) = f(x,
1
2
x) = x(
1
2
x). Te only iemaining constiaint is that x cannot be
negative, and that y =
1
2
x also cannot be negative. Tus we want to maximize F(x) =
x(
1
2
x) ovei all x in the inteival 0 x
1
2
.
102 . MAXlMA AND MlNlMA
12.3. Example maximize x + 2y over the unit circle. We aie asked to nd the
maximal value of f(x, y) = x + 2y wheie (x, y) is allowed to be any point that satises
the constiaint g(x, y) = x
2
+ y
2
= 1. lf we tiy to solve foi y we nd that theie aie two
solutions, y =

1 x
2
, and so the function F(x) = x + 2y = x 2

1 x
2
is not
ieally a function at all. ln this case we can still solve the pioblem by noting that any point
on the unit ciicle can be wiiuen as (x, y) = (cos , sin ) foi some angle , and thus we
have to maximize the function
F() = f(cos , sin ) = cos + 2 sin .
Heie theie aie no constiaints on , and we again have a ist semestei calculus pioblem.
12.4. Solution by Lagrange multipliers. ln both examples above we weie lucky be-
cause we could eithei solve the constiaint equation oi we could paiametiize all possible
points that satisfy the constiaint. Teie is a method due to Joseph-Louis Lagiange (known
fiom the iemaindei teim) that does not iequiie this kind of luck. His method is based on
the following obseivation (see Figuie 8).
f=0.69
f=0.68
f=0.66
f=0.65
f=0.67
g(x,y) = C
A
B
g
f
f
g
Figure 8. Lagrange multipliers: if, at some point like B on the constraint set the gradients of f
and g are not parallel, then we can increase f by moving along the constraint set in the direction
of
#
f. At a point (such as A) where the function f reaches a maximum, the gradients
#
f and
#
g must be parallel.
Let B = (x, y) be a point on the constiaint set as in the guie. Assume that
#
g =
#
0
at B, then neai B the lmplicit Function Teoiem says that the constiaint set g(x, y) = C
is a cuive, and that its tangent is peipendiculai to
#
g(B).
lf
#
f(B) is not peipendiculai to the constiaint set at B, then it piovides us a diiection
along the constiaint set in which f will inciease (see Figuie 8). Teiefoie f does not have
a maximum at B. lt follows that at a maximum of f on the constiaint set g(x, y) = C
the giadient
#
f(B) must be peipendiculai to the constiaint set, and hence it must be
paiallel to
#
g(B). Since one vectoi is paiallel to anothei if it is a multiple of the othei
vectoi, we have found the following fact.
12.5. eorem (Lagrange multipliers). If the function z = f(x, y) aains its largest
value among all points that satisfy the constraint g(x, y) = C at the point (a, b), and if
(113)
#
g(a, b) = 0,
12. OPTlMlZATlON WlTH CONSTRAlNTS AND THE METHOD OF LAGRANGE MULTlPLlERS 103
then the point (a, b) satises the Lagrange Multiplier equation,
(114)
#
f(a, b) =
#
g(a, b)
Te numbei is called the Lagrange multiplier, and it is one of the unknowns in the
equations we must solve when we use Lagianges method.
12.6. Example. We again tiy to nd the laigest iectangle with peiimetei 1, as in
example 12.2.
Te pioblem is to maximize f(x, y) = xy with constiaint g(x, y) = 2x + 2y = 1. We
compute the giadients
#
f =
_
y
x
_
,
#
g =
_
2
2
_
,
Te giadient of g nevei vanishes, i.e.
#
g(x, y) =
#
0 foi all (x, y), so Lagiange tells us
that at any minimum oi maximum the following equations hold
f
x
= g
x
, i.e. y = 2
f
y
= g
y
, i.e. x = 2
g(x, y) = C, i.e. 2x + 2y = 1.
Te ist two equations come fiom
#
f =
#
g, and the last equation is the constiaint. We
have thiee equations, and we also have thiee unknowns x, y and the Lagiange multipliei
.
ln this case it is easy to solve the equations the ist two say that both y and x equal
2, so in paiticulai, they equal each othei y = x. Tis alieady tells us that the solution
is a squaie' To complete the pioblem we must still solve foi x, y, . Since x = y the
constiaint implies 4x = 1, so x = y =
1
4
. Finally, eithei of the ist two equations
piovides =
1
2
x =
1
2
y =
1
8
.
What is the meaning of ? ln this example you see that we ist found the solution
(x, y), and then computed . Te multipliei is the iatio between the lengths of the
giadients of f and g at the maximum, and is usually of no inteiest. Nonetheless, when
using Lagianges method, you must always also nd , oi at least make suie that a exists
foi the x and y you have found.
Did we nd a maximum or a minimum? Lagianges method does not tell us if we
have a maximum oi a minimum, and we will have to use dieient methods to guie this
out. Teie does exist a second deiivative test foi constiained minimization pioblems, but
it falls outside the scope of this couise.
12.7. A three variable example. Find the largest value of x+y +z on the sphere with
equation x
2
+y
2
+z
2
= 1.
Solution: We must maximize f(x, y, z) = x + y + z with constiaint g(x, y, z) =
x
2
+y
2
+z
2
= 1.
Lagianges method says that the minimumand maximumeithei occui at a point (x
0
, y
0
, z
0
)
with
#
g(x
0
, y
0
, z
0
) =
#
0, oi else at a point that satises Lagianges equations. Te gia-
dient of g is
#
g(x, y, z) =
_
_
2x
2y
2z
_
_
,
104 . MAXlMA AND MlNlMA
and the only point wheie
#
g =
#
0 is at the oiigin. Te oiigin does not satisfy the
constiaint g(x, y, z) = 1, so we can iule out the possibility of the maximum oi minimum
occuiiing at a point with
#
g =
#
0.
Tis leads us to considei the Lagiange multipliei equations, which aie
1 = 2x (f
x
= g
x
)
1 = 2y (f
y
= g
y
)
1 = 2z (f
z
= g
z
)
x
2
+y
2
+z
2
= 1 (g(x, y, z) = C)
Solve the ist thiee equations foi x, y, z and substitute the iesult in the constiaint, and
we nd
1
4
2
+
1
4
2
+
1
4
2
= 1
3
4
2
= 1 =
1
2

3.
We theiefoie nd two points on the spheie,
(x, y, z) =
_
1
3

3,
1
3

3,
1
3

3
_
and (x, y, z) =
_

1
3

3,
1
3

3,
1
3

3
_
By computing the function values we nd that the ist point maximizes x + y + z, and
the second minimizes x +y +z.
13. Problems
1. Minimize xy subject to the constraint
x
2
+
1
4
y
2
= 1.
Draw the constraint set.
2. A six-sided rectangular box is to hold 1/2
cubic meter. Which shape should the box be
to minimize surface area?
(a) Find the solution without using La-
granges method.
(b) Use Lagrange multipliers to solve this
problem.
3. Using the methods of this section, find
the shortest distance from the origin to the
plane x+y+z = 10. (suggestion: instead of
minimizing the distance, you can also mini-
mize the square of the distance.)
4. Use Lagrange multipliers to find the
largest and smallest values of f(x, y) = x
under the constraint g(x, y) = y
2
x
3
+
x
4
= 0.
5. (a) Using Lagrange multipliers, find the
shortest distance from the point (2, 1, 4) to
the plane 2x y + 3z = 1.
(b) Using Lagrange multipliers, find the
shortest distance from the point (x0, y0, z0)
to the plane ax +by +cz = d.
6. (a) Find the shortest distance from the
point (0, b) to the parabola y = x
2
, using
Lagrange multipliers.
(b) Find the shortest distance fromthe point
(0, 0, b) to the paraboloid z = x
2
+y
2
.
(c) Find the shortest distance from the point
(0, 0, b) to the paraboloid z = x
2
+
1
4
y
2
.
7. Find the volume of the largest rectangu-
lar box with edges parallel to the axes that
can be inscribed in the ellipsoid
2x
2
+ 72y
2
+ 18z
2
= 288.
8. A six-sided rectangular box is to hold 1/2
cubic meter; what shape should the box be
to minimize surface area?
9. A circular cone has height H, and its
base has radius R. If the volume of the
cone is fixed, then which ratio of radius to
height (R : H) minimizes the surface area
of the cone? (The area of the cone is A =
R

R
2
+H
2
, its volume is V =
1
3
R
2
H,
and instead of minmizing the area you could
also minimize the square of the area.)
10. The post oice will accept packages
whose combined length and girth are at
most 130 inches (girth is the maximum dis-
tance around the package perpendicular to
13. PROBLEMS 10
the length). What is the largest volume that
can be sent in a rectangular box?
11. The boom of a rectangular box costs
twice as much per unit area as the sides and
top. Find the shape for a given volume that
will minimize cost.
12. Find all points on the surface
xy z
2
+ 1 = 0
that are closest to the origin.
13. The material for the boom of an aquar-
ium costs half as much as the high strength
glass for the four sides. Find the shape of the
cheapest aquariumthat hold a given volume
V .
14. The plane x y + z = 2 intersects the
cylinder x
2
+ y
2
= 4 in an ellipse. Find
the points on the ellipse closest to and far-
thest from the origin. (Hint: on the plane
you always have z = 2 x +y, so you can
eliminate z and make this a problem about
functions of (x, y) only.)
CHAPTER
Integrals
1. Ways of Integrating
ln this chaptei we will see seveial dieient ways of integiating functions of seveial
vaiiables. Befoie intioducing them one by one, we spend this section ieviewing how
integiation was dened in ist semestei calculus and outlining the geneial featuies that
all dieient ways of integiating have in common.
1.1. e one variable integral. To begin, let us quickly iecall how the integial of a
function of one vaiiable is dened. Given a function y = f(x) and an inteival [a, b], we
choose a partition of the inteival [a, b], which means that
we split the inteival [a, b] into shoitei inteivals [x
0
, x
1
], [x
1
, x
2
], , [x
N1
, x
N
],
wheie a = x
0
< x
1
< < x
N
= b,
and we choose one sample point
k
fiom each inteival [x
k1
, x
k
].
Fiom these ingiedients we compute the Riemann sum
R = f(
1
)x
1
+ +f(
N
)x
N
=
N

k=1
f(
k
)x
k
wheie x
k
= x
k
x
k1
is the length of the k
th
inteival.
a = x
0
x
1
x
2
x
3
x
4
x
5
b = x
6
a b
Figure 1. Riemann sums for

b
a
f(x)dx with one partition on the le, and a finer partition on the
right. The dashed lines in the figure on the le indicate where the sample points
k
were chosen.
Foi most functions y = f(x) it is tiue that upon making the inteivals [x
k1
, x
k
]
shoitei (and hence choosing moie paitition inteivals), the iesulting Riemann sums ap-
pioach a limiting value. When this happens we call the limiting value of the Riemann
sums the integral of the function f(x) over the interval [a, b]

b
a
f(x)dx = lim
as the partition
gets finer
f(
1
)x
1
+ +f(
N
)x
N
.
10
108 . lNTEGRALS
Te individual teims in the Riemann sum aie aieas of the naiiow iectangles in the guie.
Added togethei they appioximate the aiea of the iegion undei the giaph, so that the
integial is the aiea between the giaph of y = f(x) and the x-axis (at least in the case that
f is a positive function, so that its giaph lies above the x-axis.)
A note about rigor. Oui quick desciiption of the single vaiiable integial is lacking in
mathematical piecision. lt is based on a belief that we know what aiea is. ln the late
19
th
and eaily 20
th
centuiies many examples of geometiic guies weie found in which
aiea computations give unexpected and counteiintuitive iesults. Teiefoie one cannot
base a theoiy on oui intuitive idea of aiea, and instead the integial, dened as limit of
Riemann sums is used a way of giving a iigoious denition of the notion of aiea. Foi
a piopei tieatment of these issues the student is iefeiied to a moie advanced couise on
Real Analysis (e.g. Math 421 oi 21).
1.2. Generalizing the one variable integral. While theie is essentially only one kind
of integial in single vaiiable calculus, theie aie many dieient ways of integiating func-
tions of seveial vaiiables. All these dieient notions of integial t the following bioad
desciiption.
ln any kind of integial we have these ingiedients
a domain. Depending on the kind of integial, this can be a iegion in the plane,
a iegion in space, a plane cuive, a space cuive, oi even some suiface in thiee
dimensional space.
a function that is dened on the domain
a way of measuiing the size of pieces of the domain
To dene the integial we paitition the iegion, i.e. we divide it into lots of liule pieces.
Given any such paitition of the iegion into smallei pieces, we then foim the following
Riemann sum

pieces in the
partition
_
f at sample point
in piece #k
_

_
Size of piece #k
_
Tis gives us a numbei foi each way of paititioning the iegion. As we make the paitition
nei, i.e. as we choose moie, smallei, pieces, the Riemann sums tend to get closei to one
paiticulai numbei, which is called the integial of the function. ln shoit, the integial is
the limit of the Riemann sums we nd as we take nei and nei paititions

some region
f(x) dx = lim
as the
partition
gets finer

pieces in the
partition
_
f at sample point
in piece #k
_

_
Size of piece #k
_
Depending on what kind of function we have, and what kind of iegion the function is
dened on, and also howwe decide to measuie the size of the small pieces in the paitition,
this piocess can lead to many dieient kinds of integials. Te integials we will meet in
this chaptei aie double integrals and triple integrals, in the next chaptei on vectoi
calculus we will also see line integrals and surface integrals. See Table 1.
2. Double Integrals
Let z = f(x, y) be a function of two vaiiables dened on some iegion D in the plane.
Te double integral of f over D is dened in teims of Riemann sums, following the
geneial scheme desciibed in the pievious section. To foim a Riemann sum we ist need a
2. DOUBLE lNTEGRALS 109
Kind of integral
Domain
Typical piece of
partition
Size of piece
Good old 221
Integral

b
a
f(x) dx
interval
a x b
small subinterval
(x
k1
, x
k
)
length of subinterval
x
k
= x
k
x
k1
Multiple integral

D
f(x, y)dA
region in
the plane
tiny sub domain
area A of
tiny sub domain
Multiple integral

D
f(x, y, z)dV
region
in space
tiny sub domain
volume V of
tiny sub domain
Line integral

C
f(x, y) ds
curve in
the plane
short sub arc
of the curve
length s of
the sub arc
Line integral

C
f(x, y, z) ds
curve
in space
short sub arc
of curve
length s of
the sub arc
Surface integral

S
f(x, y, z) dA
surface
in space
small patch
on the surface
area A of
the patch
Table 1. A list of the dierent kinds of integrals that we will encounter in math 234.
paitition of the iegion Dinto smallei iegions D
1
, , D
N
, and we need to choose a sample
point (x
k
, y
k
) fiom each iegion D
k
. lf A
k
is the aiea of iegion D
k
, then the Riemann
sum coiiesponding to the paitition D
1
, , D
N
and the choice of sample points (x
1
, y
1
),
, (x
N
, y
N
) is
(11) R = f(x
1
, y
1
)A
1
+ +f(x
N
, y
N
)A
N
=
N

k=1
f(x
k
, y
k
) A
k
.
lf the paitition is suciently ne then this Riemann sum will in many cases be close to
one paiticulai numbei, which we will call the integial of the function f ovei the iegion
D. Tus
(11)

D
f(x, y) dA = lim
as the partition
gets finer&finer
N

k=1
f(x
k
, y
k
) A
k
.
Figure 2. On the le: a region in the plane with some partition. Many pieces of the partition
are rectangles. This is a common choice, but the pieces dont have to be rectangles: here the pieces
that touch the boundary of the domain have at least one curved edge. On the right: the same
region with two finer partitions.
110 . lNTEGRALS
To make this moie piecise one has to iesoit to s and s, which iesults in the following
denition.
2.1. Denition. If for every > 0 there is a > 0 such that the Riemann sum corre-
sponding to any partition of the region D into smaller pieces D
1
, , D
N
, whose pieces have
diameter no more than satises

I
N

k=1
f(x
k
, y
k
) A
k

<
then we say that

D
f(x, y) dA = I.
On one hand it can be shown in many cases that that the integial of a function ex-
ists accoiding to the above denition. On the othei hand the - denition is neithei a
piactical method of computing such integials, noi does it piovide an easy intuitive un-
deistanding of the piopeities of the integial. Teiefoie, we will stick to the less piecise
denition (11) in this couise.
2.2. e integral is the volume under the graph, when f 0. lf the function f is
positive, then its giaph lies above the xy-plane, and theie is a simple inteipietation of the
integial, namely

D
f(x, y) dA = Volume of R,
wheie R is the iegion undei the giaph of f above the domain D in symbols,
(11) R =
_
(x, y, z) : (x, y) lies in D, and 0 z f(x, y)
_
.
To see why this is so, imagine that we have a positive function z = f(x, y) dened on
some iegion D in the xy-plane, and let us tiy to compute the integial

D
f(x, y)dA
geometiically. To compute the integial we begin by nely paititioning the iegion D
into smallei iegions D
1
, D
2
, , D
N
(see Figuie 3 on the lef wheie the small pieces weie
themselves chosen to be iectangles). We also choose one sample point (x
k
, y
k
) in each
iegion D
k
. Te Riemann sum we get this way is
R = f(x
1
, y
1
)A
1
+ +f(x
N
, y
N
)A
N
wheie A
k
is the aiea of D
k
. Te k
th
teim, f(x
k
, y
k
)A
k
, is the volume of a block
whose base is D
k
and whose top is some point on the giaph of the function above the
iegion D
k
. Tis volume is almost, but usually not exactly the same as the volume of the
iegion between the giaph of the function and the small iegion D
k
in the xy-plane. Te
volume f(x
k
, y
k
)A
k
of the block above D
k
is not exactly the same as the volume of the
iegion undei the giaph because the top of the block is a piece of a hoiizontal plane while
the giaph of f will usually have a slope (see Figuie 3).
Te total Riemann sum is theiefoie the sum of the volumes of such blocks, (see Fig-
uie 4) and this will appioximate the volume between the giaph of f and the domain of
integiation D. Te nei the paitition, the beuei the appioximation and so we can con-
clude' that the limit of the Riemann sums is the volume undei the giaph, i.e. the volume
of the iegion R dened in (11).
'As piomised befoie, this is not a veiy piecise pioof, a pioof that the limit of Riemann sums exists quickly
lead us to & aiguments.
2. DOUBLE lNTEGRALS 111
f(x
k
, y
k
)
x
z
y
A
k
D
k
a b
c
d
(x
k
, y
k
)
Figure 3. On the le: the domain of the function f partitioned into 6 5 pieces, each with
the same width x and height y. To form a Riemann sum we have to choose one sample point
(x
k
, y
k
) in each piece D
k
of the partition. Below we will always choose the upper-right-hand
corner of the rectangle to be the sample point. Onthe right: Any piece in the partition corresponds
to a term in the Riemann sum of the form f(x
k
, y
k
)A
k
. This is the volume of a block of height
f(x
k
, y
k
), and base D
k
, which is approximately the volume of the region under the graph of f and
above the piece D
k
. Adding all these volumes together we see that a Riemann sum approximates
the total volume between the graph and the region D.
x
y
z
N=4
M=3
x
y
z
N=8
M=6
Figure 4. Approximating the region under the graph of z = f(x, y) from Figure 3 by vertical
blocks. The base of each block is a rectangle in a partition of the domain of f. As we choose finer
and finer partitions, the region occupied by the vertical blocks gets closer to the region under the
graph of f.
2.3. How to compute a double integral. So fai, we have a denition foi the double
integial

D
f(x, y)dA, and an inteipietation of the integial as volume undei the giaph
of f. What is missing is a method of actually computing the integial. ln this section well
see how one can compute a double integial by doing two one-vaiiable integials.
Let us take anothei look at the integial of the function f ovei the iectangle
D =
_
(x, y) : a x b, c y d
_
,
fiom the pievious section.
112 . lNTEGRALS
We again paitition Dinto smallei iectangles, as in Figuie 3, but instead of just counting
them and aibitiaiily numbeiing the pieces 1, 2, , N, we can use the fact that the smallei
iectangles appeai in iows and columns. lf we take N iectangles in the x diiection, and
M in the y diiection, then the smallei iectangles will measuie x by y, wheie
x =
b a
N
, y =
d c
M
.
We let (x
k
, y
l
) be the uppei-iight-hand coinei of the iectangle in the k
th
column fiom
the lef, and the l
th
iow fiom below. Ten
(118) x
k
= a +kx, y
l
= c +ly.
Te Riemann sum coiiesponding to this paitition and choice of sample points (x
k
, y
l
) is
(119)
R =

f(x
k
, y
l
)xy
= f(x
1
, y
1
)xy + + f(x
N
, y
1
)xy+
f(x
1
, y
2
)xy + + f(x
N
, y
2
)xy+
.
.
.
f(x
1
, y
M
)xy + + f(x
N
, y
M
)xy
Since we aie choosing the uppei-iight-hand coinei of each iectangle as sample point
in that iectangle, the sample point foi the iectangle at the top-iight is (x
N
, y
M
). (See
Figuie 3 on the lef.) Teiefoie, in this summation k can have any value with 1 k N
and l can be any integei with 1 l M.
Te teim coiiesponding to iectangle (k, l) iepiesents the volume of a block whose
height is f(x
k
, y
l
) and whose base is a x y iectangle. Togethei these blocks ap-
pioximate the iegion between the giaph of the function and the xy-plane.
Considei the teims on the k
th
iow in equation (119), afei factoiing out y we get
iow -k of (119) = y
_
f(x
1
, y
k
)x +f(x
2
, y
k
)x + +f(x
N
, y
k
)x
_
.
x
y
z
y = c
y = d
x = a
x = b
Figure 5. This picture shows the blocks corresponding to all those terms in the Riemann sum R
from equation (119) in which y = y
k
. These terms
{
f(x1, y
k
)x + +f(xN, y
k
)x
}
y give
you the total volume of one row of matchsticks from Figure 4. In this sumy is frozen at the value
y = y
k
, so we can think of f(x1, y
k
)x + +f(xN, y
k
)x as a Riemann sum for the integral

b
a
f(x, y
k
) dx.
2. DOUBLE lNTEGRALS 113
Note that in this sum the function is always evaluated at the same value of y, namely y
k
.
Te sum between biaces { } is actually a Riemann sum foi the one-vaiiable integial
I =

b
a
f(x, y
k
)dx
in which we tieat f(x, y
k
) as a function of x only and considei the vaiiable y to be fiozen
at y = y
k
. Te value of this integial depends on the value at which y is fiozen, so it is
beuei to wiite
I(y) =

b
a
f(x, y)dx.
With this notation we nd that
iow -k of (119) y
_
I(y
k
)
_
= I(y
k
)y.
To nd the value of the Riemann sumthat appioximates the double integial

D
f(x, y)dA
we add the iows in (119) and nd
R I(y
1
)y +I(y
2
)y + +I(y
M
)y.
Te sumon the iight is again a Riemann sumfoi a one vaiiable integial, namely,

d
c
I(y)dy.
Teiefoie we nd that
R

d
c
I(y)dy
lf we now take the limit in which we let the size of the pieces in the paitition go to zeio,
then it can be shown (with quite a bit of eoit) that the appioximation above gets beuei,
and that one has

D
f(x, y)dA =

d
c
I(y)dy.
Teiefoie, iemembeiing the denition of I(y), we have found the following method of
computing a double integial.
2.4. eorem. If f(x, y) is a function dened on a rectangle
D = {(x, y) : a x b, c y d} ,
then the double integral of f over D is given by
(120)

D
f(x, y)dA =

d
c
_

b
a
f(x, y)dx
_
dy.
One can also rst integrate with respect to y and then x, so that
(121)

D
f(x, y)dA =

b
a
_

d
c
f(x, y)dy
_
dx.
Te second way of computing the double integial

D
f(x, y) dA, i.e. equation (121),
follows by the same ieasoning that led us to (120), except in (119) one gioups the teims
by columns iathei than iows.
To compute the iight hand side in this equation we have to compute two one-vaiiable
integials. Te expiession

d
c
_

b
a
f(x, y)dx
_
dy =

d
c

b
a
f(x, y) dxdy
is called an iterated integral.
114 . lNTEGRALS
Te two integials that appeai in an iteiated integial aie ofen called innei and outei
integial

d
c
_

b
a
f(x, y)dx
. .
innei integial
_
dy
. .
outei integial
.
2.5. Example: the volume under the graph of the paraboloid z = x
2
+y
2
above the
square Q = {(x, y) : 0 x 1, 0 y 1}. Te double integial we have to compute
is
Volume =

Q
_
x
2
+y
2
_
dA
and to compute it we wiite it as an iteiated integial

Q
_
x
2
+y
2
_
dA =

1
0
_

1
0
(x
2
+y
2
)dx
_
dy.
ln the innei integial the vaiiable y is fiozen, so to compute the innei integial, we simply
tieat y as a constant, and integiate with iespect to x. We get

1
0
(x
2
+y
2
)dx =
_
1
3
x
3
+y
2
x

1
x=0
=
1
3
+y
2
.
(Tis is I(y) in the notation of the pievious section.)
x
y
z
a
b
x
y
z
Figure 6. The graph of z = x
2
+ y
2
above the unit square Q on the le, and rectangle {(x, y) :
0 x a and 0 y b}, on the right, together with the surrounding block. What fraction of
the volume of the block lies below the graph?
2. DOUBLE lNTEGRALS 11
To get the double integial we must still do the outei integial

Q
_
x
2
+y
2
_
dA =

1
0
_

1
0
(x
2
+y
2
)dx
_
dy
=

1
0
_
1
3
+y
2
_
dy
=
_
1
3
y +
1
3
y
3

1
0
=
1
3
+
1
3
=
2
3
.
Since the suiiounding block (Figuie ) is a 1 1 2 block, its volume is 2, and the iegion
undei the giaph occupies exactly one thiid of the whole block.
To compute the volume of the iegion undei the giaph of the same function above the
iectangle {(x, y) : 0 x a, 0 y b} one can compute eithei of the iteiated
integials

a
0

b
0
_
x
2
+y
2
_
dy dx oi

b
0

a
0
_
x
2
+y
2
_
dx dy.
2.6. Double integrals when the domain is not a rectangle. We have seen how to
compute a double integial when the domain is a iectangle. Te ieasoning that led us fiom
a double integial to an iteiated integial also woiks foi non iectangulai domains, piovided
they aie not too complicated. Suppose we want to compute

D
f(x, y)dA wheie the
domain D is the iegion caught between the giaphs of two functions
D =
_
(x, y) : a x b, f(x) y g(x)
_
.
We again paitition the iegion by cuuing it along many veitical lines x = x
1
, x = x
2
,
, x = x
N
, and many hoiizontal lines y = y
1
, , y = y
M
. Most of the pieces of the
paitition will be iectangles, but those that oveilap with the boundaiy of the iegion D
may have cuived edges. See Figuies and 8.
T
o
p
: y
=
d
(x
)
D
x
k
x
k-1
e strip c(x) yd(x), x
k-1
xx
k
a b
B
o

o
m
:
y
=
c
(
x
)
Figure 7. The region between the graphs of y = f(x) and y = g(x).
11 . lNTEGRALS
Tis time, all the teims in a Riemann sumcoiiesponding to one paiticulai stiip x
k1

x x
k
add up to a Riemann sum foi an integial ovei the y vaiiable,

d(x)
c(x)
f(x
k
, y) dy x,
and adding all these we get the iteiated integial
(122)

D
f(x, y) dA =

b
a
_

d(x)
c(x)
f(x, y) dy
_
dx.
2.7. Anexamplethe parabolic oce building. Considei the iegion undei the giaph
of f(x, y) = x +y, above the domain
D =
_
(x, y) : 0 x 1, (1 x)
2
y 1
_
.
Te volume of this iegion is given by
V =

D
(x +y)dA.
We can compute this volume by nding the following iteiated integial
(123) V =

1
x=0

1
(1x)
2
(x +y) dy dx.
Alteinatively, the iegion D can also be desciibed as
D = {(x, y) : 0 y 1, 1

y x 1} .
Tis leads to the following iteiated integial foi the volume
(124) V =

1
y=0

1
1

y
(x +y) dx dy.
Both iteiated integials should give the same answei. Lets compute the ist one
V =

1
0

1
(1x)
2
(x +y) dy dx
=

1
0
_
1
2
xy +
1
2
y
2

1
(1x)
2
dx
=

1
0
_
x
_
1 (1 x)
2
_
+
1
2
_
1
2
(1 x)
4
_
dx
=

1
0
_
2x
2
x
3
+
1
2
_
4x 6x
2
+ 4x
3
x
4
_
dx
=

1
0
_
2x
2
x
3
+ 2x 3x
2
+ 2x
3

1
2
x
4

dx
=
2
3

1
4
+ 1 1 + 2
1
4

1
2

1
5
=
16
15
.
Note that even though the function we integiated is veiy simple (its just x + y) the
integial can still become complicated because of the shape of the domain D ovei which
we aie integiating.
2. DOUBLE lNTEGRALS 11
x
y
1
1
D
x
y
1
1
x
y
z
Figure 8. On the le: the domain of integration, a partition, and all pieces in the partition
corresponding to one value of y. On the right: The parabolic oice building, being the region
whose volume is computed in example 2.7
2.8. Double integrals in Polar Coordinates. Sometimes Caitesian cooidinates aie
just not the best choice. Foi instance, a disc oi iadius R, centeied at the oiigin, is veiy
easy to desciibe in polai cooidinates as all points with r R. ln Caitesian cooidinates
we need Pythagoias, and we have to say all points with x
2
+ y
2
R
2
. ln the same
x
y
r
r

r
x
y
Figure 9. Le: A polar rectangle and a partition by lines of constant (the spokes) and curves
of constant r (the arcs). Right: The area of a small piece of such a partition is approximately
A r r.
118 . lNTEGRALS
spiiit a polai iectangle is a domain of the foim
R = {all points with
0

1
, r
0
r r
1
} .
See Figuie 9 (on the lef). Teie is a veiy natuial way of paititioning such a iegion into
many smallei iegions, by cuuing the iegion along cuives of constant r (aics centeied at
the oiigin) oi constant (iays emanating fiom the oiigin). lf the paitition is suciently
ne, then the pieces in the paitition will almost be ieal Caitesian iectangles, with sides
r and r ( being the angle between adjacent iays, and r being the dieience in
iadius between two consecutive aics). Te aiea of such a small paitition piece is theiefoie
A r r, and one aiiives at the following foimula foi the integial of a function
of a polai iectangle
(12)

R
f(x, y) dA =

r1
r0

1
0
F(r, ) rd dr =

1
0

r1
r0
F(r, ) rdr d.
Heie F(r, ) = f(r cos , r sin ) is the function f(x, y) wiiuen in polai cooidinates.
Figure 10. The gray region is the region between the polar graphs r = a() and r = b().
Teie is a similai foimula foi moie complicated domains. lf a domain can be desciibed
in polai cooidinates by
D = {all points with , a(x) r b(x)}
and if we want to integiate a function z = f(x, y) of this domain, then we can again
paitition the domain D into many small pieces that aie bounded by ciiculai aics centeied
at the oiigin, and iays emanating fiomthe oiigin. Te aiea of a small piece in the paitition
is once again given by A r r, and theiefoie the integial of f ovei D is
(12)

D
f(x, y) dA =

b()
a()
F(r, ) r dr d.
lt is veiy common to use the same leuei f foi both functions, i.e. to wiite f(x, y) foi f as a function of
Caitesian cooidinates, and also f(r, ) foi the same function but wiiuen in Polai cooidinates. Tis begs the
question of what f(0.3, 1.24) means aie (0.3, 1.24) the polai oi the Caitesian cooidinates of the point at
which f is to be evaluated` To avoid this kind of ambiguity we will tiy to use dieient leueis foi the same
quantity iegaided as a function of Caitesian cooidinates, and of Polai cooidinates.
2. DOUBLE lNTEGRALS 119
x
y
z
/4
Figure 11. The graph of the function z = a in polar coordinates is called the helicoid. Here
we see one quarter turn of a helicoid with a =
1
2
. The volume under the helicoid is given by a
double integral which is best computed using polar coordinates. Which fraction of the volume in
the surrounding quarter cylinder lies beneath the helicoid?
2.9. Example: the volume under a quarter turn of a helicoid. A helicoid is the
suiface that in polai cooidinates is given by
z = a
wheie a > 0 is some constant. (See Chaptei lll, 4.2)
lf we choose the constant a =
1
2
, and take the ist quaitei tuin of this suiface, on
which 0
1
2
, then we get the pictuie in Figuie 11. ln that diawing we have only
included the pait with 0 r 1. To compute the volume of the iegion undei the quaitei
helicoid using Caitesian cooidinates, we would have to compute this integial
V =

1
0


1x
2
0
1
2
aictan
y
x
dy dx.
(Tiy to set up this integial youiself')
ln Polai cooidinates things aie easiei. Te domain is a polai iectangle,
0 r 1, 0
1
2
,
and the function is veiy simple,
F(r, ) =
1
2
.
Te double integial that iepiesents the volume is theiefoie
V =

D
1
2
dA =

1
0

/2
0
1
2
r d dr =

2
32
.
120 . lNTEGRALS
3. Problems
1. Compute these iterated integrals:
(a)

1
0

4
0
x dy dx
(b)

1
0

4
0
x dx dy
(c)

1
1

x
2
0
dy dx
(d)

y
0
sin y
y
dx dy
(e)


0
sin

dr d
(f)

1
0


1x
2
0
dy dx
2. What is wrong with the iterated integral

1
x
{

1
0
sin(x)dx
}
dy ?
Is the answer a number does it depend on x or y?
3. (a) Is the following true or false? For any two functions f(x) and g(y) one has

1
0

2
0
f(x)g(y) dx dy =
(
1
0
f(x) dx
)

(
2
0
g(y) dy
)
.
Explain your answer (if you claim true give a proof, if you claim false give a counterexample.)
(b) Is the following true or false? For any two functions f(x) and g(y) one has

2
0

1
0
f(x)g(y) dy dx =
(
1
0
f(x) dx
)

(
2
0
g(y) dy
)
.
Explain your answer (no, this is not the same question as before. Look at the integration bounds.)

(c) Suppose D is the unit disc, D = {(x, y) : x


2
+ y
2
< 1}. True or False: For any two functions
f(x) and g(y) one has

D
f(x)g(y) dx dy =
(
1
1
f(x) dx
)

(
1
1
g(y) dy
)
.
Again, explain your answer.
4. Answer the question posed in Figure 6.
5. Compute the following double integrals. In each case sketch the domain of integration and
show which iterated integral you must compute to find the given double integral.
(a)

D
(1 +x) dA D = {(x, y) : 0 x 2, 0 y 4}.
(b)

D
(x +y) dA D = {(x, y) : |x| 1, 0 y 4}.
(c)

D
xy dA D = {(x, y) : 0 x y, 1 y 2}.
(d)

D
dA D =
{
(x, y) :
1
2
y
2
x

y, 0 y 1
}
.
(e)

D
x
2
y
2
dA D = {(x, y) : 1 x 2, 1 y x}.
(f)

D
y
e
x
dA D =
{
(x, y) : 0 x 1, 0 y x
2
}
.
(g)

D
xcos y dA D =
{
(x, y) : 0 x

/2, 0 y x
2
}
.
(h)

x
3
+ 1 dA D = {(x, y) : 0 y 1,

y x 1}.
4. TRlPLE lNTEGRALS 121
(i)

D
y sin(x
2
) dA D =
{
(x, y) : 0 y 1, y
2
x 1
}
.
(j)

D
x

1 +y
2
dA D =
{
(x, y) : 0 x 1, x
2
y 1
}
.
(k)

D
2

1 x
2
dA D is the triangle bounded by the y axis, the line y = 1
and the line y = x.

6. Find the volumes of the following regions by computing a double integral.


(a) the region bounded by z = x
2
+y
2
and z = 4.
(b) the region in the first octant bounded by y
2
= 4 x and y = 2z.
(c) the region in the first octant bounded by y
2
= 4x, 2x +y = 4, z = y, and y = 0.
(d) the region in the first octant bounded by
x +y +z = 9, 2x + 3y = 18, and x + 3y = 9.
(e) the region in the first octant bounded by x
2
+y
2
= a
2
and z = x +y.
(f) the region bounded by x
2
+y
2
= 4z and z = 2.
(g) the region bounded by z = x
2
+y
2
and z = y.
7. The average value of a function f(x, y)
over a domain D is by definition
average f over D =

D
f(x, y) dA
area of D
Find the average value of f(x, y) =
e
y

x +e
y
on the rectangle with vertices
(0, 0), (4, 0), (4, 1) and (0, 1).
8. Suppose f(x) is a positive function de-
fined on an interval a x b. Let A
be the area under the graph of y = f(x),
(a c b), and let B be the area under the
graph of y = f(x)
2
(a c b)
(a) Compute

b
a

f(x)
0
dydx.
(b) Compute

b
a

f(x)
0
ydydx.
9. Let V be the volume under the graph of
the function
z =
2xy
x
2
+y
2
,
above the region
D =
{
(x, y) : x 0, y 0, x
2
+y
2
1
}
.
(a) Write an iterated integral for the volume
V , using Cartesian coordinates. (You dont
have to compute the integral you get.)
(b) Compute V using polar coordinates.
10. Let V be the volume under the graph of
z = xy above the domain
D =
{
(x, y) : x 0, y 0, x
2
+y
2
4
}
.
Try to draw the region D, and the graph of
z = xy above D.
(a) Use Cartesian coordinates to compute V .
(Hint: this is similar to part (i) of the previ-
ous problem, but the integral in this problem
isnt as bad.)
(b) Use Polar Coordinates to compute V .
4. Triple integrals
lnstead of integiating ovei two-dimensional iegions in the plane, we can also integiate
ovei thiee-dimensional iegions in space. ln this section we will see the denition, how
to compute tiiple integials using iteiated integials, and some examples of how tiiple
integials come up in the ieal woild.
4.1. Denition, and how to compute triple integrals. Te denition of tiiple in-
tegials follows the same pauein as that of double integials. Let D be some thiee di-
mensional iegion in thiee dimensional space D could be a cube, a block, a cylindei, a
122 . lNTEGRALS
spheie, oi in geneial, the iegion enclosed by some suiface. A paiticulai case is that of a
rectangular blo, which is a iegion dened by the inequalities
(12) a
x
x b
x
, a
y
y b
y
, a
z
z b
z
.
To dene the tiiple integial of a function w = f(x, y, z) ovei such a iegion we considei
a paitition of D into many smallei pieces. We numbei the pieces 1, 2, , N and foi
each j we choose a sample point (x
j
, y
j
, z
j
) fiom the j
th
paitition piece. Let V
j
be the
volume of the j
th
paitition piece and considei the Riemann sum
f(x
1
, y
1
, z
1
)V
1
+ +f(x
N
, y
N
, z
N
)V
N
=
N

j=1
f(x
j
, y
j
, z
j
)V
j
.
lf these Riemann sums conveige to some numbei as we choose nei and nei paititions,
then we call this limit is called the triple integral, or volume integral, of f over D. Te
notation we use is
(128)

D
f(x, y, z) dV = lim
as the
partition
gets finer
N

j=1
f(x
j
, y
j
, z
j
)V
j
.
lf the domain D is a iectangulai block, dened by the inequalities (12), then the tiiple
integial can be computed by an iteiated integial
(129)

D
f(x, y, z) dV =

bz
az

by
ay

bx
ax
f(x, y, z) dx dy dz.
Tis follows fiom the same kind of aiguments that allowed us to tuin a double integial
into an iteiated integial in 2.3.
We can use (129) to compute a tiiple integial ovei any thiee dimensional block. To
compute tiiple integials ovei moie geneial domains we can use the same slicing method
as in 2.. lf the domain D is given by inequalities of the type
(130) a
x
(y, z) x b
x
(y, z), a
y
(z) y b
y
(z), a
z
z b
z
.
wheie a
y
(z), b
y
(z), a
z
(y, z), and b
z
(y, z) now aie functions iathei than constants, then
the tiiple integial of a function f(x, y, z) ovei D is given by

D
f(x, y, z) dV =

bz
az

by(z)
ay(z)

bx(y,z)
ax(y,z)
f(x, y, z) dx dy dz.
4.2. Example the integral of f(x, y, z) = x
2
+y
2
over a rectangular blo. Lets
compute the integial of f(x, y, z) = x
2
+y
2
ovei the domain
D = {(x, y, z) : 0 x A, 0 y B, 0 z C} ,
wheie A, B, and C aie the sides of the block.
Te integial of f ovei D is

D
_
x
2
+y
2
_
dV =

C
0

B
0

A
0
_
x
2
+y
2
_
dx dy dz
4. TRlPLE lNTEGRALS 123
lt is a good idea to wiite such an integial as

D
_
x
2
+y
2
_
dV =
C

z=0
B

y=0
A

x=0
_
x
2
+y
2
_
dx dy dz =
1
3
ABC
_
A
2
+B
2
_
,
to emphasize which integial goes with which vaiiable.
Te computation goes in thiee steps (theie aie thiee integials). Te inneimost integial
is

A
0
(x
2
+y
2
) dx =
1
3
A
3
+y
2
A.
Next we integiate this with iespect to y

B
0

A
0
_
x
2
+y
2
_
dx dy =

B
0
_
1
3
A
3
+y
2
A
_
dy =
1
3
A
3
B +
1
3
AB
3
.
nally, we integiate with iespect to z

C
0

B
0

A
0
_
x
2
+y
2
_
dx dy dz =

C
0
_
1
3
A
3
B +
1
3
AB
3
_
dz
=
1
3
A
3
BC +
1
3
AB
3
C
=
1
3
ABC
_
A
2
+B
2
_
.
4.3. Example of setting up a triple iterated integral the integral of e
x
over the
unit sphere. Suppose we needed to know the integial

D
e
x
dV,
wheie the domain
D =
_
(x, y, z) : x
2
+y
2
+z
2
1
_
is the unit spheie. By slicing the domain D in the x, y, and z diiections we can desciibe
following the geneial template in (130)
z can take any value between 1 and +1,
foi given z the cooidinate y can be anything between

1 z
2
and +

1 z
2
,
foi given y and z the iemaining cooidinate xcan have all values fiom

1 y
2
z
2
to +

1 y
2
z
2
.
(See Figuie 12.)
Tis lets us wiite the tiiple integial as an iteiated integial

D
e
x
dV =

1
1


1z
2

1z
2


1y
2
z
2

1y
2
z
2
e
z
dx dy dz.
Even though it can be computed this is not an easy integial the point of this example
was to nd the integiation bounds in the iteiated integial.
124 . lNTEGRALS
z
y
x
Figure 12. Turning a triple integral over the unit sphere into an iterated integral. The hor-
izontal gray disc contains all points with a given fixed value of z; the solid line in that disc contains
all points at height z whose y coordinate is also fixed at a particular value. From this drawing we
can see that z runs between 1 and +1; for any given z, the y coordinate runs between

1 z
2
and +

1 z
2
; for fixed y and z, the x coordinate can take any value between

1 y
2
z
2
and

1 y
2
z
2
.
5. Why compute a Triple Integral?
5.1. e 4D-volume under a graph. Just as

b
a
f(x) dx is the aiea between the
giaph of the function y = f(x) and the inteival [a, b] on the x-axis, and

D
f(x, y) dA
is the volume caught between the giaph of z = f(x, y) and the domain Din the xy plane,
theie should be a similai desciiption of

D
f(x, y, z) dV . Teie is, but it iequiies
some imagination the giaph of f is the set of points in four dimensional space whose
cooidinates (x, y, z, w) satisfy w = f(x, y, z), and the tiiple integial

D
f(x, y, z)dV
is the foui dimensional volume of the foui dimensional iegion caught between the giaph
of f and the domain D in xyz-space. Of couise, even though people will diaw caitoon
like iepiesentations of the situation like this,
w=f(x, y, z)
xyz-space
w
What appears as thex-axis in this
drawing really is meant to represent
the three dimensional xyz-space.
The region between the graph and
the horizontal axis is four dimensional
. WHY COMPUTE A TRlPLE lNTEGRAL` 12
we cannot ieally visualize foui dimensional volumes. Rathei than telling us what the
tiiple integial is, the inteipietation integial=volume gives a denition of what foui
dimensional volume should be.
5.2. e average of a function over a domain D. Teie is a foimula foi the aveiage
value of a function on a iegion. Te only iigoious denition foi the aveiage is just that
foimula, so we could simply state the foimula be done with it. Heie it is the aveiage of
a function w = f(x, y, z) ovei a iegion D is dened to be
(131)
Average of f
over D
=
1
V
D

D
f(x, y, z) dV.
Teie is howevei an intuitive deiivation (a stoiy) that justies why we call this paitic-
ulai quantity the aveiage. Undeistanding this deiivation is at least as impoitant as just
knowing the foimula (131).
Why (131) deserves to be called the average. What is an aveiage` lf we have nitely
many numbeis a
1
, , a
N
then theii aveiage is just
Average =
a
1
+ +a
N
N
.
lf we only have nitely many points (x
1
, y
1
, z
1
), , (x
N
, y
N
, z
N
) in the iegion D then
the aveiage function value at these points is
Average function value
at given points
=
f(x
1
, y
1
, z
1
) + +f(x
N
, y
N
, z
N
)
N
.
To dene the aveiage of a function ovei a iegion D, we cannot simply add all the function
values of f at all the points in D because theie aie innitely many such points. lnstead,
we spiinkle the iegion D with a veiy laige but nite numbei of points, and calculate the
aveiage value of the function at all these points. lf the points aie evenly distiibuted, and
if theie aie enough of them, then the aveiage value of the function at the dots should be a
good appioximation foi the aveiage value of the function on the iegion. E.g. the aveiage
of oui function ovei the iegion on the lef should be appioximately the aveiage of the
D
D
1
D
2
D
3
D
n
function at the dots diawn in that iegion.
To appioximate the aveiage at the dots we paitition the iegion into many small pieces,
which we label D
1
, , D
n
. We wiite V
j
foi the volume of the j
th
piece D
j
, and V
D
foi
the volume of the whole iegion D. We assume that the pieces aie so small that we may
assume that the function is piactically constant in each piece.
Since the dots aie evenly distiibuted ovei D, the numbei of dots in the j
th
paitition
piece is piopoitional to the volume of that piece, so
(132)
N
j
N

V
j
V
D
wheie N
j
is the numbei of dots in the j
th
piece, and N is the total numbei of dots.
12 . lNTEGRALS
To compute the aveiage value of f at all the dots we begin with
sum of f at all dots =

j
sum of f at all dots in j
th
piece .
lf we pick a sample point (x
j
, y
j
, z
j
) in each piece D
j
, then, since the pieces aie assumed
to be small, we may appioximate the function value at eveiy dot in D
j
by the value of
the function at the sample point. Teie aie N
j
dots in D
j
, so we nd that
sum of f at all dots

j
N
j
f(x
j
, y
j
, z
j
)
Using (132) we theiefoie nd that the aveiage function value at all the dots is
sum of f at all dots
number of dots in D

1
N

j
N
j
f(x
j
, y
j
, z
j
)
=

j
V
j
V
D
f(x
j
, y
j
, z
j
)
=
1
V
D

j
f(x
j
, y
j
, z
j
) V
j

1
V
D

D
f(x, y, z) dV.
Tis is exactly how we had dened the aveiage of f ovei the iegion D.
Keep in mind that the above is not a pioof of the equation (131), but iathei an intuitive
justication foi taking (131) as denition of the aveiage.
5.3. Example 4.2 continued. ln 4.2 we computed the volume integial of
f(x, y, z) = x
2
+y
2
ovei the iectangulai block D given by 0 x A, 0 y B, 0 z C and we found

D
_
x
2
+y
2
_
dV =
1
3
ABC
_
A
2
+B
2
_
.
Since the volume of the block is ABC, the aveiage value of f(x, y, z) = x
2
+y
2
ovei the
block D is
Average of x
2
+y
2
over D
=
1
3
ABC
_
A
2
+B
2
_
ABC
=
1
3
_
A
2
+B
2
_
.
5.4. Densities. lf a substance (foi an example, think of a gas in a cylindei) occupies
a ceitain iegion D in space, then its density is dened to be
= density =
mass in D
volume of D
.
lf the substance is evenly distiibuted thioughout the iegion D, then the mass-to-volume
iatio will be the same foi any subiegion D

. Tus the mass contained in any smallei


iegion D

will be piopoitional to the volume of that iegion


mass in D

= volume of D

.
When the substance is not distiibuted evenly this piopoitionality will no longei hold,
and we say that the density vaiies fiom point to point. lf we now want to give a piecise
denition of the density at any point P, we iun into the same kind of pioblem we had
. WHY COMPUTE A TRlPLE lNTEGRAL` 12
in ist semestei calculus when we tiied to dene the slope of a tangent, oi the velocity
at one moment in time. Namely, the density at P should be the mass of the substance
at P divided by the volume of the point P but theie is no mass at one point, and the
volume of one point is zeio, so this leads to density=
0
0
= Te way out of this is to
calculate the aveiage density foi veiy small iegions D

suiiounding the point P, and to


declaie those as appioximations of the density at P. To get a beuei appioximation we
should choose a smallei iegion D

.
Tis is summaiized in the following foimula,
(133) (x, y, z) = lim
D

P
mass in D

volume of D

wheie D

P means that we aie taking the limit as the iegion D

shiinks to the point


P.
D D D D
4
P
Figure 13. Density of gas in a container; in these drawings most of the gas concentrates in the
boom of the container. Le: The total mass in two regions, D

and D

, depends on their
location, even though they have the same shape and volume. Right: To define the density at a
point P, we compute the average density over smaller and smaller regions D1, D2, which shrink
to the given point P. If the average densities converge to some number, then we call that limit the
density at P.
5.5. Mass as integral of the density. Suppose the density of a substance is given to
us as a function (x, y, z), how do we nd the total mass of the substance piesent in a
paiticulai iegion D` Te answei is in teims of a tiiple integial, and the way this integial
comes about is typical foi a laige numbei of applications of double and tiiple integials.
To nd the total mass piesent in a iegion D we paitition it into many small pieces, and
compute the mass in each small piece. Considei one such piece. lf it is small enough, then
we assume that the density (x, y, z) is neaily constant in that small piece, and hence the
total mass in one small piece will be
mass in a piece of the partition = (x, y, z) V.
Heie (x, y, z) is a sample point in the paitition piece, and V is the volume of the piece.
So when we compute the total mass by adding all the masses of the paitition pieces, each
piece in the paitition contiibutes one teim of the foimf(x, y, z)V . Oui foimula foi the
total mass is theiefoie a Riemann sum foi the following tiiple integial
(134) total mass =

D
(x, y, z) dV.
128 . lNTEGRALS
5.6. Example: air in the atmosphere. How much air is there in the atmosphere in a
vertical column of height H above one square meter?
Accoiding to one model of the atmospheie, the density of the atmospheie decays ex-
ponentially with height, so that
(13) (x, y, z) = Ce
z/L
(kg/m
3
)
wheie z is the height above sea level, and x, y aie hoiizontal cooidinates. Te constant
x
y
z
1
1
H
C is the density of aii at sea level, and L is anothei constant (L must have the units of
length).
We adapt oui cooidinates to the 11 squaie which is the base of the aii column whose
mass we aie to compute, namely, we let the oiigin be one of the coineis of the squaie,
and we let the sides at this coinei be the x and y axes. Te iegion occupied by the aii
column is then a iectangulai block
D = {(x, y, z) : 0 x 1, 0 y 1, 0 z H}
and the mass of the aii in this block is
M =

D
(x, y, z) dV
=

H
0

1
y=0

1
x=0
Ce
z/L
dx dy dz
= LC
_
1 e
H/L
_
.
To get the mass of all the aii above oui 1 1 squaie, we let H which leads to
Total Mass = LC.
5.7. e moment of inertia of a solid about an axis of rotation. An object of mass
m that moves with velocity v has kinetic eneigy given by
(13) K =
1
2
mv
2
.
lf a solid object is iotating about an axis, then it also has kinetic eneigy, but the foimula
(13) does not apply, because dieient paits of the solid will be moving with dieient
velocities. Te pioblem is that v is not a constant it vaiies fiom place to place, and thus
it is a function of wheie we measuie the velocity.
(radians/second)
v=r
r
the kinetic energy of a
whirling potato
To compute the kinetic eneigy of a iotating solid we bieak it up into small pieces
if each of the pieces is small enough, then all the paiticles in that small piece will have
neaily the same velocity. A well known foimula fiom tiigonometiy says that if the object
is iotating with angulai velocity about an axis, then the velocity of a paiticle in the
object is given by v = r wheie r is the distance fiom the paiticle to the axis of iotation.
On the othei hand the mass of such a small piece will be V , wheie is the density of
the mateiial (which we assume to be constant heie), and V is the volume of the small
piece. Teiefoie if we bieak the object into many small pieces (paitition the object), the
kinetic eneigy of any one of the small pieces is
K.E. of one piece =
1
2
(r)
2
V.
. lNTEGRATlON lN SPEClAL COORDlNATE SYSTEMS 129
Adding the kinetic eneigies of all the small pieces again gives us a Riemann sum foi an
integial, and this leads us to the foimula
(13) K =

D
1
2

2
r
2
dV =
1
2
M
2
,
wheie
(138) M
def
=

D
r
2
dV.
is called the moment of inertia of the given object about the given axis of iotation.
5.8. Example. Compute the moment of inertia of a wooden rectangular block
D =
_
(x, y, z) : 0 x A, 0 y B, 0 z C
_
.
around the z axis. e density of the wood is .
Te integial we have to calculate is
M =

D
r
2
dV.
To compute this we have to guie out what r is since r is the distance fiom the point
(x, y, z) to the axis of iotation, and since this axis is the z-axis, we get, by Pythagoias,
r
2
= x
2
+y
2
. Teiefoie we have to compute
M =

D
_
x
2
+y
2
_
dV.
We have alieady computed this integial in 4.2, wheie we found that
r
x
y
P (x,y,z)
a
x
i
s

o
f

r
o
t
a
t
i
o
n
M =
1
3
ABC(A
2
+B
2
).
6. Integration in special coordinate systems
Many volume integials aiise in situations wheie theie is a lot of symmetiy. When this
happens Caitesian x, y, z cooidinates aie usually not the best choice to compute the
integial. Teie aie many dieient cooidinates besides Caitesian. ln this section we will
look at the two most commonly used cooidinate systems. Tey can both be thought of as
thiee-dimensional vaiiations on polai cooidinates in the plane.
6.1. Cylindrical coordinates. Let P be some point in thiee dimensional space. lf we
piovide the z cooidinate as well as the polai cooidinates (r, ) of the piojection of P on
the xy plane, then the location of P is completely deteimined. See the diawing on the
lef in Figuie 14. Fiom this diawing it is easy to deiive the ielation between cylindiical
Caitesian cooidinates
(139)
x = r cos
y = r sin
z = z
130 . lNTEGRALS

r
z
x
y
z

sin
x
y
z
Figure 14. Le: In cylindrical coordinates we specify the location of a point by its height z above
the xy-plane, and the polar coordinates (r, ) of its projection on the xy-plane. Right: In spherical
coordinates we specify the location of a point by its distance to the origin, the polar angle of
its projection on the xy-plane, and the angle between the z-axis and the line segment from the
point to the origin.
6.2. Spherical coordinates. We can also specify the location of a point P by piovid-
ing these thiee numbeis
the distance fiom P to the oiigin
the angle between the positive z-axis and the line fiom the oiigin to the point P
the polai angle of the piojection of P onto the xy-plane.
See the diawing on the iight in Figuie 14, fiom which we can deiive the following
DO NOT MEMORIZE.
What if the north pole hap-
pens to be on the x-axis?
Can you still relate spher-
ical and Cartesian coordi-
nates?
ielation between the spheiical cooidinates (, , ) and the Caitesian cooidinates (x, y, z)
of a point
(140)
x = sin cos
y = sin sin
z = cos
Te angle takes values between 0 and +, with = 0 on the noith pole, and = on
the south pole. Te polai angle can take all values fiom 0 to 2, oi moie geneially any
value in some inteival of length 2 (like < < ).
6.3. Triple integral in cylindrical coordinates. Suppose we wanted to nd a tiiple
integial

D
f(x, y, z) dV
ovei a domain D which is a iectangulai block in cylindiical cooidinates, i.e. suppose D
is given by the inequalities
r
0
r r
1
, z
0
z z
1
,
0

1
.
Lets tiy to wiite it as an iteiated integial. To do this we paitition the iegion D into many
small pieces by dividing the inteival r
0
r r
1
into pieces of length r, the inteival
z
0
z z
1
into pieces of length z, and inteival
0

1
into pieces of length
. Te whole iegion D then gets bioken up into small iegions in which the iadius
is constiained to lie in the inteival (r, r + r), the height to the inteival (z, z + z)
and the polai angle to (, ). See Figuie 1. Such a small iegion is appioximately a
. lNTEGRATlON lN SPEClAL COORDlNATE SYSTEMS 131
x
y
z
dr
r
rd
dz

d
Figure 15. The cylindrical volume element.
iectangulai block, so that we can appioximate its volume by multiplying the lengths of
its sides, which leads to
V r r z.
Aiguing as in the case of polai cooidinates (see 2.8) we get the following iteiated integial
foimula foi a tiiple integial ovei a iectangulai block in cylindiical cooidinates
(141)

D
f(x, y, z)dV =

r1
r0

z1
z0

1
0
f(x, y, z)r d dz dr
lf the function to be integiated is given in teims of the Caitesian cooidinates x, y, z, then
we ist have to iewiite it in teims of cylindiical cooidinates using (139).
6.4. Triple integral in spherical coordinates. A spheiical block is a iegion D which
in spheiical cooidinates is given by the inequalities

0

1
,
0

1
,
0

1
To integiate a function ovei such a block we divide into many small spheiical blocks. ln
each of these blocks incieases by , by , and by . See Figuie 1. Any su-
ciently small spheiical block is appioximately iectangulai, and we can theiefoie compute
its volume by multiplying the lengths of its sides. lf we caiefully look at the diawing on
the iight in Figuie 1, then we nd that
V sin .
Tis leads us to the foimula foi integiation in spheiical cooidinates
A spherical block
(142)

D
f(x, y, z)dV =

1
0

1
0

1
0
f(x, y, z)
2
sin d d d
As in the cases of polai cooidinates and cylindiical cooidinates we ist have to expiess
the function f(x, y, z) in teims of the vaiiables , , and , using (140).
132 . lNTEGRALS

d
sin
sin d
d
Figure 16. Le: a number of small spherical blocks with varying but the same and stacked
together. Right: the volume of a small spherical block is approximately the product of the lengths
of its sides, so V
2
sin .
6.5. Example Rotational Kinetic energy of the Earth. Te eaith is ioughly a spheie
with iadius a 6400km, which iotates aiound its axis with angulai velocity =
2iad/day. Lets assume that the density of the eaith is constant, say, kg/m
3
.
To compute the total kinetic eneigy of the eaith we can use foimula (13), which tells
us that we have to nd

Earth
r
2
dV,
wheie r is the distance to the eaiths axis of iotation.
Tis integial is best computed using spheiical cooidinates, in which
r = sin (see Figuie 14, iight).
Tus the kinetic eneigy is
K =
1
2

Earth

2
sin
2
dV
=
1
2


=0

a
=0

2
=0

2
sin
2

2
sin d d d
. .
dV
.
Afei doing the and integials, we get
K =
2
a
5
5


0
sin
3
d.
Tis last integial can be done seveial ways (integiate by paits and nd a ieduction foi-
mula, oi substitute u = cos ). Te iesult is
K =
4
15

2
a
5
.
7. Problems
. PROBLEMS 133
1. Describe the following sets (given in
spherical coordinates):
(a) All points with = /6.
(b) All points with = .
(c) All points with = /2.
(d) All points with = /2
2. Let E be the part of the sphere with ra-
dius a, centered at the origin, and contained
in the first octant
(a) Describe E in terms of spherical coordi-
nates.
(b) Describe E in terms of cylindrical coordi-
nates. There are two possible answers, find
both.
3. Draw the volume elements in cylindrical
and in spherical coordinates and show how
these lead to dV = rdrddz, and dV =

2
sin d d d, respectively.

4. Look at Figure 12. Suppose the grey disc


has height z, and suppose all points on the
line segment drawn in this disc have the
same y-coordinate (y).
(a) What are the radii of the two circles
drawn in the xy plane?
(b) What are the coordinates of the two end-
points of the drawn line segment?
5. The potential energy in a pile of honey.
If you li an object to height h above the
ground, then the potential energy you give
it is mgh, where m is the mass of the ob-
ject, and g is the acceleration of gravitation
(g 9.8m/sec
2
).
Suppose that a certain substance occu-
pies a three dimensional region D (think of
honey that has just been poured into a jar,
see the drawing which gives a two dimen-
sional side view of the situation).
h
e
i
g
h
t
=
f
(
x
,

y
)
e potential energy
of a small piece is
m g z
z
Assuming the base of the jar is an A B
rectangle, the honey occupies the region
D =
{
(x, y, z) :
0 x A, 0 y B,
0 z f(x, y)
}
.
Here f(x, y) is the height of the honey
above the point (x, y) in the base of the jar.
(a) What is the potential energy of a small
piece of the honey at (x, y, z) (assume the
density of the honey is , and that this is
constant.) Is your formula an exact formula?

(b) Write a volume integral for the total


(gravitational) potential energy contained in
the honey.
(c) Write your triple integral as an iterated
integral, and show that you can do the inte-
gration in the z direction even if you dont
know the height function f(x, y).
6. The kinetic energy in a tornado.
Assume an airmass is whirling around
the z-axis, and assume that the wind veloc-
ity v(r) only depends on the distance from
the z-axis.
Assume furthermore that the air has con-
stant density .
(a) Derive a volume integral for the total ki-
netic energy of the airmass in a given region
D. (The kinetic energy of an object of mass
m and velocity v is
1
2
mv
2
. See the deriva-
tion of the moment of inertia in 5.7).
(b) Suppose the velocity is actually given by
v(r) = 1/

1 +r
2
, the density is = 1.
Let Dbe the cylinder of height H and radius
R, with the z-axis as its central axis. How
much kinetic energy does the airmass in D
have? (Hint: which coordinates should you
use?)
134 . lNTEGRALS
7. For each of the following iterated inte-
grals, describe and draw the domain of in-
tegration. Then compute the integral.
(a)

2
0

x
2
1

y
1
xyz dz dy dx.
(b)

1
0

x
0

ln y
0
e
x+y+z
dz dy dx.
(c)

/2
0

sin
0

r cos
0
r
2
dz dr d.
(d)

sin
0

r sin
0
r cos
2
dz dr d.
(e)

1
0

y
2
0

x+y
0
xdz dxdy.
(f)

2
1

y
2
y

ln(y+z)
0
e
x
dxdz dy.
8. Find the mass of a cube with edge length
2 and density equal to the square of the dis-
tance from one corner.
9. Find the mass of a cube with edge length
2 and density equal to the square of the dis-
tance from one edge.

If a mass is distributed throughout a re-


gion D with density (x, y, z), then, by def-
inition the coordinates (X, Y, Z) of the cen-
ter of mass
X
def
=

D
x(x, y, z)) dV
Mass of D
,
and similarly for Y and Z.
10. An object occupies the volume of the up-
per hemisphere of x
2
+ y
2
+ z
2
= 4 and
has density z at (x, y, z). Find the center of
mass.

11. An object occupies the volume of the


pyramid with corners at (1, 1, 0), (1, 1, 0),
(1, 1, 0), (1, 1, 0), and (0, 0, 2) and has
density x
2
+y
2
at (x, y, z). Find the center
of mass.
12. Let z = f(x, y) be a function on some
domain D, and assume that D is split into
two parts: D+, on which f 0, and D, on
which f(x, y) < 0.
Let V+ be the volume of the region be-
neath the graph of f and above the domain
D+ in the xy-plane, and, similarly, let V
be the volume of the region above the graph
of f and beneath the region D in the xy-
plane.
Reminder: volumes are never negative, so
both V+ 0 and V 0.
(a) Express the following integrals in terms
of V+ and V:
I =

D
+
f(x, y) dA,
J =

f(x, y) dA
K =

D
f(x, y) dA
L =

D
|f(x, y)| dA.

(b) Find the region E in three dimensional


space for which

E
(1 x
2
y
2
z
2
) dV
is a maximum. [Hint: Suppose E is some
region; consider then what happens to the
integral if you make E larger by adding on
a piece.]
13. Evaluate

1
0

x
0


x
2
+y
2
0
(x
2
+y
2
)
3/2
x
2
+y
2
+z
2
dz dy dx.

14. Evaluate

x
2
dV over the interior of
the cylinder x
2
+y
2
= 1 between z = 0 and
z = 5.
15. Evaluate

xy dV over the interior of


the cylinder x
2
+y
2
= 1 between z = 0 and
z = 5.
16. Evaluate

z dV over the region


above the x-y plane, inside x
2
+y
2
2x = 0
and under x
2
+y
2
+z
2
= 4.
17. Evaluate

yz dV over the region in


the first octant, inside x
2
+y
2
2x = 0 and
under x
2
+y
2
+z
2
= 4.
. PROBLEMS 13
18. Evaluate

x
2
+y
2
dV over the inte-
rior of x
2
+y
2
+z
2
= 4.
19. Evaluate


x
2
+y
2
dV over the in-
terior of x
2
+y
2
+z
2
= 4.
20. Find the mass of a right circular cone of
height h and base radius a if the density is
proportional to the distance from the base.
21. Find the mass of a right circular cone of
height h and base radius a if the density is
proportional to the distance from its axis of
symmetry.
22. An object occupies the region inside the
unit sphere at the origin, and has density
equal to the distance from the x-axis. Find
the mass.
23. An object occupies the region inside the
unit sphere at the origin, and has density
equal to the square of the distance from the
origin. Find the mass.
24. An object occupies the region between
the unit sphere at the origin and a sphere
of radius 2 with center at the origin, and has
density equal to the distance fromthe origin.
Find the mass.
CHAPTER
Vector Calculus
1. Vector Fields
So fai we have been studying the calculus of functions of seveial vaiiables. Functions
aie used to desciibe things that have dieient values at dieient locations, e.g. quantities
like tempeiatuie, oi density. Many othei physical phenomena aie desciibed by vector
elds, i.e. by vectois whose diiection and magnitude can vaiy fiom place to place. Vectoi
calculus is the theoiy of integiation and dieientiation of vectoi elds.
By denition, a vectoi eld in the plane is a vectoi valued function of two vaiiables
wheieas an oidinaiy function of two vaiiables gives us a numbei foi each (x, y) in its
domain, a vectoi eld gives us a vectoi in the plane foi each point (x, y) in its domain.
Such a vectoi is deteimined by its two components, both of which aie oidinaiy functions
of (x, y). Te notation we will use in this couise is as follows
(143)
#
v(x, y) =
_
P(x, y)
Q(x, y)
_
= P(x, y)
#
+Q(x, y)
#
.
Foi a vectoi eld in thiee dimensional space we must specify a vectoi
#
v(x, y, z) at
each point (x, y, z) in a thiee dimensional domain
#
v(x, y) =
_
_
P(x, y)
Q(x, y)
R(x, y)
_
_
= P(x, y)
#
+Q(x, y)
#
+R(x, y)
#
k.
To diaw a vectoi eld in the plane we would have compute
#
v(x, y) at lots of points and
simply plot them. Te moie points we pick, the busiei the pictuie gets. See foi example
Figuie 1, in which the vectoi eld
(144)
#
v(x, y) =
_
y/(x
2
+y
2
)
x/(x
2
+y
2
)
_
is diawn.
2. Examples of vector elds
2.1. Gradients as vector elds. We have alieady seen examples of vectoi elds be-
foie, namely the gradient of any function f(x, y) is a vector eld:
#
f(x, y) =
_
f
x
(x, y)
f
y
(x, y)
_
.
ln fact, the example (144) is such a vectoi eld it is the giadient of the polai angle .
ln lll.4.2 we saw that foi x > 0 this angle is given by (x, y), and we checked in piob-
lem lll.1.1 that
#
v given by (144) and shown below in Figuie 1 satises
#
v =
#
.
13
138 . VECTOR CALCULUS
Figure 1. A vector field in the plane. This vector field is
#
v(x, y) =
#
(x, y) =
y
x
2
+y
2
#
+
x
x
2
+y
2
#
.
In general, drawings of vector fields become messy in the region where the vectors are long, because
they tend to overlap. Drawing a three dimensional vector field is challenging.
2.2. Fluid ow. Vectoi elds appeai in vaiious ways in physics. Te easiest way to
visualize a vectoi eld is by thinking of it as the velocity eld of a uid ow. Suppose
a uid is owing thiough a ceitain iegion in space. Te velocities of the uid paiticles
will geneially vaiy fiom place to place, and also with time. A uid ow is called steady if
the velocity of a uid paiticle only depends on its location. Tis means that the velocity
vectoi
#
v of a uid paiticle is a function of its cooidinates (x, y, z) only, and does not
depend on time.
central axis
r
wall
Figure 2. Fluid flow in a cylindrical pipe. Le: as a viscous fluid flows through a pipe it
sticks to the walls, and so its velocity will be highest at the center of the pipe. Right: a drawing
of a cross section the flow on the le. We see the vector field corresponding to so-called Poiseuille
flow, given by Equation (145).
Foi instance, if a viscous uid ows thiough a cylindiical pipe, the velocity of the uid
will only depend on the distance to the cential axis of the pipe. On the walls the velocity
will vanish (the uid sticks to the wall of the pipe), and in the centei the uid will move
fastest. Undei ceitain ciicumstances it follows fiom the laws of uid mechanics that the
velocity eld
is always paiallel to the cential axis, and
depends quadiatically on the distance to the cential axis.
2. EXAMPLES OF VECTOR FlELDS 139
lt is given by
(14)
#
v(x, y, z) = v
c
_
1
r
2
R
2
_
#
=
_
_
v
c
_
1 (r/R)
2
_
0
0
_
_
,
wheie Ris the iadius of the pipe, r is the distance to its cential axis, and v
c
is the velocity
at the centei of the pipe.
Tis example desciibes the motion of a uid, but a vectoi eld can be the velocity eld
of anything that moves, in paiticulai, a gas ow has a velocity eld, and the velocities in
a moving elastic solid (think Jello) must also be desciibed by a vectoi eld.
2.3. Force elds. lf we assume the Eaith is at, then the giavitational foice it exeits
on a mass m is always the vectoi
#
F =
_
0
mg
_
. We can think of this as a constant vectoi
eld its magnitude and diiection aie the same eveiywheie.
But the Eaith is not at, and accoiding to Newton the giavitational foice
#
F is a vectoi
pointing towaids the centei of the eaith, whose magnitude is inveisely piopoitional to
the distance to the centei of the Eaith. lf we choose the Eaiths centei to be the oiigin,
then Newtons law looks like this
(14)
#
F(x, y, z) = C
#
x

#
x
3
,
#
x =
_
_
x
y
z
_
_
.
Heie C is a constant that depends on the mass m of the object, and the mass M of the
x
F
E
a
rth
M
m
m
F
at Earth
Eaith (physics tells us that C = GMm, wheie G is called the univeisal giavitational
constant.)
Othei piominent examples of vectoi elds appeai in the theoiy of electiomagnetism.
Te electiic cuiients and chaiges aiound us cieate an electiic eld and a magnetic eld,
which at each point in space aie given by vectois
#
E and
#
B. Tese vectois change fiom
place to place, and so they dene vectoi elds
#
E =
#
E(x, y, z),
#
B =
#
B(x, y, z).
Foi example, Coulombs law states that the electiic eld geneiated by a chaiged paiticle
at the oiigin is given by
(14)
#
E(x, y, z) =
Q
4
0
#
x

#
x
3
,
which is almost the same as Newtons law (14) foi the giavitational eld. Heie
0
is
some constant, and Q is the electiic chaige of the paiticle.
lf an electiic cuiient of stiength I iuns upwaid thiough the z-axis, then this cuiient
will cieate a magnetic eld which is given by
(148)
#
B(x, y, z) =

0
I
2
_
_
y/(x
2
+y
2
)
x/(x
2
+y
2
)
0
_
_
.
Again, a constant (
0
) appeais. lf we compaie (148) with (144), then we see that, except
foi the constant factoi
0
I/2 this vectoi eld is a thiee dimensional veision of the one
diawn in Figuie 1 we can iegaid Figuie 1 as a top view of the magnetic eld
#
B of an
electiic cuiient.
140 . VECTOR CALCULUS
3. Line integrals
3.1. Line integrals of functions. lnstead of integiating ovei plane domains, oi ie-
gions in space, it ofen tuins out to be useful to integiate ovei a cuive in the plane, oi a
cuive in space.
lf C is a cuive in the plane, oi in space, (think of a line segment, a ciiculai aic, oi a
fanciei cuive), and if w = f(x, y, z) is a function then the basic pauein foi dening the
integial of f ovei the cuive C is the same as foi all the othei integials we have dened in
the pievious chaptei.
s
Figure 3. Partitioning a curve.
To dene the integial we divide the cuive C into many shoit aics, and label them C
1
,
, C
n
, we choose one sample point (x
k
, y
k
, z
k
) on aic C
k
foi eveiy k = 1, , n, and we
compute the length s
k
of each aic C
k
. With these data we foim the Riemann sum
(149) R = f(x
1
, y
1
, z
1
)s
1
+ +f(x
n
, y
n
, z
n
)s
n
,
and if these Riemann sums conveige as one makes the paitition aibitiaiily ne, then we
call the limit the line integial of f with iespect to aic length ovei the cuive C
(10)

C
f(x, y, z) ds = lim
as the partition
gets finer
n

k=1
f(x
k
, y
k
, z
k
)s
k
.
Te length of the cuive C can be expiessed as a line integial
Length of C =

C
ds.
3.2. How to calculate a line integral. Recall that a cuive C is usually given by a
paiametiization
#
x =
#
x(t) =
_
_
x(t)
y(t)
z(t)
_
_
, (a t b)
also wiiuen as
#
x(t) = x(t)
#
+y(t)
#
+z(t)
#
k.
Given such a paiametiization it is easy to make paititions by just paititioning the
paiametei inteival a t b into many shoit sub inteivals, a = t
0
< t
1
< < t
n
= b.
We could choose the k
th
sample point to be the point
#
x(t
k
). Te length of the aic fiom
#
x(t
k1
) to
#
x(t
k
) is appioximately the same as the distance between these two points (foi
as one makes the paitition nei, the aics become moie and moie like shoit line segments).
Tus we nd
s
k

#
x(t
k
)
#
x(t
k1
) =
_
_
_
_
#
x(t
k
)
#
x(t
k1
)
t
k
_
_
_
_
t
k

#
x

(t
k
) t
k
,
3. LlNE lNTEGRALS 141
#
x(t)
#
x

(t)
C
#
x(t)

#
x
#
x

(t)t
C
#
x(t + t)
Figure 4. A parametrized curve: Le: The vector
#
x

(t) is tangent to the curve at the point


#
x(t). The vector
#
x(t) is the position vector of a point on the curve. Right: Increasing the param-
eter t by a small amount t changes the position vector to
#
x(t +t), causing the corresponding
point on the curve to move by
#
x(t + t)
#
x(t)
#
x

(t)t.
wheie t
k
= t
k
t
k1
. Te Riemann sum foi the line integial is
R
n

k=1
f(
#
x(t
k
))
#
x

(t
k
) t
k
.
As the paitition is made nei the appioximation gets beuei, and in the limit we get
(11)

C
f(
#
x) ds =

b
a
f(
#
x(t))
#
x

(t) dt.
3.3. Example What is the average of f(x, y) = x over the quarter unit circle
in the rst quadrant? Just as with double and tiiple integials, the aveiage of a function
ovei a cuive C is dened to be
Aveiage of f(
#
x) =

C
f(
#
x) ds

C
ds
,
wheie

C
ds is the length of C.
x
y
1 2/
Figure 5. The average x-coordinate on a quarter circle
To compute these integials we must ist paiametiize the cuive. Since the cuive is the
unit ciicle, we can paiametiize points on the cuive by theii polai cooidinate , which
gives us
#
x(t) =
_
cos
sin
_
and thus
#
x

(t) =
_
_
_
_
_
sin
cos
__
_
_
_
= 1.
Teiefoie

C
x ds =

/2
0
cos d = 1.
142 . VECTOR CALCULUS
Te length of the cuive is /2, so the aveiage value of x on the quaitei ciicle is
1
/2
=
2

0.636 619 8 . . . .
4. Problems
1. If C is the quarter of the unit circle that
lies in the first quadrant, then
(a) What is the average distance to the ori-
gin on C?
(b) what is the average polar angle ?
2. (a) Compute the average x and y coordi-
nates of the polygon fromA(1, 0) to B(1, a)
to C(0, a) (a > 0 is a constant; the polygon
has the shape of an upside-down L).
(b) Compute the average polar angle on
the same polygon ABC.
3. Find the average x and y-coordinates on
the part that lies above the x-axis of the cir-
cle with radius R and center at the origin.

4. Compute

C
x ds where C is the parabola
y = x
2
, with 0 x 1.
5. A wire is made in the shape of a helix, of
radius a and height H, with parametrization
#
x(t) =

a cos t
a sin t
Ht/2

(0 t 2).
x
y
z

2
1
1
(cos , sin ,

4
)
Suppose the temperature at (x, y, z) is T =
T0e
z/L
for constants L and T0.
(a) What are the units of a, H, T0, and L?
(b) What values do a and H have for the he-
lix in the drawing?
(c) What is the average temperature on the
wire? (Check that your answer has the right
units.)
5. Line integrals of vector elds
5.1. Denition. If C is a curve in three dimensional space, and
#
F(x, y, z) is a vector
eld, then the line integral of
#
F over C is dened to be
(12)

C
#
F d
#
x = lim
as the partition
gets finer
n

k=1
#
F(x
k
, y
k
, z
k
)
#
x
k
To dene the Riemann sum we have paititioned the cuive into n pieces, (x
k
, y
k
, z
k
) is a
sample point on the k
th
shoit aic in the paitition, and
#
x
k
is the vectoi connecting the
initial and nal points of the k
th
paitition aic. See Figuie (iight).
5.2. Integrals over closed curves. A cuive C is closed if its initial and nal points
coincide. lf we aie integiating a vectoi eld ovei a closed cuive, and if we want to em-
phasize this in the notation, then we can wiite

C
#
F d
#
x, oi

C
Pdx +Qdy +Rdz.
. LlNE lNTEGRALS OF VECTOR FlELDS 143
displacement
vector = x

F
C
F
x
k
th
partition
piece
start
end
Figure 6. Le: The work done by a force
#
F acting on an object is equal to the product of the
length of the displacement
#
x and the magnitude of the force in the direction of the displacement.
If the angle between force and displacement is , then this is W =
#
F
#
x cos =
#
F
#
x.
Right: To define

C
#
F d
#
x we partition the curve into small pieces, and add the work done by
the force
#
F over all partition pieces.
5.3. Dierential form notation for line integrals. Te d
#
x that appeais in line in-
tegials is ofen inteipieted as an innitesimally shoit vectoi connecting two adjacent
points on the cuive C. lts components give us the amounts by which the cooidinates x, y,
and z change as we go fiom one point to the next on the cuive, and theiefoie one ofen
wiites
d
#
x =
_
dx
dy
dz
_
.
lf the vectoi eld
#
F has components
#
F =
_
P
Q
R
_
, wheie P, Q, and R aie functions of
(x, y, z), then the expiession
#
F d
#
x can be wiiuen as
#
F d
#
x =
_
P
Q
R
_

_
dx
dy
dz
_
= P(x, y, z)dx +Q(x, y, z)dy +R(x, y, z)dz.
Because of this the following notation foi line integials is ofen used

C
#
F d
#
x =

C
Pdx +Qdy +Rdz.
Foi instance, the integial

C
xdx +zdy xydz
stands foi the line integial of the vectoi eld
#
F =
_
x
z
xy
_
ovei the cuive C.
Expiessions of the foim Pdx + Qdy + Rdz, such as xdx + z dy xy dz above, aie
called dierential forms.
5.4. e orientation of a curve. Te Riemann sum (12) contains the vectois
#
x
k
,
which connect two adjacent points in oui paitition of the cuive C. Whenevei we have two
points Aand B, theie aie two vectois connecting them, namely
#
AB and
#
BA =
#
AB. To
make suie that the diiection of the vectoi
#
x
k
in the Riemann sum(12) is unambiguous,
we have to agiee on a diiection in which the cuive C is tiaveised. Such a diiection is
144 . VECTOR CALCULUS
called an orientation of the cuive. A cuive can have exactly two oiientations, and to
distinguish between a cuive and the same cuive with the opposite oiientation, one wiites
C = the curve C with its orientation reversed.
lf one ieveises the oiientation of a cuive (e.g. by switching its begin and end points, see
Figuie ), then each vectoi
#
x
k
in the Riemann sum in (12) changes its sign, and as a
iesult the whole Riemann sum changes its sign. ln the limit the integial changes its sign.
Tus we have
(13)

C
#
F d
#
x =

C
#
F d
#
x.
lt is impoitant to iealize that the integial changes its sign heie because it is the line
integial of a vectoi eld. lf w = f(x, y, z) is a function of thiee vaiiables then

C
f(x, y, z) ds = +

C
f(x, y, z) ds.
Foi instance, if f = 1 is constant then

C
ds and

C
ds aie the length of C and C. Since
the length of a cuive is always a positive numbei and does not depend on its oiientation,
we have

C
ds = length of C = length of C =

C
ds.
5.5. Integrating over piecewise dened curves. To compute a line integial it is of-
ten best to stait with a paiametiization of the cuive and use (11). ln piactice it can be
veiy dicult to nd such a paiametiization of the whole cuive, even though the cuive
can be bioken into a few pieces, each of which does have a simple paiametiizations. Foi
instance, the edges of a squaie togethei foim a closed cuive. lt is dicult to nd one
paiametiization foi all foui edges at once, but each edge of the squaie is a simple line
segment foi which one can easily nd a paiametiization. ln this situation one can wiite
A B
D C
C
1
C
2
e curve C=C
1
+C
2
the line integial ovei the whole cuive as a sum of line integials ovei the sepaiate pieces.
Going back to the example of the squaie, we have

C
#
F d
#
x =

AB
#
F d
#
x +

BC
#
F d
#
x +

CD
#
F d
#
x +

DA
#
F d
#
x.
ln geneial, if a cuive C consists of two paits, C
1
and C
2
, then we expiess this by wiiting
C = C
1
+C
2
.
. LlNE lNTEGRALS OF VECTOR FlELDS 14
A line integial ovei the whole cuive is the sum of the line integials ovei the sepaiate
pieces
(14)

C
#
F d
#
x =

C1
#
F d
#
x +

C2
#
F d
#
x.
5.6. e line integral as the integral of the tangential component of a vector eld.
ln the Riemann-sum (12) the k
th
teim contains the vectoi
#
x
k
, which connects two
adjacent points in the paitition of the cuive (see guie , on the iight). We can wiite this
vectoi as the pioduct of a unit vectoi and a positive numbei

#
x
k
=

#
x
k

#
x
k


#
x
k
.
Te vectoi
#
x
k
will be almost tangent to the cuive, and the nei one makes the paitition,
the smallei the angle between
#
x
k
and the tangent to the cuive will be. lf the paitition
is suciently ne, then we will have

#
x
k

#
x
k


#
T
k
,
wheie
#
T
k
is the unit tangent vectoi to the cuive at the point (x
k
, y
k
, z
k
). Fuitheimoie,

#
x
k
s
k
appioximates the length of the k
th
aic in the paitition, and hence we can
wiite the Riemann sum as
n

k=1
#
F(x
k
, y
k
, z
k
)
#
x
k

n

k=1
#
F(x
k
, y
k
, z
k
)
#
T
k
s
k
.
Te sum on the iight is a Riemann sum foi the line integial

C
#
F(
#
x)
#
T ds. Taking the
limit of aibitiaiily ne paititions, we conclude that
(1)

C
#
F d
#
x =

C
#
F
#
T ds.
Since
#
T is a unit vectoi, the quantity
#
F
#
T is the length of component of the vectoi
eld
#
F tangential to the cuive. Equation (1) theiefoie says that the line integial of the
vectoi eld
#
F along a cuive C is the same as the line integial (in the sense of 3.1) of the
tangential component of
#
F along C.
5.7. Example work around a circle. Te foimula (1) foi the line integial is useful
if we know the angle between the foice
#
F and the cuive, and the magnitude of the foice.
Foi instance, considei this pioblem
Compute the work done by the vector eld
#
F(x, y) = x
#
+y
#
= (
x
y
) along the curve
C, where C is some piece of the unit circle in the plane.
We aie asked to compute

C
#
F d
#
x. Te vectoi eld
#
F = (
x
y
) always points away
fiom the oiigin, and thus it is always peipendiculai to the tangent
#
T to the unit ciicle.
(See Figuie .) Hence
#
F
#
T = 0, and we nd that

C
#
F d
#
x =

C
#
F
#
T ds =

0 ds = 0.
Tose who piefei the dieiential foim notation ( .3) can wiite this as

C
x dx +y dy =

C
#
F d
#
x = 0.
14 . VECTOR CALCULUS
Figure 7. Le: the vector field
#
F(x, y) = x
#
+ y
#
from 5.7. Right: the vector field
#
F is
perpendicular to the path C and hence does no work.
5.8. How to compute a line integral. lf a paiametiization of a cuive C is given, so
that the cuive C is the image of
#
x =
#
x(t) =
_
x(t)
y(t)
z(t)
_
, a t b,
then we can paitition the cuive C by paititioning the paiametei inteival a t b by
choosing paitition points a = t
0
< t
1
< . . . < t
n
= b, just as in 3.2. Te k
th
teim in the
Riemann sum (12) dening

C
#
F d
#
x is
#
F(x
k
, y
k
, z
k
)
#
x
k
, with

#
x
k
=
#
x(t
k
)
#
x(t
k1
)
#
x

(t
k
)t
k
(again as in 3.2). Te Riemann sum foi

C
#
F d
#
x is theiefoie
n

k=1
#
F(x
k
, y
k
, z
k
)
#
x
k

n

k=1
#
F(x
k
, y
k
, z
k
)
#
x

(t
k
)t
k
.
Te sum on the iight conveiges to the integial

b
a
#
F(
#
x(t))
#
x

(t) dt, and thus we have


found
(1)

C
#
F d
#
x =

b
a
#
F(x(t), y(t), z(t))
#
x

(t)dt.
We can think of this as a substitution foimula foi integials, in which we substitute
#
x =
#
x(t) in the integial

C
#
F(
#
x) d
#
x, using the iule
d
#
x =
#
x

(t)dt.
5.9. ree examples. Let C
1
be the line segment fiom the oiigin to the point (1, 1),
and let C
2
be the piece of the paiabola y = x
2
between the oiigin and the point (1, 1).
Compute the woik done by the vectoi eld
#
F(x, y) = (
y
x
) along each of these two
paths.
Te two cuives C
1
and C
2
togethei bound a iegion R. Let C
3
be the boundaiy of
this iegion, tiaveised in clockwise diiection, and compute the woik done by
#
F along the
closed cuive C
3
.
To nd these integials we need paiametiizations of the cuives. Foi C
1
we can use
#
x
1
(t) =
_
t
t
_
, 0 t 1,
. LlNE lNTEGRALS OF VECTOR FlELDS 14
C
1
C
2
(1, 1)
#
F
C
1
C
2
(1, 1)
#
F
R
Figure 8. Le: Two dierent paths from the origin to the point (1, 1), and the vector field
#
F.
Right: By reversing the orientation of the second path C2 we can create a closed path that starts
and ends at the origin. This path (C1 combined with C2) is the boundary of the shaded region R,
traversed in the clockwise sense.
and foi C
2
we can use
#
x
2
(t) =
_
t
t
2
_
, 0 t 1.
To show how both notations woik, we will do the ist integial using vectoi notation, and
the second using the dieiential foim notation.
Integral over C
1
. Te ist integial is computed as follows

C1
#
F d
#
x =

C1
_
y
x
_
d
#
x substitute
#
x =
#
x
1
(t) =
_
t
t
_
=

1
t=0
_
t
t
_
. .
#
F

_
1
1
_
dt
. .
d
#
x
since d
#
x =
#
x

1
(t)dt =
_
1
1
_
dt
=

1
0
0 dt
= 0.
Integral over C
2
. Te second integial wiiuen using dieiential foims is

C2
#
F d
#
x =

C2
y dx +x dy.
Heie we substitute the paiametiization of the path
x = x
2
(t) = t and y = y
2
(t) = t
2
,
with
dx = dt, dy = dt
2
= 2t dt,
148 . VECTOR CALCULUS
and we nd

C2
#
F d
#
x =

1
t=0
t
2
dt
. .
ydx
+t 2t dt
. .
xdy
=

1
0
t
2
dt =
1
3
.
Integral over C
3
. We had dened C
3
to be the combination of the cuives C
1
and C
2
(which is C
2
with its oiientation ieveised). Teiefoie

C3
#
F d
#
x =

C1
#
F d
#
x +

C2
#
F d
#
x
=

C1
#
F d
#
x

C2
#
F d
#
x.
We have alieady computed these two integials so theie is no need to do a newintegiation.
Te iesult we aie looking foi is

C3
#
F d
#
x = 0
1
3
=
1
3
.
6. Another Fundamental eorem of Calculus
lf we know the deiivative f

(x) of a function y = f(x) of one vaiiable then the Fun-


damental Teoiem of Calculus tells us that we can iecovei the function by integiating
the deiivative
(1) f(b) = f(a) +

b
a
f

(x) dx
Tis semestei we saw in chaptei lV, 14 that one can do the same foi functions of seveial
vaiiables, i.e. following a somewhat complicated pioceduie one can iecovei a function of
two oi moie vaiiables if one knows it s paitial deiivatives. ln this section we show that
the pioceduie has a much shoitei desciiption in teims of a line integial.
6.1. eorem. For any path C and any dierentiable function f one has
(18) f(B) f(A) =

C
#
f(
#
x) d
#
x,
where A and B are the initial and nal points, respectively, of the path C.
ln dieiential foim notation the same statement is wiiuen as

C
_
f
x
dx +
f
y
dy +
f
z
dz
_
= f(B) f(A).
6.2. Line integral of a gradient does not depend on the path. Te examples in .9
show that the line integial

C
#
F d
#
x of some vectoi eld
#
F noimally depends on the
path C. Howevei, it follows fiom Teoiem .1 that if the vectoi eld
#
F happens to be the
giadient of a function,
#
F =
#
f, then the line integial

C
#
F d
#
x only depends on the
initial and nal points, A and B, of the path C, but not on the way that C gets fiom A to
B.
6.3. Line integral of a gradient around a closed curve vanishes. An impoitant spe-
cial case of Teoiem .1 is that in which the cuive C is closed. lf C is a closed cuive, then
its initial and nal points coincide, so that one always has
(19)

C
#
f(
#
x) d
#
x = 0.
. ANOTHER FUNDAMENTAL THEOREM 149
C
P
Figure 9. If we knowthe gradient of a function, and its value at one point (say, the origin), then we
can compute f(P) at any other point P by choosing a path C from the origin to P, and computing
the line integral of the gradient. We have f(P) = f(0, 0) +

C
#
f d
#
x. It does not maer which
path we choose.
6.4. Proof of the Fundamental eorem. Suppose
#
F =
#
f(x, y, z), and let the
cuive C be paiametiized by
#
x =
#
x(t), a t b. Ten
#
F =
#
f =
_
_
f
x
f
y
f
z
_
_
,
and hence

C
#
f d
#
x =

C
f
x
dx +
f
y
dy +
f
z
dz
=

b
a
_
f
x
(x(t), y(t), z(t)) x

(t)
+
f
y
(x(t), y(t), z(t)) y

(t)
+
f
z
(x(t), y(t), z(t)) z

(t)
_
dt
Te expiession between { } is what the Chain Rule would give us if we tiied to diei-
entiate f(x(t), y(t), z(t)) with iespect to t. So we get

C
#
f d
#
x =

b
a
df(x(t), y(t), z(t))
dt
dt
= f
_
x(b), y(b), z(b)
_
f
_
x(a), y(a), z(a)
_
.
Te point B = (x(b), y(b), z(b)) is the end point of the cuive C, and A = (x(a), y(a), z(a))
is its initial point, so we have found the fundamental theoiem (18).
10 . VECTOR CALCULUS
7. Conservative vector elds
7.1. Denition. A vector eld
#
F is called conservative if one has
(10)

C
#
F d
#
x = 0
for every closed curve C.
Te name conseivative deiives fiom the inteipietation of the integial in (10) as the
amount of woik done by the foice eld
#
F aiound the closed cuive C. As an object moves
thioughout the plane along the cuive C, the foice
#
F acts on it, does woik, and theiefoie
piovides eneigy to the object. Te line integial (10) measuies how much eneigy the
foice adds to the object afei going aiound the cuive C once. Foi a conseivative vectoi
eld the total eneigy piovided to the object is exactly zeio, suggesting that its eneigy is
conserved.
C
#
F
P
Figure 10. As an object moves along the closed curve C the force
#
F acts on it. At times the force
works in the direction of the motion, at other times it works against the motion. If the object starts
at P, and goes around once, will it have gained energy when returning to P?
lt follows fiom .3 that any vectoi eld
#
F that is the giadient of a function is a
conseivative vectoi eld. Te following theoiem says that these aie actually the only
conseivative vectoi elds.
7.2. eorem. If
#
F is a conservative vector eld then there is a function f such that
#
F =
#
f.
lf
#
F =
#
f the function
V = f
is called a potential of the vectoi eld
#
F. Tus a function V is a potential of the vectoi
eld F if
#
F =
#
V.
Te potential V can be found by choosing one xed point A, at which we declaie V (A) =
0, and then computing the line integial
(11) V (P)
def
=

P
A
#
F d
#
x
wheie the integial is a line integial ovei a path fiom the point A to the point P. Te
assumption that
#
F is a conseivative vectoi eld implies that the integial in (11) does
not depend on the path that is chosen.
9. FLUX lNTEGRALS 11
8. Problems
1. Is the gravitational vector field
#
g (x, y) = g
#
e2 =
(
0
g
)
a conservative vector field?
2. Newtons gravitational vector field
#
F(x, y) =
#
x

#
x
3
from 2.3, equation (14) is a conservative
vector field. Show this by finding a potential
of the form f(x, y, z) = K
#
x
a
for suit-
able constants a and K.
3. Reread the section in Chapter IV about
Clairauts theorem. You now have two ways
to tell that a vector field
#
F = P(x, y)
#
e1 +
Q(x, y)
#
e2 cannot be a gradient. Which are
they?
4. (a) Compute the line integrals of the vec-
tor fields
#
F =
(
x
0
)
and
#
G =
(
0
x
)
around the unit circle
#
x() = cos
#
e1 +
sin
#
e2.
(b) Which of the vector fields
#
F or
#
Gcannot
be a gradient, based on your answer to (a)?

(c) Can you conclude from your answer to


(a) that any of the vector fields
#
F or
#
Gmust
be a gradient?
9. Flux integrals
9.1. Denition of ux. ln .1 we dened the integial of a vectoi eld along a cuive
C as the line integial of the tangential component of the vectoi eld. lf the cuive C is not
a space cuive, but lies in the xy-plane, then one can also dene the ux of the vector
eld across the curve.
To dene the ux we must ist choose a unit noimal vectoi
#
N foi the cuive C, i.e. at
each point on C we must choose a vectoi
#
N that has unit length and that is peipendiculai
to the cuive

#
N = 1, and
#
N
#
T = 0.
Once a unit noimal foi the cuive C has been chosen, the ux of a vectoi eld
#
v acioss
the cuive C in the diiection of
#
N is dened to be
(12) Flux =

C
#
v
#
N ds
Te ux integial has a veiy natuial inteipietation if the vectoi eld
#
v is the velocity
eld of a two dimensional uid owing in the plane. lf C is an aic in the plane, and if
#
N
is a unit noimal to C, then uid will ow acioss this aic, and one can ask how much uid
ows acioss the aic in the diiection of
#
N. Te answei is given by the ux integial (12).
Foi an explanation see Figuie 11. Teie the aic is divided into many small sub aics, which
may be consideied neaily stiaight. Duiing a time inteival of length t the uid owing
thiough one such shoit aic sweeps out a paiallelogiam of which one side has length s,
while the othei is given by the vectoi
#
vt. Te aiea of the small paiallelogiam is then
#
N
#
vt s. To get the iate at which uid ows acioss the shoit aic we divide this by
t to get
#
v
#
Ns. Adding ovei all shoit aics that compiise the cuive C leads to the ux
integial (12).
9.2. Flux across a closed curve. lf C is a closed cuive without self inteisections, and
R is the iegion it encloses then the ux of a vectoi eld
#
v acioss the cuive C can again
be inteipieted as the iate at which uid ows acioss the cuive C. Since the cuive now
encloses the bounded iegion R, we can also say that the ux of
#
v acioss the cuive C is
the net iate at which uid leaves the iegion R (piovided
#
N is the outwaid pointing unit
noimal).
12 . VECTOR CALCULUS
C
#
T

#
N
C
#
vt
#
vt
#
vt
#
vt
The unit tangent
and a choice
of unit normals
s
#
N
#
vt
Area =
#
N
#
vts
Water flowing across
the curve C #
N
C
Figure 11. Le: At each point on a plane curve there are two choices of unit normal. If a unit
tangent is given, then the most common choice of normal is to rotate the unit tangent counter-
clockwise by 90

.
Top, right: if water is flowing over the plane with velocity field
#
v, then the rate at which water
flows across the curve C in the direction of the normal
#
N is given by the flux integral (12) of the
velocity.
Boom, right: the amount of water flowing across a short arc of length s on the curve C in
time t is the area of a parallelogram one of whose sides is
#
vt. The area of this parallelogram
is the length of the normal component of
#
vt times s.
Figure 12. The flux of a vector field
#
v across a closed curve measures the rate at which fluid is
flowing out of the enclosed region, if
#
N is the outward normal to the curve.
9.3. Example water under the bridge. An endless iivei R occupies the stiip
R = {(x, y) : 1 y 1}
in the xy-plane. (Te width of the iivei is 2.) Te watei in the iivei ows with velocity
#
v(x, y) = V (1 y
2
)
#
e
1
=
_
V (1 y
2
)
0
_
,
wheie V is a constant (it is the maximal velocity of the watei, which is auained at y = 0,
i.e. in the middle of the iivei, this ow is a two dimensional veision of the Poiseuille ow
fiom 2.2.)
9. FLUX lNTEGRALS 13
#
v = V (1 y
2
)
#
e
1
B
r
i
d
g
e
Figure 13. The shaded region represents the water that passed under the bridge during one time
unit.
estion: How much water ows from le to right through the line segment AB, where
A is the point (0, 1), and B is the point (0, 1)?
Solution: We paiametiize the line segment by
#
x(u) =
_
0
u
_
, 1 u 1.
Noimally one iefeis to the paiametei as time, but since we aie consideiing owing
watei, time is alieady pait of the pioblem. Teiefoie we have called the paiametei on the
cuive u instead of t.
Te line segment is veitical, so the unit noimal is a hoiizontal vectoi of length 1,
i.e. eithei
#
N =
#
e
1
oi
#
N =
#
e
1
. We aie asked to nd how much watei ows from
le to right, so we need the noimal that points to the iight
#
N = +
#
e
1
.
We can now compute the integial. We begin with
ds =
#
x

(u)du =
_
_
_
_
_
0
1
__
_
_
_
du = du,
and
#
N
#
v =
_
1
0
_

_
V (1 y
2
)
0
_
= V (1 y
2
) = V
_
1 u
2
_
,
which gives us

C
#
v
#
N ds =

1
u=1
V (1 u
2
)du = V [u
1
3
u
3
]
1
1
=
4
3
V.
9.4. An expanding ow. A substance, peihaps a uid, oi a gas, is spieading fiom
the oiigin and is moving with velocity eld
#
v = V
_
x/R
y/R
_
=
V
R
#
x,
wheie V and R aie constants V has the units of a velocity, and R has the units of a
length. Te inteipietation of these constants is that V is the speed at which uid paiticles
aie moving when they aie at a distance R fiom the oiigin.
14 . VECTOR CALCULUS
a b
in
out
Figure 14. Le: The vector field
#
v(x, y) =
V
R
(
x
#
e1 + y
#
e2
)
and a circle with radius a. Right:
This vector field cannot describe the flow of an incompressible fluid like water since more fluid
flows out of the circle with radius b than through the circle with radius a: water would have to be
created in the annular region between the two circles.
estion: How much uid ows out of the circle with radius a?
Befoie we compute anything let us decide on the units that the answei should have.
Te question of how much uid ows acioss the cuive is ambiguous since we could
answei in teims of mass (pounds oi kilogiams of uid pei second), oi in teims of volume
(gallons pei second). Tese two aie ielated by the density (pounds pei gallon, kilos pei
litei, etc.) of the substance, and since we do not know anything about the density we
will measuie how much in teims of the volume of substance owing acioss the cuive
pei second. ln fact, since we aie dealing with a two dimensional model (the substance is
owing in the plane iathei than thiee dimensional space, we will measuie the aiea that
ows acioss the cuive instead of the volume.
Solution: We need to compute

Ca
#
v
#
N ds
wheie C
a
is the ciicle with iadius a centeied at the oiigin. Te unit noimal
#
N is the
outwaid pointing noimal, because we aie asked to nd how much uid ows out of the
ciicle.
ln this case
#
N and
#
v aie paiallel so that on the ciicle C
a
we have
#
N
#
v =
#
v = V
a
R
.
Teiefoie the ux integial is veiy simple, namely

Ca
#
v
#
N ds =

Ca
V
a
R
ds = V
a
R

Ca
ds
. .
Length of Ca
= 2
V a
2
R
.
Tis answei is uniealistic if we assume that
#
v ieally is the velocity eld of a noimal uid
(like watei). To see what is wiong we compute how much uid ows thiough ciicles of
dieient iadii a and b. lf a < b then the iate at which uid ows thiough the smallei
10. GREENS THEOREM 1
ciicle is less than the iate at which uid ows out of the laigei ciicle. Te dieience,
(13)
2V
R
_
b
2
a
2
_
,
iepiesents the amount of uid that is (appaiently) being cieated eveiy second in the iing-
shaped iegion between C
a
and C
b
.
Howevei, the computation could apply to a owing gas. ln this case we have computed
the volume of gas that ows acioss each ciicle pei time unit (oi the aiea of gas, because
we aie using a two dimensional model heie). A laigei volume could ow acioss C
b
than
acioss C
a
, piovided the gas is less dense at the ciicle C
b
than it is at the smallei ciicle
C
a
. Tis kind of ieasoning is impoitant foi uid and gas dynamics, and in fact appeais in
many othei bianches in physics.
10. Greens eorem
We have seen that the line integial

C
#
F
#
T ds of a vectoi eld along a closed cuive
vanishes if the vectoi eld happens to be the giadient of some function ( .3), but if the
vectoi eld
#
F is not the giadient of a function then its line integial aiound a closed cuive
need not vanish (see the example in .8). We have also seen examples wheie a ux
integial

C
#
v
#
N ds is non-zeio.
Gieens theoiem ielates the line integial of any vectoi eld on the boundaiy cuive C
of some domain R with a double integial involving paitial deiivatives of the vectoi eld
on the domain R itself. Teie aie two veisions of the theoiem, depending on what kind
of line integial one consideis. Te ist veision is foi woik-type integials, and is best
wiiuen in dieiential foim notation. Te second veision is about ux integials.
10.1. Simply connected domains. ln both veisions of Gieens theoiem one has a
plane iegion R and its boundaiy cuive(s). Te boundaiy cuives of a iegion can be some-
what complicated. Te simplest situation is wheie the domain R is simply connected.
Tis means that R is the iegion enclosed by one cuive C (the cuive C is not allowed to
inteisect itself.) Anothei way of desciibing what a simply connected iegion is, is to say
that a iegion is simply connected if it has no holes. See Figuie 1. lf a domain is not
simply connected, then its boundaiy may consist of moie than one cuive (Figuie 1 on
the iight).
Greens theorem. Let R be a simply connected region in the plane, and let C be the
boundary curve of the region R, with the counter clockwise orientation. Let
#
v(x, y) = P(x, y)
#
e
1
+Q(x, y)
#
e
2
be a vector eld that is dened and has continuous derivatives everywhere in R. en one
has
(14)

C
P(x, y)dx +Q(x, y)dy =

R
_
Q
x

P
y
_
dA.
Te second foim of Gieens Teoiem is about ux integials and is ofen called the
divergence theorem.
1 . VECTOR CALCULUS
R
C
R
C
1
C
2
Figure 15. Le: A simply connected domain, i.e. a domainwithout holes. Right: a non-simply
connected domain, i.e. a domain with a hole. For this non-simply connected domain the boundary
consists of two closed curves rather than one.
Flux version of Greens theorem. Let R be a bounded domain in the plane that is
enclosed by a curve C. If
#
v =
_
P(x, y)
Q(x, y)
_
is a vector eld that is everywhere dened and dierentiable on R, then
(1)

C
#
v
#
N ds =

R
_
P
x
+
Q
y
_
dA
where
#
N is the outward unit normal for the domain R.
Te quantity
(1)
P
x
+
Q
y
is called the divergence of the vectoi eld
#
v, and is wiiuen as div
#
v. lt is one of seveial
combinations of paitial deiivatives of vectoi elds that tuin out to be useful. See 1 foi
moie of these.
10.2. Examples illustrating Greens eorem.
An example where the line integral vanishes on any closed curve. Considei the vectoi
eld
#
F(x, y) = x
#
e
1
+y
#
e
2
= (
x
y
) ,
and let C be a closed cuive in the plane, that encloses the iegion R. Ten the line integial
of
#
F along C is given by

C
#
F d
#
x =

R
_
y
x

x
y
_
dA
=

R
0 dA
= 0.
11. CONSERVATlVE VECTOR FlELDS AND CLAlRAUTS THEOREM 1
We nd that the integial is always zeio, no mauei what the iegion R is. lf we weie lucky
enough to note that this paiticulai vectoi eld is a giadient,
#
F =
_
x
y
_
=
#

_
1
2
x
2
+
1
2
y
2
_
,
then we could also have used (19) to conclude that

C
#
F d
#
x = 0 foi any closed cuive
C.
e expanding gas example again. Let
#
v(x, y) =
V
R
#
x =
V
R
x
#
e
1
+
V
R
y
#
e
2
be the velocity eld of the expanding gas fiom 9.4, and let C be any closed cuive that is
the boundaiy cuive of some domain R. We again compute the ux of the velocity eld
acioss the cuive C in the diiection of its outwaid noimal, but this time we use Gieens
Teoiem.
Figure 16. A gas is flow-
ing in the plane with velocity
field
#
v. At what rate is gas
flowing out of the shaded re-
gion?
The answer turns out to be
proportional to the area of
the region.
Accoiding to Gieens Teoiem we have

C
#
v
#
N ds =

R
div
#
v dA
wheie div
#
v is the diveigence of
#
v, dened in (1). Tus
div
#
v =
v
1
x
+
v
2
y
=
{V x/R}
x
+
{V y/R}
y
=
V
R
+
V
R
= 2
V
R
,
and nally,
(1)

C
#
v
#
N ds =

R
2
V
R
dA = 2
V
R
area of R.
Tis is consistent with oui pievious computation in 9.4. Teie we found in (13) that the
amount of uid pioduced in an annulus of innei and outei iadii a and b is 2
V
R
(b
2
a
2
).
Since the aiea of the annulus is b
2
a
2
this is the same iesult that we just found in
(1).
11. Conservative vector elds and Clairauts theorem
Let
#
F(x, y) = P(x, y)
#
e
1
+Q(x, y)
#
e
2
be a vectoi eld on some iegion R in the plane.
Te fundamental theoiem foi line integials and Claiiauts Teoiem (lll.13.3) piovide con-
nections between conseivative vectoi elds, giadient vectoi elds, and the paitial deiiva-
tives of P and Q. To summaiize what we have seen so fai, iecall that
18 . VECTOR CALCULUS
if
#
F =
#
f for some function f(x, y) then
#
F is conservative,
if
#
F is conservative then
#
F =
#
f for some function f(x, y)
if
#
F =
#
f, then
P
y
=
Q
x
.
Looking at this list we see that the missing statement would be that P
y
= Q
x
implies
that
#
F is a giadient vectoi eld. Tis tuins out only to be tiue if we impose an extia
assumption on the domain R, namely, R must be simply connected (see 10.1.) We
foimulate this moie piecisely in a theoiem.
11.1. eorem. If the domain R is simply connected and if
#
F = P
#
e
1
+ Q
#
e
2
is a
vector eld on R for which
(18)
P
y
=
Q
x
,
then
#
F is conservative, and hence
#
F =
#
f for some function f.
Te pioof is an instiuctive application of Gieens theoiem, so we include it heie
Pvooi. We will show that (18) implies that
#
F is conseivative, i.e. that the line inte-
gial of
#
F aiound any closed cuive in R vanishes.
Let C be a closed cuive in R, and assume to begin with that the cuive does not inteisect
itself. Ten it must enclose a domain D, and since R is simply connected, the domain D
enclosed by the cuive C lies entiiely within R. We can theiefoie apply Gieens theoiem
to the cuive C and conclude that

C
#
F d
#
x =

C
Pdx +Qdy =

D
_
Q
x

P
y
_
dA = 0.
Tis is what we have to show. Foi a complete pioof we would still have to iemove the
assumption we made that the cuive C does not inteisect itself. We will not do this in
detail, but meiely point out that if C has one self inteisection, then one can bieak the
C
C = C
1
+C
2
R
C
D
R
Figure 17. Le: In the proof of Theorem 11.1 the case that C has no self intersections. Right:
the case where C has at least one self intersection.
cuive into pieces, each of which foims a closed cuive without self inteisections, to which
we can apply the pievious aiguments.

12. PROBLEMS 19
12. Problems
1. Use Greens theorem to compute the line
integrals
I =

C
y dx xdy
J =

C
y dx xdy
K =

C
(x sin y) dy
where C is this curve:
1
1
-1
-1
In this drawing the circle has radius 1, and
the height of the triangle is also 1. The orien-
tation of the curve is in the direction of the
arrows.
2. Let Rbe the unit square, i.e. R = {(x, y) :
0 x, y 1}. Let C be the boundary of
the square R traversed in counterclockwise
sense.
(a) Compute

C
2y dx + 3xdy by finding
parametrizations of the edges and applying
the definition of the line integral.
(b) Compute

C
2y dx + 3xdy by applying
Greens theorem and computing a suitable
double integral over R.
3. Compute

C
#
(x
2
y
2
)
#
T ds where C is
the counter clockwise traversed boundary of
the region R defined by x
2
+y
2
< 16.
4. A gas is flowing in the plane with velocity
field
#
v(x, y) =
(
1
y
)
.
(a) Draw the vector field.
(b) Howmuch gas flows out of the rectangle
R defined by 0 < x < L, H < y < H?
5. In each of the following problems C is the counter clockwise traversed boundary of the region
D and you are asked to compute the indicated line integral in two ways: directly, and by using
Greens Theorem.
(a)

C
xy dx +xy dy, R : 0 x, y 1.
(b)

C
e
2x+3y
dx +e
xy
dy, R : 2 x 2, 1 y 1.
(c)

C
#
F
#
Tds,
#
F(x, y) =
(
y cos x
y sin x
)
, R : 0 x /2, 1 y 2.
(d)

C
xy
2
dx +x
2
y dy, R : 0 x 1, 0 y x.
(e)

C
x
2
y dx +xy
2
dy, R : 0 x 1, 0 y x.
(f)

C
x

y dx +

x +y dy, R : 1 x 2, 2x y 4.
(g)

C
(x/y) dx + (2 + 3x) dy, R : 1 x 2, 1 y x
2
.
(h)

C
sin y dx + sin xdy, R : 0 x /2, x y /2.
(i)

C
xln y dx, R : 1 x 2, e
x
y e
x
2
.
(j)

1 +x
2
dy, R : 1 x 1, x
2
y 1.
10 . VECTOR CALCULUS
(k)

C
x
2
y dx xy
2
dy, R : x
2
+y
2
1.
(l)

C
#
v
#
N ds,
#
v(x, y) =
(
xy
2
x
2
y
)
, R : x
2
+y
2
1,
#
N the outward normal.
(m)

C
y
3
dx + 2x
3
dy, R : x
2
+y
2
4.
13. Surfaces and Surface integrals
ln addition to integials ovei two and thiee dimensional domains, and line integials
ovei cuives in the plane oi in space, one can also integiate ovei suifaces. ln this section
we will give a quick intioduction to suifaces and suiface integials. Foi an in-depth study
of the subject, students should considei taking a moie advanced couise on vectoi calculus,
such as Math 321.
Figure 18. Two dimensional surfaces.
13.1. Surfaces and surface pates. We can think of a cuive as the iesult of taking
a line and bending it into some cuived shape. ln the same way a suiface can be thought
of as the iesult of taking a poition of a at plane and bending and twisting it into some
othei shape. Just as some cuives appeai as the boundaiies (oi edges) of plane domains,
some suifaces appeai as boundaiies of domains in thiee dimensional space. Foi example,
the spheie centeied at the oiigin and with iadius R
(19) x
2
+y
2
+z
2
= R
2
is the boundaiy of the thiee dimensional ball it encloses.
Suifaces can be desciibed using dening equations, i.e. by specifying an equation
whose zeio set is the intended suiface. Foi example, the spheie of iadius R has (19) as
dening equation. Foi puiposes of integiation it is moie convenient to iepiesent suifaces
in teims of surface pates. Tese aie the suiface analog of paiametiized cuives.
Denition. A surface patch is a dierentiable vector function of two variables
#
x =
#
x(u, v), a u b, c v d.
13.2. Example the graph of a function is a surface pat. lf z = f(x, y) is a
function dened foi a x b, c y d, then its giaph can be thought of as a suiface
patch, wheie
(10)
#
x(u, v) =
_
_
u
v
f(u, v)
_
_
.
ln woids we take the x and y cooidinates as paiameteis, seuing x = u and y = v. Te
z component of any point on the patch is then z = f(x, y) = f(u, v).
13. SURFACES AND SURFACE lNTEGRALS 11
S
u
v
d
c
a b
(u, v)
#
x(u, v)
v constant,
a u b
u constant,
c v d
Figure 19. A surface patch. A vector function
#
x of two variables u and v maps a piece of the
uv-plane into three dimensional space. The rectangular grid in the uv domain gets mapped onto
a network of curves on the surface patch S. If the rectangular grid in the uv-domain is suiciently
fine, then the corresponding curves on the surface divide the surface patch into small pieces that
are approximately parallelograms.
#
x
(
u
,
v
)
x
y
z
u
v
z = f(u, v)
Figure 20. A graph as a surface patch: the graph of a function z = f(x, y) can be represented
as a surface patch. The vector function
#
x that parametrizes the graph is
#
x(u, v) = u
#
e1 +v
#
e2 +
f(u, v)
#
e3.
13.3. Example the sphere as a surface pat. Te spheie is a two dimensional
suiface, and one way to paiametiize it is to use spheiical cooidinates. Tus
(11)
#
x(, ) =
_
_
Rcos sin
Rsin sin
Rcos
_
_
with
0 , 0 2
is a suiface patch that paiametiizes the spheie it is a paiametiization of the spheie. See
Vl-.2 wheie spheiical cooidinates weie dened, and see Figuie 21 foi a pictuie.
All points with = 0 aie mapped to the noith pole, all points with = coiiespond
to the south pole, the points with =
1
2
foim the equatoi.
12 . VECTOR CALCULUS
Figure 21. Sphere: a piece of the sphere parametrized by the surface patch in (11). Shown is
the piece with 0.1 0.9 and 0.1 1.9.
13.4. Area of a surface pat. Foi any given suiface we can ask what is its suiface
aiea` Te intuitive inteipietation of this could be
(12) how much paint do we need to cover one side of the surface?
oi
(13) how much paper to we need to make the surface?
Neithei inteipietation stands up to closei sciutiny theie aie suifaces, like the Mbius
stiip in Figuie 22, that only have one side, so that questions (12) and (13) will give
Figure 22. A Mbius strip. What is the surface area of this strip, and how many square inches
of paper do we need to make one?
dieient answeis. On the othei hand, while it is possible to take a at piece of papei and
bend it in the shape of a cylindei, a cone, oi a Mbius stiip, it is not possible to bend a at
piece of papei into a spheie without iipping oi stietching it (and thus changing its aiea.)
ln spite of these (and othei) issues we will aigue fiom intuition and deiive a foimula
foi the aiea of a suiface patch. Te stoiy is veiy similai to the deiivation of the aic length
of a paiametiized cuive in ll.13.
lf
#
x(u, v) is a suiface patch with domain a u b, c v d, then we divide its
domain into many small iectangulai pieces of size u by v by paititioning both the u
and v inteivals. See the lef half of Figuie 23. Tis leads to a paititioning of the suiface
patch into small iegions, each of which is appioximately a paiallelogiam (on the iight
13. SURFACES AND SURFACE lNTEGRALS 13
#
N
#
x
v
v
#
x
u
u

A
S
u
v
d
c
a b
(u, v)
#
x(u, v)
v
u
Figure 23. Computing the area and normal to a surface patch. The small rectangle in the uv-
domain gets mapped to a small region on the surface patch. This small region is almost a parallel-
ogram whose sides are given by the vectors
#
xuu and
#
xvv.
in Figuie 23). We compute the aiea of the suiface patch by adding the aieas of all these
smallei pieces. Since any such piece is appioximately a paiallelogiam, we can nd its
aiea by computing the cioss pioduct of the vectois dened by its edges. To nd these
u
0 u
0
+ u
v
0
v
0
+ v
#
x(u
0
+ u, v
0
)
#
x(u
0
, v
0
)
#
x(u
0
, v
0
+ v)
#
x(u
0
, v
0
)
x
z
y
Figure 24. The small blue rectangle in the uv-plane from Figure 23, and its image on the surface
patch.
edges considei Figuie 24. ln a small paitition piece on the suiface patch, the paiametei
u is allowed to vaiy between some value u
0
and u
0
+ u, while the othei paiametei
v is allowed to vaiy between some v
0
and v
0
+ v. One edge of the suiface patch (on
the iight in Figuie 24) iepiesents the change in
#
x(u, v) as u is incieased by u, while
keeping v constant, i.e. it is
#
x(u
0
+ u, v
0
)
#
x(u
0
, v
0
)

#
x
u
(u
0
, v
0
) u.
Te othei edge iepiesents the change in
#
x(u, v) when v is incieased by v and is thus
given by
#
x(u
0
, v
0
+ v)
#
x(u
0
, v
0
)

#
x
v
(u
0
, v
0
) v.
14 . VECTOR CALCULUS
Te aiea of the small paiallelogiam on the suiface patch is theiefoie the length of the
cioss-pioduct of these two vectois
A
_
_
_
_

#
x
u
(u
0
, v
0
) u

#
x
v
(u
0
, v
0
) v
_
_
_
_
=
#
x
u

#
x
v
uv.
Adding this ovei all pieces that make up the suiface patch gives us the total aiea of the
patch
(14) Area of S =

d
c

b
a

#
x
u

#
x
v
dudv.
Te quantity that appeais in this integial appeais in many othei suiface integials and is
called the aiea element of the suiface patch
#
x. Te usual notation foi this quantity is
(1) dA =
#
x
u

#
x
v
dudv,
and it is thought of as the aiea of an innitesimally small piece of the suiface.
13.5. Surface integrals. lf f(x, y, z) is some function that is dened on the suiface
(e.g. a density of some kind), then one denes its integial ovei the suiface to be
(1)

S
f(x, y, z) dA =

d
c

b
a
f(
#
x(u, v))
#
x
u

#
x
v
dudv.
Heie f(
#
x(u, v)) is the iesult of substituting the suiface paiametiization
#
x(u, v) in the
function.
13.6. Unit normal to a surface pat. Fiom Figuies 23 and 24 it appeais that both
vectois
#
x
u
and
#
x
v
aie tangent to the suiface, and that theii cioss pioduct
#
x
u

#
x
v
is
peipendiculai to the suiface. We adopt this as the denition of the tangent plane and
noimal diiection to the suiface
Denition. Let
#
x be a surface patch, and let X
0
be a point with position vector
#
x(u
0
, v
0
)
on the surface patch. If
#
m
def
=
#
x
u
(u
0
, v
0
)
#
x
v
(u
0
, v
0
) =
#
0,
then the vector
#
mdenes the normal direction to the surface. e tangent plane to the surface
through X
0
is the plane with normal vector
#
mthat goes through X
0
.
ln geneial the vectoi
#
mdoes not have unit length, and one ofen needs a noimal vectoi
with length one foi the suiface. Tus one denes
(1)
#
N =
#
m

#
m
=
#
x
u

#
x
v

#
x
u

#
x
v

to be the unit normal foi the suiface patch


#
x. Note that
#
N also is a unit vectoi that
is noimal to the suiface.
14. EXAMPLES 1
13.7. Flux across a surface pat. ln 9 we dened the ux acioss a cuive of a vectoi
eld
#
v (which we think of as the velocity eld of some owing liquid oi gas). Te set-up
in 9 was puiely two dimensional. Now that we have intioduced suiface integials we
can foimulate the same concept foi the moie iealistic situation of a uid owing thiough
thiee dimensional space with velocity eld
#
v. We dene the ux of a vector eld
#
v
across a surface pat to be
(18) Flux =

S
#
v
#
N dA
We have expiessions foi both
#
N and dA (namely, (1) and (1)). When put togethei,
they simplify to
#
N dA =
#
x
u

#
x
v

#
x
u

#
x
v


#
x
u

#
x
v
dudv =
#
x
u

#
x
v
dudv
Teiefoie the ux integial can be computed as
(19)

S
#
v
#
N dA =

d
c

b
a
#
v (
#
x
u

#
x
v
) dudv.
14. Examples
14.1. Area and unit normal of a sphere. Te spheie with iadius R can be iepie-
sented by the suiface patch
(180)
#
x(, ) =
_
_
Rcos sin
Rsin sin
Rcos
_
_
,
foi which we have
#
x

= R
_
_
cos cos
sin cos
sin
_
_
,
#
x

= R
_
_
sin sin
cos sin
0
_
_
and hence
#
x

#
x

= R
2
_
_
cos sin
2

sin sin
2

cos
2
sin cos + sin
2
sin cos
_
_
= R
2
_
_
cos sin
2

sin sin
2

sin cos
_
_
= R
2
sin
_
_
cos sin
sin sin
cos
_
_
.
Te length of
#
x

#
x

is

#
x

#
x

= R
2
sin
_
_
_
_
_
_
_
_
cos sin
sin sin
cos
_
_
_
_
_
_
_
_
= R
2
sin .
and the aiea element on the spheie is
dA = R
2
sin d d.
1 . VECTOR CALCULUS
lntegiating ovei the spheie gives us the aiea of the spheie
(181) Area of sphere =

2
=0


=0
R
2
sin d d = 4R
2
,
which is the familiai answei.
We also nd fiom oui foimula foi
#
x

#
x

that the unit noimal at the point with


position vectoi
#
x(, ) is
#
N =
#
x

#
x

#
x

#
x

=
_
_
cos sin
sin sin
cos
_
_
#
x
#
N =
#
x
R
Figure 25. The unit normal at a point on a sphere centered at the origin has the same direction
as the position vector of the point.
Looking back at the denition (180) of oui suiface patch we see that
#
x = R
#
N, oi,
#
N =
#
x
R
.
ln woids the unit noimal is just the position vectoi
#
x iescaled to length one. Peihaps
with hindsight, this should be cleai fiom a diawing of the spheie (e.g. Figuie 2). ln
many geometiically simple situations it is ofen easiei to guess the unit noimal fiom a
diawing than by going thiough a computation like the one we did in this example. And
sometimes it is even possible to compute the aiea element without woiking out
#
x
u
,
#
x
v
,
and theii cioss pioduct. Foi instance, it is possible to deiive oui foimula foi the aiea
element dA = R
2
sin dd fiom a diawing like Figuie Vl.1.
14.2. e ux of a vector eld across the sphere. We considei the velocity eld
of the expanding gas fiom 9.4 again, except we now considei a gas occupying thiee
dimensional space
#
v =
V
0
R
0
#
x.
Heie V
0
and R
0
aie constants V
0
is the velocity of the gas when it has ieached distance
R
0
fiom the oiigin.
We compute the ux
Flux =

SR
#
v
#
N dA
of this velocity eld acioss the spheie S
R
with iadius R in two ways.
1. THE DlVERGENCE THEOREM AND STOKES THEOREM 1
Fiist, we use the foimula foi
#
N dA
#
N dA =
#
x

#
x

dd = R
2
sin
_
_
cos sin
sin sin
cos
_
_
dd
and compute
Flux =


=0

2
=0
V
0
R
0
#
x R
2
sin
_
_
cos sin
sin sin
cos
_
_
dd
=
V
0
R
0
R
2


=0

2
=0
_
_
Rcos sin
Rsin sin
Rcos
_
_

_
_
cos sin
sin sin
cos
_
_
sin dd
=
V
0
R
0
R
3


=0

2
=0
sin dd
= 4
V
0
R
0
R
3
.
Te second appioach is moie geometiical and avoids computing any integials. We
begin by noting that the unit noimal on the spheie at the point with position vectoi
#
x is
#
N =
#
x/R, and hence that
#
v
#
N =
V
0
R
0
#
x
#
x
R
=
V
0
R
0
#
x
#
x
R
=
V
0
R
0
R
2
R
= V
0
R
R
0
.
Te quantity we want to integiate is theiefoie constant. We nd that the ux is
Flux =

S
V
0
R
R
0
dA = V
0
R
R
0
Area of S = V
0
R
R
0
4R
2
,
which is the same as we got using the ist appioach.
15. e divergence theorem and Stokes theorem
15.1. e divergence theorem in three dimensions. If S is a surface that encloses a
three dimensional region R, if
#
v is a vector eld that is dened and dierentiable on all of
R, and if
#
N is the outward unit normal on S, then
(182)

S
#
v
#
N dA =

R
div
#
v dV
where div
#
v is the divergence of the vector eld
#
v.
By denition the diveigence of the vectoi eld
#
v =
_
_
v
1
(x, y, z)
v
2
(x, y, z)
v
3
(x, y, z)
_
_
= v
1
(x, y, z)
#
e
1
+v
2
(x, y, z)
#
e
2
+v
3
(x, y, z)
#
e
3
is
div
#
v =
v
1
x
+
v
2
y
+
v
3
z
.
18 . VECTOR CALCULUS
15.2. Stokes eorem. If S is a surface patch, if the curve C is the boundary of S, and
if
#
F is a dierentiable vector eld dened everywhere on the surface, then
(183)

C
#
F d
#
x =

S
(curl
#
F)
#
N dA
where the curl of a vector eld is dened by
curl
#
F =
_
_
_
_
F3
y

F2
z
F1
z

F3
x
F2
x

F1
y
_
_
_
_
15.3. Example involving the divergence theorem. We ietuin to the computation
in 14.2 of the ux acioss the spheie S of iadius R of the expanding gas vectoi eld
#
v =
V0
R0
#
x. Accoiding to the diveigence theoiem we have

S
#
v
#
N dA =

B
div
#
v dV
wheie B is the iegion enclosed by the spheie (the ball of iadius R).
Te diveigence of
#
v is easy to compute
div
#
v =

x
_
V
0
x
R
0
_
+

y
_
V
0
y
R
0
_
+

z
_
V
0
z
R
0
_
= 3
V
0
R
0
Since the diveigence is constant its integial ovei B is easy

B
div
#
v dV = 3
V
0
R
0
Volume of B
= 3
V
0
R
0
4
3
R
3
= 4
V
0
R
0
R
3
wheie we have used that the volume of the ball B is
4
3
R
3
.
16.
#
dierentiating vector elds
Te components of a vectoi eld aie functions, and theiefoie we can dieientiate them.
As we have seen in the diveigence theoiem and Stokes theoiem, vaiious combinations
of the paitial deiivatives of vectoi elds tuin out to be veiy useful. Te easiest way to
desciibe these is to intioduce the so-called nabla opeiatoi (oi del opeiatoi) dened by
(184)
#
=
_
_
_

z
_
_
_ =

x
#
+

y
#
+

z
#
k.
At ist sight something is missing heie theie aie paitial deiivatives, but the function
whose deiivative is supposed to be taken is missing. Tis is intentional, and the way
#

is to be inteipieted is as follows
1.
#
DlFFERENTlATlNG VECTOR FlELDS 19
in any formula containing
#
,
the partial derivatives are to be taken of
all functions appearing to the right of the
#
.
Foi example, if f(x, y, z) is a function of (x, y, z), then
#
f =
_
_
_

z
_
_
_ f(x, y, z) =
_
_
_
f
x
(x, y, z)
f
y
(x, y, z)
f
z
(x, y, z)
_
_
_.
So
#
f is the giadient of the function f, just as we had dened it befoie. Sometimes a
dieient notation is used, namely
#
f = grad f.
Next, supposing we have a vectoi eld
#
v =
_
_
P(x, y, z)
Q(x, y, z)
R(x, y, z)
_
_
what would be the iesult of multiplying
#
with
#
v` Since we think of
#
as a vectoi, the
multiplication can be eithei a dot pioduct, oi a cioss pioduct. lf we take the dot pioduct
of
#
and
#
v, we get
#

#
v =
_
_
_

z
_
_
_
_
_
P
Q
R
_
_
=
P
x
+
Q
y
+
R
z
.
Othei commonly used notation foi the diveigence is
div
#
v =
#

#
v.
Tis combination of deiivatives of the components of
#
v is called the divergence of the
vector eld
#
v.
lf we take the cioss pioduct of
#
and
#
v we nd the so-called curl of the vector eld
#
v,
#

#
v =


x
P
#


y
Q
#
k

z
R

=
_
_
R
y
Q
z
P
z
R
x
Q
x
P
y
_
_
.
Te cuil of a vectoi eld
#
v is sometimes called the iotation of
#
v, and the following
alteinative notations also get used
#

#
v = curl
#
v = rot
#
v.
16.1. Example compute the divergence of
#
v(x, y, z) =
#
x and
#
w =
#
x. Te
vectoi elds aie
#
v(x, y, z) =
#
x =
_
_
x
y
z
_
_
, and
#
w(x, y, z) =
#
x =
_
_
x
y
z
_
_
,
in which is the iadius fiom spheiical cooidinates, i.e.
=

x
2
+y
2
+z
2
.
10 . VECTOR CALCULUS
Te diveigence of
#
v is easy
#

#
v =
x
x
+
y
y
+
z
z
= 3, oi div
#
v = 3.
Te diveigence of
#
w is a liule haidei. To begin with, we have
#

#
w =
x
x
+
y
y
+
z
z
.
lt helps to nd the paitial deiivatives of sepaiately. Tey aie

x
=
x

,

y
=
y

,

z
=
z

.
Tese foimulas look nicei in vectoi foim, namely
(18)
#
=
_
_
x/
y/
z/
_
_
=
1

_
_
x
y
z
_
_
=
#
x

.
(Pioblem 1.8 will ask you to check this.) Aimed with these paitial deiivatives we nd
x
x
=
x

x +
x
x
=
x
2

+.
We get similai teims foi
y
y
and
z
z
. Adding these togethei leads to
#

#
w =
x
2

+
y
2

+
z
2

+ 3 =
x
2
+y
2
+z
2

+ 3 =

2

+ 3 = 4.
16.2. Example compute the curl of the Poiseuille ow from 2.2. Te ow is
given in Equation (14). Foi simplicity we will assume R = 1 and v
c
= 1. lf we assume
that the cential axis is the x axis, then the distance r to the cential axis is r =

y
2
+z
2
,
and the velocity eld in the cylindei is given by
#
v(x, y, z) =
_
_
1 y
2
z
2
0
0
_
_
.
lts cuil is then
#

#
v =


x
1 y
2
z
2
#


y
0
#
k

z
0

=
_
_
0
2z
+2y
_
_
16.3. e curl of a gradient always vanishes. lf f(x, y, z) is any function of thiee
vaiiables, then its giadient is a vectoi eld. What is the cuil of this vectoi eld` Te
computation is stiaightfoiwaid,
(18)
#

#
f =
#

_
_
f
x
f
y
f
z
_
_
=


x
f
x
#


y
f
y
#
k

z
f
z

=
_
_
(f
z
)
y
(f
y
)
z
(f
x
)
z
(f
z
)
x
(f
y
)
x
(f
x
)
y
_
_
.
We know that foi any function of seveial vaiiables mixed paitials aie equal (when they
aie continuous), meaning (f
x
)
y
= (f
y
)
x
, etc. Anothei look at the cuil we just computed
tells us that
(18)
#

#
f =
#
0, oi, curl grad f =
#
0,
foi any function f (whose second deiivatives aie continuous).
1. PROBLEMS 11
Function
grad
Vector field
curl
Vector field
div
Function
f
grad

#
(f),
#
v
curl

#
v,
#
w
div

#
w
Figure 26. The three basic operations of vector calculus. If we apply two consecutive operations
in this diagram, we get zero. See Equations (187) and (188).
16.4. e divergence of a curl always vanishes. A computation just like the one
above shows that if we have a vectoi eld
#
v and we compute the diveigence of its cuil,
we always get zeio
(188)
#
(
#

#
v) = 0, oi, div curl
#
v = 0.
Both Equations (18) and (188) aie easy to iemembei in theii
#
foim, if we pietend that
#
is a ieal vectoi.
To get (18) iemembei that the cioss pioduct of any vectoi with itself always vanishes
#
a
#
a =
#
0 foi any
#
a. Te expiession
#

#
f contains the cioss pioduct of
#
with
itself, and so it should vanish. Te aigument doesnt hold because
#
is not ieally a
vectoi, but oui computation (18) shows that the conclusion is tiue anyway.
To get (188), we use that
#
a
#
b is always peipendiculai to
#
b , no mauei what
#
a and
#
b
aie, so that
#
a (
#
a
#
b ) = 0 always holds. Equation (188) is exactly that, with
#
a =
#

and
#
b =
#
v.
16.5. Other combinations of gradient, curl and divergence. Te diveigence of the
giadient does not noimally vanish. lf we expand the denitions we nd
#

#
f =

2
f
x
2
+

2
f
y
2
+

2
f
z
2
.
Tis combination of second deiivatives of a function, which occuis veiy ofen is called
the Laplacian of the function f. Te following notation is used
(f) =
#

#
f = f
xx
+f
yy
+f
zz
.
Te othei combination of deiivatives that one can considei is the cuil of the cuil. lf
#
v is a vectoi eld then its cuil
#

#
v is again a vectoi eld, and thus one can compute
the cuil of the cuil
#
(
#

#
v). Tis combination usually does not vanish.
Foi a given vectoi eld one can also considei its diveigence,
#

#
v, which is a function,
and of which one can compute the giadient,
#
(
#

#
v). Tis quantity usually also does
not vanish.
Teie is a ielation between the cuil of the cuil and the giadient of the diveigence,
which is useful in mathematical physics, and which we state heie foi iefeience only foi
any vectoi eld
#
v one has
#
(
#

#
v) = (
#
v)
#
(
#

#
v).
17. Problems
1. If the central axis of the cylinder in Fig-
ure 2 is the x-axis, and if the vector field is
as given in (145), then write
#
v in terms of
x, y, z instead of r.
12 . VECTOR CALCULUS
2. It is always said that Newton discovered
the inverse square law for gravitation. Ac-
cording to this law the strength of the grav-
itational force is inversely proportional to
the square of the distance to the center of
the Earth. But the exponent in our equa-
tion (14) is three instead of two!
Could this be a dierent law? A typo?
To find out, compute the length
#
F of the
gravitational force in (14).
3. Show that the magnetic field in (148) can
be wrien as
#
B(x, y, z) = C
#
k
#
x

#
k
#
x
n
for some integer n and some constant C.
Find the right n and C.
4. Let
#
a and
#
m be two constant vectors,
with components
#
a =
(
a
1
a
2
a
3
)
, and
#
m =
(
m
1
m
2
m
3
)
.
Let
#
v(x, y, z) be the vector field
#
v = (
#
m
#
x)
#
a.
(a) Write
#
v in terms of its components:
#
v =
(
?
?
?
)
.
(b) Compute
#

#
v.
(c) Compute
#

#
v.
(d) If
#
v is the gradient of some function f,
what can you say about the vectors
#
a and
#
m?
(e) If
#
v is the curl of some vector field
#
w,
what can you say about the vectors
#
a and
#
m?
5. Let
#
a and
#
mbe as in the previous prob-
lem. Consider the vector field
#
v(x, y, z) = e
#
m
#
x #
a
= e
m
1
x+m
2
y+m
3
z
(
a
1
a
2
a
3
)
.
(a) Show by computing the derivatives that
#

(
e
#
m
#
x
)
= e
#
m
#
x #
m.
(b) Compute
#

#
v. (Find the shortest way
to write the answer.)
(c) Compute
#

#
v. Again, simplify your
answer.
(d) Which condition must the vectors
#
a and
#
m satisfy if
#
v is to be divergence free, i.e.
if div
#
v = 0?
(e) Suppose that
#
v =
#
for some func-
tion. What do you know about
#
a and
#
m?

6. If
#
v =
(
P
Q
R
)
is a vector field and f is a
function, then what is
#
v
#
f?
7. Product rules. Let f be a function of three
variables, and let
#
v be a three dimensional
vector field.
(a)
#
(f
#
v) = (
#
f)
#
v +f
#

#
v
(b) Guess a product rule for
#
(f
#
v) and
prove it.
8. In this problem, as in all the problems in
this section, =

x
2
+y
2
+z
2
=
#
x is
the radius in spherical coordinates.
Check the following formulas
#
=
#
x

, and
#

#
x = 3.

9. Use the product rule from Problem 17.7


and the formulas from problem 17.8 to com-
pute the following quantities
(a)
#
(
2 #
x)
(b)
#
x
#

(c) div
#
x

#
x
3
. What does this say about the
Earths gravitational field?
10.
(a) Show that
#
x =
1
2
#
(
2
).
(b) Compute
#

#
x without doing any
derivatives.
(c) Compute
#
(
#
x) using the product
rule from problem 17.7.
11. Compute
#

#
v for the vector field
#
v(x, y, z) =
#
k
#
x.
1. PROBLEMS 13
12. Consider the vector field
#
v(x, y, z) =
n #
x,
where n is a constant. (Both Newtons law
of gravitation and Coulombs law have this
vector field with n = 3.)
(a) Write
#
v(x, y, z) in the form
(

)
, using
only Cartesian coordinates x, y, z.
(b) Compute
#

#
v. (Use one of the product
rules from Problem 17.7; you can also avoid
computing the derivatives of by looking
them up in the text.)
(c) For which value(s) of n does one have
div
#
v = 0?
13. A function of three variables is called ra-
dially symmetric if it only depends on the
radius =

x
2
+y
2
+z
2
, i.e. if it can
be wrien as F() for some function F of
one variable. E.g. f(x, y, z) =
2
, or
g(x, y, z) = e

are radially symmetric


functions.
Find the gradient of a radially symmetric func-
tion F().
(You may want to use x = x/,
etc. from(185) to speed up the computation.)

(a) Let
#
v =
n #
x, as in problem 17.12. Does
there exist a function f(x, y, z) such that
#
v =
#
f? (Hint: try a radially symmetric
function, and use problem 17.13.)
Math 234 Answers and Hints
(I12.3e) (a) 3 (b)
_
_
2
4
4
_
_
(c) 36 (d)
_
_
3
3
3
_
_
(e)
_
_
1
5
5
_
_
(I12.4) Every vector is a position vector. To see of which point it is the position vector translate it so its initial point is
the origin.
Here
#
AB =
_
3
3
_
, so
#
AB is the position vector of the point (3, 3).
(I12.5) One always labels the vertices of a parallelogram counterclockwise (see ).
ABCDis a parallelogramif
#
AB+
#
AD =
#
AC.
#
AB =
_
1
1
_
,
#
AC =
_
2
3
_
,
#
AD =
_
3
1
_
. So
#
AB+
#
AD =
#
AC, and ABCD is not a parallelogram.
(I12.6a) As in the previous problem, we want
#
AB +
#
AD =
#
AC. If D is the point (d
1
, d
2
, d
3
) then
#
AB =
_
_
0
1
1
_
_
,
#
AD =
_
_
d
1
d
2
2
d
3
1
_
_
,
#
AC =
_
_
4
1
3
_
_
, so that
#
AB +
#
AD =
#
AC will hold if d
1
= 4, d
2
= 0 and d
3
= 3.
(I12.6b) Now we want
#
AB +
#
AC =
#
AD, so d
1
= 4, d
2
= 2, d
3
= 5.
(I12.9) Compute the dot product:
#
a
#
b = 2s + 3(1 s) = 3 s. When the dot-product vanishes the vectors are
perpendicular; this happens when s = 3. The angle between the vectors is acute is the dot-product is positive.
This happens when 3 s > 0, i.e. when s < 3.
(I12.11a) The problem is open-ended because it doesnt specify what draw means.
If you are allowed to use a calculator and a protractor, then you could use the dot product to compute
the angle between the two vectors; then, using your protractor, draw two line segments that make this
angle, and mark o lengths 3 and 5 to get the vectors. From the dot-product and the two lengths you find
3 5 cos = 12, so cos =
12
15
= 0.8, which implies = aiccos(0.8) 2.498 . . . radians, or
143.13 . . . degrees.
This turns out to be only half the answer: we have forgoen that the equation cos = 0.8 has many
more solutions than just aiccos(0.8). One other solution is aiccos(0.8). This gives us two vectors
#
b
with
#
b = 5 and
#
b = 5 and
#
a
#
b = 12.
A dierent approach goes like this: you could assume
#
a = 3
#
e
1
, which has length 3, and
#
b =
_
b
1
b
2
_
.
The condition that
#
b have length 5 then says b
2
1
+b
2
2
= 5
2
= 25, while the dot-product is
#
a
#
b = a
1
b
1
+
a
2
b
2
= 3b
1
. Since the dot-product must be 12 we find b
1
=
12
3
= 4. Using the length of
#
b leads to
b
2
=

25 (4)
2
= 3. Thus we find two solutions:
#
b =
_
4
3
_
= 4
#
e
1
3
#
e
2
.
You make the drawing.
(I12.11b) No. The inner product of two vectors is
#
a
#
b =
#
a
#
b cos , and therefore it can never be larger than

#
a
#
b .
1
1 MATH 234 ANSWERS AND HlNTS
(I12.13a) True:
(
#
a +
#
b ) (
#
a
#
b ) = (
#
a +
#
b )
#
a (
#
a +
#
b )
#
b
=
#
a
#
a +
#
b
#
a
#
a
#
b
#
b
#
b
=
#
a
2
+
#
b
2
.
(I12.13b) True: This is Pythagoras theorem. Here is an algebraic derivation:

#
a +
#
b
2
= (
#
a +
#
b ) (
#
a +
#
b )
= (
#
a +
#
b )
#
a + (
#
a +
#
b )
#
b
=
#
a
#
a +
#
b
#
a +
#
a
#
b +
#
b
#
b
=
#
a
2
+ 2
#
a
#
b +
#
b
2
=
#
a
2
+
#
b
2
.
(I12.13c) Not so. The same computation as for the previous problem shows

#
a
#
b
2
= (
#
a
#
b ) (
#
a
#
b )
= (
#
a
#
b )
#
a (
#
a
#
b )
#
b
=
#
a
#
a
#
b
#
a
#
a
#
b +
#
b
#
b
=
#
a
2
2
#
a
#
b +
#
b
2
=
#
a
2
+
#
b
2
.
Therefore

#
a
#
b
2
=
#
a
2

#
b
2
only is true if
#
b =
#
0.
(I12.15a) (
#
a +
#
b )(
#
a +
#
b ) =
#
0
(I12.15b) (
#
a +
#
b +
#
c )(
#
a +
#
b +
#
c ) =
#
0
(I12.15c) (
#
a +
#
b )(
#
a
#
b ) = 2
#
a
#
b .
(I12.16a)
#
a
#
c =
#
a (
#
a
#
b ) = 0, but for the two given vectors in the problem
#
a
#
c = 1 = 0, so there cannot be
a vector
#
b with
#
a
#
b =
#
c as
#
c is not perpendicular to
#
a.
(I12.16b) In this case
#
a
#
c , so the argument from the first part of this problem doesnt rule out that there might be a
solution. So lets try
#
b =
_
b
1
b
2
b
3
_
. Then
#
a
#
b =
_
_
b
2
b
1
2b
3
2b
2
_
_
?
=
#
c =
_
_
1
3
2
_
_
.
Solving this for b
1
, b
2
, and b
3
leads to b
2
= 1, and b
1
2b
3
= 3 as only remaining equation. Since we have
found b
2
there are still two unknowns le. We can choose an arbitrary b
3
and set b
1
= 3 2b
3
, e.g. b
3
= 0
works, provided we choose b
1
= 3.
(II17.7d) (x) =
e
x
_
1 +e
2x
_
3/2
.
To find the point with largest curvature:

(t) =
e
t
_
1 +e
2t
_
5/2
_
1 2e
2t
_
, so the maximal curvature
(smallest radius of curvature) occurs when x =
1
2
ln 2.
(III5.1) d(x, y).
MATH 234 ANSWERS AND HlNTS 1
(III5.2) a < 0, b > 0, c > 0.
(III5.3a) x = 2 for the x-axis, y = 6 for the y-axis, z = 6 for the z-axis.
(III5.3b) z = 3
3
4
x
3
2
y.
(III5.3c)
x
a
+
y
b
+
z
c
= 1 is a nice symmetric way of writing the equation.
(III5.4) The distance is
|c|

1 +a
2
+b
2
.
(III5.5a) This one is already the sum of squares. We dont have to do anything, and can immediately conclude that
f(x, y) > 0 for all (x, y) in the plane except the origin, where x = y = 0 and f(x, y) = 0.
(III5.5b) The square containing x is already complete (no xy terms) and we can immediately factor Q(x, y) = (x
y)(x +y).
(III5.5c) We complete the square:
g(x, y) = (x 2y)
2
y
2
.
We get the dierence of two squares, so we can factor the quadratic form:
g(x, y) = (x 2y y)(x 2y +y) = (x 3y)(x y).
(III5.5d) This one is positive definite:
Q = 9
_
s
2
4st + 9t
2
_
= 9
_
(s 2t)
2
4t
2
+ 9t
2
_
= 9
_
(s 2t)
2
+ 5t
2
_
= 9(s 2t)
2
+ 45t
2
.
(III5.5e) Positive definite:
M =
1
2
_

2
2 + 2
2
_
=
1
2
_
( )
2
+
2
_
.
(III5.5f) This quadratic form has no x
2
term. When that happens you cna immediately factor the form, because all
terms contain y:
Q(x, y) = xy +y
2
= (x +y)y.
This form is indefinite.
(III5.5g) Now this form does have an x
2
term, so we can complete the square if we want to but if we look carefully
then we see that theres not y
2
term. Because of this we can factor out x, and we get
Q = x
2
+ 2xy = x(x + 2y).
The form is indefinite.
What if we dont notice that y
2
is missing and just blindly complete the square? Nothing goes wrong
and we get the same answer:
Q = x
2
+ 2xy = x
2
+ 2xy +y
2
y
2
= (x +y)
2
y
2
= (x +y y)(x +y +y) = x(x + 2y).
We did work too hard though :-(
(III5.6) Complete the square:
Q = (x +ky)
2
k
2
y
2
+y
2
= (x +ky)
2
+ (1 k
2
)y
2
.
If 1 k
2
> 0 then we have the sum of two squares. If 1 k
2
< 0, then we can rewrite Qas the dierence of
two squares
Q = (x +ky)
2
(k
2
1)y
2
= (x +ky)
2

k
2
1y
_
2
which is indefinite. That is all we need to know: we are not actually asked to factor the form when it is
indefinite. But in case youre wondering, the somewhat ugly formula is thus:
Q =
_
x + (k +

k
2
1)y
__
x + (k

k
2
1)y
_
.
18 MATH 234 ANSWERS AND HlNTS
The conclusion is that Q(x, y) is positive definite if 1 < k < 1 and indefinite when k > 1 or k < 1.
In the remaining cases k = 1 we have
Q = (x +ky)
2
k
2
y
2
+y
2
= (x +ky)
2
+ (1 k
2
)y
2
= (x y)
2
,
i.e. the form is a square (it is semidefinite).
(III5.7a) The graph is the saddle surface, the function is defined at all (x, y). The level set is given by xy = c. If c = 0
then this set consists of both branches of the hyperbola y =
c
x
. If c = 0 then xy = 0 is equivalent with x = 0
or y = 0, so the level set is the union of the x-and y-axes.
(III5.7b) z x
2
= 0. Domain R
2
. Graph is a parabolic cylinder and consists of horizontal lines perpendicular to the
xz-plane, going through the parabola y = x
2
in that plane.
Level sets: parallel straight lines x =

z if z > 0, the x axis if z = 0, the empty set if z < 0.


(III5.7c) z
2
x = 0. Implicit function. At least two functions are defined, namely z =

x. Domain: all points (x, y)


with x 0. Graph is half a parabolic cylinder and consists of horizontal lines perpendicular to the xz-plane,
going through the parabola z =

x (or z =

x, depending on which function you choose) in that plane.


Level sets (assuming we choose the function z = +

x): the line x = z


2
if z 0, empty set otherwise.
(III5.7d) z x
2
y
2
= 0. Domain is the whole plane. Graph is a paraboloid of revolution, obtained by rotating the
parabola z = x
2
in the xz-plane around the z axis.
Level sets: circle with radius

z for z > 0, the origin for z = 0 (note: this level set is a point rather than
a curve), empty for z < 0.
(III5.7e) z
2
x
2
y
2
= 0. Implicit function. Domain all of R
2
. Possible functions are z =

x
2
+y
2
. Graph is
the cone obtained by rotating the half line z = x, x 0 in the xz-plane around the z axis (or the half line
z = x, x 0, if you chose z =

x
2
+y
2
.)
Level sets (assuming we choose z = +

x
2
+y
2
): circle with radius z when z > 0, origin when z = 0,
empty when z < 0.
(III5.7f) xyz = 1. Domain the whole plain with the x and y-axes removed, i.e. all points (x, y) with xy = 0. Function
is f(x, y) =
1
xy
. For each y the graph is the hyperbola z = 1/(yx) which is just the standard hyperbola
z = 1/x stretched vertically by a factor 1/y. As y 0 this factor goes to .
(III5.7g) xy/z
2
= 1. Implicit function. Domain first and third quadrants (all points with xy > 0). Functions z =

xy. Cross sections with planes y =constant are half parabolas.


Note: Harder to see, but the surface with equation xy = z
2
is in fact the cone obtained by rotating the
x-axis around the line x = y in the xy-plane.
(III5.8a) x > 0. This one is in the text.
(III5.8b) x < 0.
(III5.8c) x > 0. This is the same region as in part (a): remember that the polar angle is only determined up to a multiple
of 2.
(III5.8d) In the upper half plane, y > 0.
(III5.8e) In the whole plane, except the origin, and the negative x-axis. This formula for the polar angle clearly is valid
in a larger region than the other formulas, but it does not look half as nice.
(III5.9) The level set for c = 24 is the empty set, since it consists of all points on the lake surface where the lake is
24 meters deepi.e. where the water reaches 24meters above the lake.
Similarly, the level set for c = +400 is also empty since the lake is not that deep anywhere.
The level set d
1
(0) consists of those points where the lake is 0meters deep. This is exactly the shore
line.
The level set d
1
(24) consists of all points on the lake surface where the lake is exactly 24meters deep.
Form the map it looks like this happens on two separate curves near the center of the lake.
(III5.10) See 4.
MATH 234 ANSWERS AND HlNTS 19
(III5.11a) The two rectangular strips 3 x 3, 2 y < and 3 x 3, < y 2.
(III5.11b) By definition aicsin(x) is only defined if 1 x 1. For aicsin(x
2
+ y
2
2) to be defined, we must
therefore have 1 x
2
+y
2
2 1, i.e. 1 x
2
+y
2
3.
The domain of this function is the ring-shaped region between the circles with radii 1 and

3, both
centered at the origin. Circles are included in the domain.
(III5.11c) The way this function is wrien both

x and

y must be defined, so the domain consists o all (x, y) with


x 0 and y 0.
(III5.11d)

xy must exist, which happens for all (x, y) in the first and third quadrants (axes included.)
(III5.11f) The region in the plane given by x
2
+ 4y
2
16, which is the region enclosed by an ellipse with major axis of
length 4, along the x axis, and minor axis of length 2 along the y-axis. The ellipse is included.
(III5.12) The level sets of the function whose graph is a cone are equally spaced circles (the level set at level c is a circle
with radius c). Hence the one on the right corresponds to the cone, and the one on the le corresponds to the
paraboloid.
(III5.13a) (0,
1
2
) is in the square Q, so it is the point closest to (0,
1
2
).
The point (0, 1) on the top edge of the square is closest to (0, 2).
The corner point (1, 1) is closest to (3, 4).
(III5.13b) f(0,
1
2
) = 0; f(0, 2) = 1 and f(3, 4)) =

2
2
+ 3
2
=

13.
(III5.13c) The zero set of f is the square Q.
(III5.13d) The level set at level 1 is empty. The others are rounded rectangles, see this drawing, in which the square
is grey, the dashed lines are given by x = 1 or y = 1.
x
y
(III5.13e) The lines x = 1 and y = 1 divide the plane into nine regions. On each region the function is given by a
dierent formula. Here they are:
f(x, y) if
0 (x, y) in Q
x 1 x 1, |y| 1
y 1 |x| 1, y 1
x 1 x 1, |y| 1
y 1 |x| 1, y 1

(x 1)
2
+ (y 1)
2
x 1 and y 1

(x 1)
2
+ (y + 1)
2
x 1 and y 1

(x + 1)
2
+ (y 1)
2
x 1 and y 1

(x + 1)
2
+ (y + 1)
2
x 1 & y 1
180 MATH 234 ANSWERS AND HlNTS
(III5.14a) At time t we have a line through the origin with slope sin t. As time progresses this lines turns up and down,
and up and down, etc.
(III5.14b) Same as previous problem, but twice as fast.
(III5.14c) At all times one sees the graph of y = sin x stretched vertically by a factor t.
(III5.14d) Same as previous problem, but twice as fast.
(III5.14e) The graph of y = sin 2x stretched vertically by a factor t.
(III5.14f) Parabola with its minimum on the x-axis at x = t. So we see the parabola y = x
2
translating from the le to
the right with constant speed 1.
(III5.14g) Parabola with its minimum on the x-axis at x = sin t. So we see the parabola y = x
2
translating back and
forth horizontally every 2 time units.
(III5.14j) At time t we see Agnesis witch, i.e. the graph y = a/(1 +x
2
) with amplitude a = 1/(1 +t
2
). Thus we see
a bump whcich starts out small at t = , grows to its maximal size at time t = 0, and then decays again,
until it vanishes at t = +.
(III5.16) The graph of y = g(x a) is obtained from the graph of y = g(x) by translating the graph of y = g(x) by a
units to the right.
Hence the graph of g(x ct) is the graph of g(x) translated by ct units to the right. As time changes
the graph of g(x ct) therefore moves with velocity c to the right.
(III5.17) If you know the graph of a function y = g(x), then you get the graph of y = cg(x) by stretching the graph
of g vertically by a factor c (here c is a constant.) If you allow this constant to depend on time, e.g. as in this
problem by seing c = cos(t), then the movie you get is of a version of the graph of g which is growing
and shrinking vertically.
y=cos(t)g(x)
y=g(x)
(IV3.2b) 2xy sin(x
2
y), x
2
sin(x
2
y) + 3y
2
(IV3.2c) (y
2
x
2
y)/(x
2
+y)
2
, x
3
/(x
2
+y)
2
(IV3.2g) 2xe
x
2
+y
2
, 2ye
x
2
+y
2
(IV3.2h) y ln(xy) +y, xln(xy) +x
(IV3.2i) x/

1 x
2
y
2
, y/

1 x
2
y
2
(IV3.2l) tan y, x/ cos
2
y
(IV3.2m) 1/(x
2
y), 1/(xy
2
)
(IV3.4a)

x
=
y
x
2
+y
2
,

x
=
x
x
2
+y
2
.
MATH 234 ANSWERS AND HlNTS 181
(IV3.5) The distance to the origin is exactly the radius in polar coordinates, so f(x, y) =

x
2
+y
2
, and
fx =
x

x
2
+y
2
, fy =
y

x
2
+y
2
.
This is the same as in problem 3.3. The only quantity that we did not compute before is
_
fx
_
2
+
_
fy
_
2
=
x
2
x
2
+y
2
+
y
2
x
2
+y
2
=
x
2
+y
2
x
2
+y
2
= 1.
(IV3.6a)
z
x
= f

(x)g(y),
z
y
= f(x)g

(y).
(IV3.6b)
z
x
= yf

(xy),
z
y
= xf

(xy).
(IV3.6c)
z
x
=
1
y
f

(
x
y
),
z
y
=
x
y
2
f

(
x
y
).
(IV7.1a) The linear approximation formula is equation (0), in which x
0
= a = 3, y
0
= b = 1, and x = xa = x3,
y = y b = y 1. So for this problem the linear approximation of f(x, y) = xy
2
at (3, 1) is
f(x, y) 3 + (x 3) + 6(y 1) = x + 6y 6.
This approximation is only expected to be good when (x, y) is close to (3, 1). The approximation contains an
error which is small compared to |x 3| and |y 1|.
FAQ: What is the relation between the linear approximation and the tangent plane?
Answer: They are very closely related: the tangent plane is the graph of the linear approximation. The linear
approximation is the equation for the tangent plane. To compute either you have to do the same thing.
(IV7.1b) x/y
2
3 + (x 3) 6(y 1) = x 6y + 6 when x is close to 3 and y is close to 1.
(IV7.1c) sin x + cos y 1 + (1)(x ) + (0)(y ) = 1 x when x is close to and y is close to .
(IV7.1d)
xy
x+y

3
4
+
1
16
(x 3) +
9
16
(y 1) when x is close to 3 and y is close to 1.
(IV7.2) z = 1
(IV7.3) z = 6(x 3) + 3(y 1) + 10
(IV7.4) z = (x 2) + 4(y 1/2)
(IV7.5a) Solve for z: z =

2x
2
+ 3y
2
4. In this problem we are looking at the point (1, 1, 1) so we have the
graph of z = f(x, y) =

2x
2
+ 3y
2
4. The partials are
f
x
=
2x

2x
2
+ 3y
2
4
,
f
y
=
3y

2x
2
+ 3y
2
4
so that, at (1, 1, 1) you get fx = 2, fy = 3. There for the equation for the tangent plane is z =
2(x 1) 3(y 1) 1
(IV7.6a) The tangent plane has equation z = z
0
+ A(x x
0
) + B(y y
0
). By puing the variables x, y, z on one
side, and all the constants on the other, you can write this as
Ax +By z = Ax
0
+By
0
z
0
.
This is the equation for a plane whose normal is
#
n =
_
A
B
1
_
. Any other multiple of this vector is also a valid
normal to the plane, in particular,
_
A
B
+1
_
is OK.
(IV7.6b) We want a normal to the graph of z = f(x, y) =
1
2
x
2
+2y
2
at the point P. By the previous problem a normal
is given by
#
n =
_
fx(2,1)
fy(2,1)
1
_
=
_
2
4
1
_
.
A line through P in the direction of
#
n is given by
#
r (t) =
_
2
1
4
_
+t
_
2
4
1
_
(IV7.7) Below you see the graph of a function and two (solid) lines which are tangent to the graph. On one line you
have x = a (hence constant), and its slope is fx(a, b); on the other you have y = b, and it has slope fy(a, b).
182 MATH 234 ANSWERS AND HlNTS
The tangent plane to the graph (not drawn here, but see Figure 4 in the notes) is the plane containing the two
lines in the drawing.
(IV7.8) The function is f(x, y) = xln(xy). We have f(2,
1
2
) = 2 ln(2
1
2
) = ln 1 = 0. The gradient of the function
is
#
f =
_
ln(xy)+1
x/y
_
. At the point (2,
1
2
) this is
#
f =
_
1
4
_
, so the linear approximation is
f(x, y) f(2,
1
2
) + 1 (x 2) + 4 (y
1
2
),
i.e.
f(x, y) 1(x 2) + 4(y
1
2
).
(This is also the answer to problem 7.4.)
Here we dont want to describe the tangent plan, but we want to find the value of f(x, y) for (x, y) =
(1.98, 0.4). Substituting these values of x and y in the linear approximation we get f(1.98, 0.4) (1.98
2) + 4(0.4 0.5) = 0.42.
This is only an approximation, and you wonder how good it is. We have x = 1.98 2 = 0.02, and
y = 0.4
1
2
= 0.1are these numbers small? To find the error in the approximation you could use
a Lagrange-type remainder term, but thats not part of math 234. Instead we grab a calculator and compute
f(1.98, 0.4) = 1.98 ln(1.98 0.4) = 0.46172 . So our linear approximation formula is o by 0.04 .
(IV7.9a) The x-and y-axes.
(IV7.9b) The heights are the z-coordinates, so z = xy and z = 2 +x + 2y. The dierence is
z z = xy (2 +x + 2y) = xy x 2y + 2.
(IV7.10a) The tangent plane has equation z = ab +b(x a) +a(y b) = bx +ay ab.
MATH 234 ANSWERS AND HlNTS 183
(IV7.10b) The point (x, y, z) lies on the intersection if z = xy and z = bx + ay ab. Therefore x and y must satisfy
xy bx ay +ab = 0. This equation factors as follows:
xy bx ay +ab = (x a)(y b) = 0,
so that the intersection contains the line x = a, z = ay, and also the line y = b, z = bx.
(IV10.2)
(f+g)
x
= fx +gx, and
(f+g)
y
= fy +gy, so
_
_
_
(f+g)
x
(f+g)
y
_
_
_ =
_
fx +gx
fy +gy
_
=
_
fx
fy
_
+
_
gx
gy
_
Hence
#
(f +g) =
#
f +
#
g.
The product and quotient rules follow in the same way.
(IV10.3b) The gradient is
#
f =
_
2x
8y
_
. This vector is parallel to
_
1
1
_
if there is a number s such that
#
f = s
_
1
1
_
, i.e.
_
fx
fy
_
= (
s
s
). This happens if fx(x, y) = fy(x, y). From our computation of the partial derivatives of f we
find that
#
f is parallel to
_
1
1
_
when 2x = 8y. This happens at every point on the line y =
1
4
x.
We are asked which points on the level set f = 4 satisfy this condition, so we must find where the line
y =
1
4
x intersects the level set x
2
+ 4y
2
= 4. Solving the two equations gives two points (
4
5

5,
1
5

5) and
(
4
5

5,
1
5

5).
(IV10.3c)
#
g =
_
4y
2
8xy
_
. This is parallel to
_
1
1
_
when y = 2x. This line intersects the level set g = 4 in the point
(
1
2
3

2,
3

2).
Note: when you solve the equations
#
g = (
s
s
), you find y = 2x, but also the line y = 0 (x-axis). On
this line the gradient actually vanishes, i.e.
#
g =
#
0 and has no direction, so you cant really say it is parallel
to
_
1
1
_
.
(IV10.4a) Its a paraboloid of revolution.
(IV10.4b)
#
f =
_
2x
2y
2
_
= s
_
1
1
2
_
if 2 = 2s, i.e. s = 1. This then implies 2x = 1, 2y = 1, so that x = y =
1
2
.
Since the point has to lie on the zero set of f, we find z =
1
2
(x
2
+y
2
) =
1
4
.
(IV10.5a) At (2, 1) the gradient is
#
T =
_
2x
9y
2
_
=
_
4
9
_
. To cool o as fast as possible the bug should go in the
opposite direction, i.e. in the direction of
_
4
9
_
, or any positive multiple of this vector.
(IV10.5b) At (1, 3) the gradient is
#
T =
_
2
81
_
. To keep its temperature constant the bug should walk in any direction
perpendicular to the gradient. The vector
_
81
2
_
is perpendicular to the gradient, so the bug should go in the
direction of
_
81
2
_
or the opposite direction,
_
81
2
_
.
Any non-zero multiple of
_
81
2
_
is also a valid answer, since we can only give the direction and not the
speed.
Remember: the vector
_
b
a
_
is perpendicular to (
a
b
).
(IV10.6) The zero set doesnt have to be a curve. For example the zero set of the function f(x, y) =distance from (x, y)
to the square Q (Problems 5.13 and 3.7) is the whole square Q.
(IV10.7)
#
f is larger at the top right, because there the function f changes faster.
(IV10.8a) The gradient at the origin is the zero vector. This was explained in the text.
(IV10.8b) The function increases in the direction of the gradient. Since it vanishes on the curve in Figure 8, the function
will be positive in the region above the curve, and it will be negative both below the curve and inside the lile
loop.
(IV10.12b) The result of a rather long calculation is that
#
f = 1 everywhere outside the square, and
#
f = 0
inside the square (because f is constant in the square.)
184 MATH 234 ANSWERS AND HlNTS
(IV10.14) ax +by +cz = R
2
.
(IV12.1) 4xt cos(x
2
+y
2
) + 6yt
2
cos(x
2
+y
2
)
(IV12.2) 2xy cos t + 2x
2
t
(IV12.3) 2xyt cos(st) + 2x
2
s, 2xys cos(st) + 2x
2
t
(IV12.4) 2xy
2
t 4yx
2
s, 2xy
2
s + 4yx
2
t
(IV12.6a)
T
B
Y
= sin
T
A
x
+ cos
T
A
y
.
(IV12.6b) Take the formulas for
T
B
X
and
T
B
Y
and work out the right hand side in this problem.
(IV12.9a)
#
E =
#
ln r =
1
r
2
_
x
y
_
.
(IV12.9b)
#
E = 1/r =
1

x
2
+y
2
.
(IV12.13a) Height = (x
2
y
2
)/(x
2
+y
2
)
(IV12.13b) Height = sin 2.
(IV12.13c) Height = cos 2.
(IV15.1) fx = 3x
2
y
2
, fy = 2x
3
y + 5y
4
, fxx = 6xy
2
, fyy = 2x
3
+ 20y
3
, fxy = 6x
2
y
(IV15.2) fx = 12x
2
+y
2
, fy = 2xy, fxx = 24x, fyy = 2x, fxy = 2y
(IV15.3) fx = sin y, fy = xcos y, fxx = 0, fyy = xsin y, fxy = cos y
(IV15.9) A function of two variables has
fxx, fxy = fyx, fyy,
so it has three dierent partial derivatives of second order.
A function of three variables has these partial derivatives:
fxx fxy fxz
fyx fyy fyz
fzx fzy fzz
The ones below the diagonal are the same as corresponding derivatives above the diagonal, so there are only
six dierent partial derivatives of second order, namely these:
fxx fxy fxz
fyy fyz
fzz
A function of two variables has
fxxx,
fxxy = fxyx = fyxx,
fxyy = fyxy = fyyx,
and fyyy
so four dierent partial derivatives of third order.
(IV15.15a) We have g(u, v) = f(u +v, u v), so
g
u
=
f
x
(u +v)
u
+
f
y
(u v)
u
= fx(u +v, u v) +fy(u +v, u v).
Similarly,
g
v
= fx(u +v, u v) fy(u +v, u v).
Dierentiate again to get

2
g
u
2
= fxx(u +v, u v) + 2fxy(u +v, u v) +fyy(u +v, u v).
MATH 234 ANSWERS AND HlNTS 18
(IV15.15b)

2
g
v
2
= fxx(u +v, u v) 2fxy(u +v, u v) +fyy(u +v, u v)
Note that this is almost the same as

2
g
u
2
: the only change is in the minus sign before fxy.
(IV15.15c)

2
g
uv
= fxx(u +v, u v) fyy(u +v, u v)
(IV15.15d)

2
g
u
2


2
g
v
2
= 4fxy
(IV15.15e)

2
g
u
2
+

2
g
v
2
= 2
_
fxx +fyy
_
.
(V3.1a) If y = 0 then you can increase x
2
x
3
y
2
by seing y = 0. To put it dierently, no maer what you choose
for y, you always have
f(x, y) = x
2
x
3
y
2
x
2
x
3
= f(x, 0).
(V3.1b) The maximum has to appear on the x axis, so the question is which x 0 maximizes f(x, 0) = x
2
x
3
?
This is a Math 221 question. The answer is at x = 2/3.
(V3.1c) No, lim
x
f(x, y) = +, so f has no largest value.
(V3.3)
1
(
3
4
,
3
8

3)
(
3
4
,
3
8

3)
The quantity 4(x
3
x
4
) = 4x
3
(1x) is negative when x < 0 or x > 1, so the region is confined to the
vertical strip 0 x 1. Within this strip Ris comprised of those points which satisfy

4(x
3
x
4
) y
+

4(x
3
x
4
). The largest x value is aained at the point with x = 1, where y = 0, so, at the point (1, 0).
The smallest x value is aained at the point (0, 0). The largest y value is aained at the point where y
2
=
4x
3
4x
4
is maximal. This happens when x =
3
4
, and the largest y value is therefore

4[(3/4)
3
(3/4)
4
] =
3
8

3. The smallest y value also occurs at x =


3
4
and is given by y =
3
8

3.
(V6.1a) fx = 2x 2, fy = 8y + 8, fxx = 2, fxy = 0, fyy = 8.
There is exactly one critical point, at (x, y) = (1, 1).
The 2nd order Taylor expansion at this point is
f(1 + x, 1 + y) = f(1, 1) + (x)
2
+ 4(y)
2
+
The quadratic part is positive definite, therefore f has a local minimum at (1, 1).
(V6.1b) fx = 2x + 6, fy = 2y 10, fxx = 2, fxy = 0, fyy = 2.
There is exactly one critical point, at (x, y) = (3, 5).
The 2nd order Taylor expansion at this point is
f(3 + x, 5 + y) = f(3, 5) + (x)
2
(y)
2
+
= f(3, 5) +
_
x y
__
x + y
_
+
The quadratic part factors, therefore f has a saddle point at (3, 5). The level set near the critical point
consists of two crossing curves whose tangents are given by the equations x = y and x = y. Since
18 MATH 234 ANSWERS AND HlNTS
x = x a = x + 3 and y = y b = y + 5, the two tangent lines have equations x + 3 = y + 5 and
x + 3 = (y + 5).
Critical point and level set
near the critical point.
(V6.1c) fx = 2x + 4y, fy = 4x + 2y, fxx = 2, fxy = 4, fyy = 2. There is one critical point: (x, y) = (2, 1).
The 2nd order Taylor expansion at this point is
f(2 + x, 1 + y) = f(2, 1) + (x)
2
+ 4xx + (y)
2
+
= f(2, 1) +
_
x + 2y
_
2
3(y)
2
+
= f(2, 1) +
_
x + (2 +

3)y
__
x + (2

3)y
_
+
The quadratic part factors, therefore f has a saddle point at (2, 1). The level set near the critical point
Critical point and level set
near the critical point.
consists of two crossing curves whose tangents are given by the equations x = (2 +

3)y and x =
(2

3)y. Since x = x a = x 2 and y = y b = y + 1, the two tangent lines have equations


x 2 = (2 +

3)(y + 1) and x 2 = (2

3)(y + 1).
(V6.1d) fx = 2x y 5, fy = x + 4y + 6, fxx = 2, fxy = 1, fyy = 4.
There is again one critical point: x = 2, y = 1.
The 2nd order Taylor expansion at this point is
f(2 + x, 1 + y) = f(2, 1) + (x)
2
xx + 2(y)
2
+
= f(2, 1) +
_
x
1
2
y
_
2
+
7
4
(y)
2
+
The second order part of the Taylor expansion is positive, so (2, 1) is a local minimum.
(V6.1e) fx = 36x + 4x
3
, fy = 2y, fxx = 36 + 12x
2
, fxy = 0, fyy = 2.
The equation fx = 0 has three solutions, x = 0 and x = 3. The equation fy = 0 has only one solution
y = 0. Therefore there are three critical points, the origin and the points (3, 0).
The taylor expansions at these points are
f(x, y) = f(0, 0) 18(x)
2
+ (y)
2
+
= f(0, 0) +
_
y

18x
__
y +

18x
_
+
f(3 + x, y) = f(3, 0) + 36(x)
2
+ (y)
2
+
f(3 + x, y) = f(3, 0) + 36(x)
2
+ (y)
2
+
The second order terms in the Taylor expansions at (3, 0) and at (3, 0) are both positive for all x and y, so
both points (3, 0) are local minima. The second order part of the expansion at the origin factors and hence the
origin is a saddle point. The tangents to the zeroset at the origin are the lines y =

18x = 3

2x.
Since here x = xa = x, and y = y, the tangents are the lines through the origin given by y = 3

2x.
You can try to draw the zeroset of this function and analyze it in the same way as the fishy example in
4.4. The zeroset of f consists of the graphs of y =

18x
2
x
4
= |x|

18 x
2
. It looks like a squashed
or a buerfly (you decide.)
-3 3
Critical points and zero set.
(V6.1f) There are nine critical points. Four global minima at (3,

3), four saddle points at (0,

3) and (3, 0)
respectively, and finally, a local but not global maximum at the origin.
(V6.1g) critical point at (1, 1/6) fx = 4 4x, fy = 1 6y, fxx = 4, fxy = 0, fyy = 6.
Second order Taylor expansion at the critical point:
f(1 + x,
1
6
+ y) = f(1,
1
6
) 2(x)
2
3(y)
2
+
The second order terms are always negative so (1,
1
6
) is a local maximum.
(V6.1h) The derivatives are:
fx = 4y 2xy 2y
2
, fy = 4x x
2
4xy, fxx = 2y, fxy = 4 2x 4y, fyy = 4x.
This function is given in factored form, so without solving the equations fx = 0, fy = 0 you can say the
following about this problem. The zero set consists of the three lines: the y-axis (x = 0), the x-axis (y = 0)
and the line with equation 4 x 2y = 0. It follows that the intersection points (0, 0), (4, 0), and (0, 2) of
these lines are saddle points. Since f > 0 in the triangle formed by the three lines this triangle must contain
at least one local maximum.
MATH 234 ANSWERS AND HlNTS 18
To find all critical points solve these equations:
fx = 4y 2xy 2y
2
= 0 y = 0 or 4 2x 2y = 0
and
fy = 4x x
2
4xy = 0 x = 0 or 4 x 4y = 0
Since both equations fx = 0 and fy = 0 lead to two possibilities, we have to consider 2 2 = 4 cases:
y = 0 & x = 0: This tells us the origin is a critical point
y = 0 & 4 x 4y = 0: Solving these equations leads to x = 4, y = 0, so (4, 0) is a critical point.
4 2x 2y = 0 & x = 0: Solve and you find that (0, 2) is a critical point.
4 2x 2y = 0 & 4 x 4y = 0: Solve these equations and you get (x, y) = (
4
3
,
2
3
).
The first three critical points are the saddle points we predicted. The fourth critical point must be a local
maximum, since there has to be one in the triangle, and of all the critical points we have found the others are
all saddle points.
(V6.1i) Two saddle points: (0, 0) and (1, 1).
(V6.1j) Two saddle points: (2, 2) and (2, 2)
(V6.1l) The origin. Neither a local max, min, nor saddle. The graph of this function is called the Monkey Saddle as it
accommodates two legs and a tail too. Draw it in your graphing program to see this.
(V6.1m) Zero set is the parabola with equation x = y
2
, and the line x = 1. They intersect at (1, 1), so the function
has two saddle points (1, 1) and (1, 1). The region between the line x = 1 and the parabola must contain
local minimum. It is located at (
1
2
, 0).
(V6.1n) Two saddle points : (2, 2) and (2, 2). Yes, this problem appeared twice.
(V6.1o) All points on the y-axis are critical points. They are all global minima, but the second derivative test doesnt
tell you so.
(V6.1p) All points on the y-axis are again critical points. Those with y > 0 are local minima, those with y < 0 are
local maxima, and the origin is neither. The second derivative test applies to none of these points.
(V6.1q) All points on the unit circle are global minima, because the function vanishes there, and is positive everywhere
else. The origin is a local maximum. The 2nd derivative test applies to the origin, but not to any of the other
critical points.
(V6.1r) All points on the y-axis are again critical points. Those with y > 0 are local minima, those with y < 0 are
local maxima, and the origin is neither. The second derivative test applies to none of these points.
(V6.5a) (3, 4/3)
(V6.5c) x = (a +c +e)/3, y = (b +d +f)/3.
(V6.6) You have to show that fx(a, b) = fy(a, b) = 0. By the product rule fx(a, b) = gx(a, b)h(a, b) +
g(a, b)hx(a, b). Since both g(a, b) = 0 and h(a, b) = 0, it follows that fx(a, b) = 0. The same reason-
ing applies to fy(a, b).
(V8.1a) One variable calculus! There is only one variable, a, and we must solve E

(a) = 0.
(V8.1b) a = (x
1
+ +x
N
)/N, i.e. the average provides the best fit.
(V8.2a) Three: a, b, and c.
188 MATH 234 ANSWERS AND HlNTS
(V8.2b) The equations for (a, b, c) are:
(

x
4
k
) a + (

x
3
k
) b + (

x
2
k
) c =

x
2
k
y
k
(

x
3
k
) a + (

x
2
k
) b + (

x
k
) c =

x
k
y
k
(

x
2
k
) a + (

x
k
) b + N c =

y
k
(V8.3) The equations are
(

x
2
k
) a + (

x
k
y
k
) b + (

x
k
) c =

x
k
z
k
(

x
k
y
k
) a + (

y
2
k
) b + (

y
k
) c =

y
k
z
k
(

x
k
) a + (

y
k
) b + N c =

z
k
(V10.1) The two x and ys are dierent. The first set of (x, y) are
x = x 0, y = y 0,
(0, 0) being the coordinates of the first critical point we studied. The second set of (x, y) is
x = x
2
3
, y = y 0,
where (
2
3
, 0) is the other critical point. In a drawing:
Critical point at (
2
/
3
, 0)
x
y
x
y
Critical point at (0,0)
(x,y) = (x, y)
(x,y) = (
2
/
3
+x, y)
(V10.2a) f(x, y) =
_
1 x + xy
_
2
= 1 2x + x
2
+ 2xy +
(V10.2b) f(1 + x, 1 + y) =
_
1 (1 + x) + (1 + x)(1 + y)
_
2
= 1 + 2y + 2xy + 2(y)
2
+
(V10.2c) f(x, y) = e
x(y)
2
= 1 + x +
1
2
(x)
2
(y)
2
+
(V10.2d) f(1 + x, 1 + y) = e
(1+x)(1+y)
2
= 1 + x 2y +
1
2
(x)
2
2xy + (y)
2
+
(V10.4) Complete the square and you get
Q(x, y) =
_
x ay
_
2
+
_
1 a
2
_
y
2
.
When 1 a
2
> 0, i.e. when 1 < a < 1 the form is positive definite. When a = 1 the form is a perfect
square, namely,
x
2
2xy +y
2
=
_
x y
_
2
.
When 1 a
2
< 0, i.e. when a > 1 or a < 1, the form is indefinite:
x
2
+ 2axy +y
2
=
_
x ay

a
2
1y
__
x ay +

a
2
1y
_
= (x k
+
y)(x k

y),
where k

= a

a
2
1.
(V10.5) See the solutions to Problem 6.1 for the solutions to this problem.
MATH 234 ANSWERS AND HlNTS 189
(V10.7a) fx = 2x
1
2
y
2
, fy = 2y xy. The equation fy = y(2 x) = 0 leads to two possibilities: x = 2 or y = 0.
If y = 0 then fx = 0 implies x = 0, which gives us one critical point, the origin (0, 0). If on the other hand
x = 2, then fx = 0 implies y
2
= 8 y = 2

2. We therefore get two more critical points (2, 2

2).
The second derivatives are fxx = 2, fxy = y, fyy = 2 x. Therefore we have the following Taylor
expansions at the three critical points:
f(x, y) = f(0, 0) + (x)
2
+ (y)
2
+ loc.min.
f(2 + x, 2

2 + y) = f(2, 2

2) + (x)
2
2

2xy + 0(y)
2
+
= f(2, 2

2) +
_
x 2

2y
_
x + saddle
f(2 + x, 2

2 + y) = f(2, 2

2) + (x)
2
+ 2

2xy + 0(y)
2
+
= f(2, 2

2) +
_
x + 2

2y
_
x + saddle
The origin is therefore a local minimum, and the points (2, 2

2) are saddle
points. At (0, 2

2) the level set consists of two crossing curves, whose tangents are given by x = 0 (a
vertical line) and x = 2

2y (a line with slope 1/2

2 =
1
4

2).
(V10.7c) fx = 1 y
2
, fy = 2 2xy. Critical points: fx = 0 holds when y = 1. If y = +1, then fy = 0
implies x = 1, and if y = 1 then fy = 0 implies x = 1. There are therefore two critical points, (1, 1) and
(1, 1).
(V13.1) f(x, y) = xy, g(x, y) = x
2
+
1
4
y
2
.
#
f = (
y
x
),
#
g =
_
2x
y/2
_
.
First we check for possible max/minima which satisfy
#
g =
#
0. But the only point (x, y) satisfying
#
g(x, y) =
_
0
0
_
is the origin (x, y) = (0, 0), and this point does not lie on the constraint set.
Therefore, if there is a minimum it is aained at a solution of Lagranges equations
fx = gx y = 2x
fy = gy x = y/2
g(x, y) = 1 x
2
+
1
4
y
2
= 1
Multiply the first equation with y and the second with 4x, then you get
y
2
= 2xy and 4x
2
= 2xy
Hence y
2
= 4x
2
. Put that in the constraint, and you find
1 = x
2
+
1
4
y
2
= 2x
2
.
Thus x =

1/2 =
1
2

2 and y =

2. In all we have found four possible solutions. Lagranges method


does not tell us which, if any, of these are minima.
A B
C D
Level sets of the function
f(x, y) = xy and the con-
stiaint set x
2
+
1
4
y
2
= 1
By looking at the constraint set (its an ellipse with horizontal axis of length 1 and vertical axis of length
2) and taking into account that f(x, y) = xy is positive in the first and third quadrants, and negative in the
second and fourth, you find out that the two points (
1
2

2,

2) and (
1
2

2,

2) (A and C in the figure)


are maximum points, while (
1
2

2,

2) and (
1
2

2,

2) (B and D in the figure) are minimum points.


190 MATH 234 ANSWERS AND HlNTS
(V13.2a) Let the sides of the box be x, y, z. We want to minimize the quantity A = 2xy+2yz+2xz, with the constraint
V = xyz =
1
2
. The constraint implies that x = 0, y = 0 and z = 0 moreover, given x and y the only z which
satisfies the constraint is z = 1/(2xy). Thus we must minimize the following function of two variables
A(x, y) = xy +
1
2x
+
1
2y
over all x > 0, y > 0.
A minimum must be an interior minimum (cant be on the x or y-axis since these are excluded), and thus
must be a critical point.
A
x
= y
1
2x
2
,
A
y
= x
1
2y
2
.
Solving Ax = Ay = 0 for (x, y) leads to x = y =
3

2, so the solution is a cube 1/


3

2 on a side
(V13.2b) We wish to minimize A(x, y, z) = 2yz +2xz +2xy with constraint V (x, y, z) = xyz =
1
2
, using Lagranges
method.
First we check for exceptional points on the constraint set, i.e. points (x, y, z) that satisfy both
V (x, y, z) =
1
2
and
#
V (x, y, z) =
#
0. Since
#
V =
_
_
yz
xz
xy
_
_
the gradient
#
V vanishes if at least two of the three coordinates x, y, z are zero. But such a point can never
satisfy the constraint xyz =
1
2
. Therefore, if there is a box with least area, its sides x, y, z must satisfy
Lagranges equations.
Lagranges equations are
Ax = Vx 2y + 2z = yz
Ay = Vy 2x + 2z = xz
Az = Vz 2x + 2y = xy
To get rid of multiply the first equation with x and the second with y to get
y(2x + 2z) = xyz = x(2y + 2z) 2xy + 2yz = 2xy + 2xz 2yz = 2xz.
Therefore we find that either z = 0 or x = y. But z = 0 is not possible, because (x, y, z) must satisfy the
constraint xyz = 0. Therefore we get x = y.
If you multiply the second Lagrange equation with y and the third with z then the same reasoning as
above tells you that y = z.
So, if there is a minimum then it happens when x = y = z, i.e. when the box is a cube. The only cube
that satisfies the constraint has sides x = y = z = 2
1/3
.
As always, Lagranges method does not rule out the possibility that the cube we have found actually
maximizes the surface area, rather than minimizing it. That this is actually not the case is something you
would have to prove by other means. We will not do that in this course.
(V13.3) Answer: the shortest distance is

100/3.
Solution: If (x, y, z) is any point than its distance to the origin is d(x, y, z) =

x
2
+y
2
+z
2
. We
want to minimize d(x, y, z) over all points (x, y, z) which satisfy the constraint g(x, y, z) = x+y +z = 10.
Instead of minimizing d(x, y, z) we will minimize f(x, y, z) = d(x, y, z)
2
= x
2
+y
2
+z
2
. You can do this
problem directly with the function d(x, y, z) and you will get the same answer the computations are just a
lile longer because f has easier derivatives than d.
We use Lagranges method. First we check for exceptional points, i.e. points on the constraint set which
satisfy
#
g =
#
0. Since
#
g =
_
1
1
1
_
the gradient of g can never be the zero vector, so there are no exceptional
points. If there is a minimum of f on the constraint set, it must be a solution of Lagranges equations.
The Lagrange equations are
fx = gx 2x =
fy = gy 2y =
fz = gz 2z =
Therefore if there is a nearest point to the origin on the plane then it must satisfy x = y = z = /2 as well as
the constraint. The only point satisfying these conditions is (
10
3
,
10
3
,
10
3
).
MATH 234 ANSWERS AND HlNTS 191
Lagranges method does not tell us that this is the nearest point. As far as Lagrange is concerned it could
also be the furthest point from the origin. (But because we know what a plane looks like we know that there
has to be a nearest point to the origin.)
(V13.5a) Minimize f(x, y, z) = (x2)
2
+(y1)
2
+(z 4)
2
subject to the constraint g(x, y, z) = 2xy+3z = 1.
First, since
#
g
_
2
1
3
_
=
#
0, there are no exceptional points, so the nearest point (if it exists) is a
solution of Lagranges equations. These are
2(x 2) = 2, 2(y 1) = , 2(z 4) = 3.
Eliminate to get
x = 2y + 4, z = 3y + 7.
Combined with the constraint you then find
y = 2, x = 0, z = 1.
The Lagrange multiplier is = x 2 = 2.
The distance from the point we found to the given point (2, 1, 4) is
d =

(x 2)
2
+ (y 1)
2
+ (z 4)
2
=

14
(V13.5b) |ax
0
+by
0
+cz
0
d|/

a
2
+b
2
+c
2
(V13.8) a cube
(V13.10) 65/3 65/3 130/3
(V13.11) It has a square base, and is one and one half times as tall as wide. If the volume is V the dimensions are
3

2V /3
3

2V /3
3

9V /4.
(V13.12) (0, 0, 1), (0, 0, 1)
(V13.13)
3

4V
3

4V
3

V /16
(V13.14) Farthest: (

2,

2, 2 + 2

2); closest: (2, 0, 0), (0, 2, 0)


(VI3.1a) 2
(VI3.1b) 8
(VI3.1c) 2/3
(VI3.1d)

y
0
sin y
y
dx dy =


0
sin y
y
y dy =


0
sin y dy = 2.
(VI3.1e) Except for a change in notation (y and x r) this is the same integral as in the previous problem. The
answer is again 2.
(VI3.1f) Which function is being integrated? Its the function f(x, y) = 1.

1
0

1x
2
0
dy dx =

1
0
_
y
_
y=

1x
2
y=0
dx =

1
0

1 x
2
dx. The last integral is the area of a quarter
circle with radius 1, so the answer is /4.
(VI3.2) Once you compute the inner integral

1
0
sin(x)dx =
_

cos x
_
1
x=0
=
1

cos
1

/4(cos 0) = 2,
you get

1
x
_

1
0
sin(x)dx
_
dy =

1
x
2dy = [2y]
1
y=x
= 2(1 x).
The result depends on x. The x in the answer and the two x-es in the inner integral refer to dierent quantities.
This is at best confusing, and should really never be done.
192 MATH 234 ANSWERS AND HlNTS
(VI3.3a) Not true! To give a counterexample for the statement in the problem, almost any two functions f and g will
do. For instance, if you choose f(x) = x, g(y) = 1, then you get

1
0

2
0
f(x)g(y) dx dy =

1
0

2
0
xdxdy = 2.
but

1
0
f(x) dx

2
0
g(y) dy =

1
0
x dx

2
0
dy =
1
2
2 = 1.
(VI3.3b) True!

1
0

2
0
f(x)g(y)dydx =

1
0
_
2
0
f(x)g(y)dy
_
dx.
Since f(x) does not depend on y, we have

2
0
f(x)g(y)dy = f(x)

2
0
g(y) dy.
Therefore

1
0
_
2
0
f(x)g(y)dy
_
dx =

1
0
f(x)
_
2
0
g(y)dy
_
dx.
The integral

2
0
g(y)dy is a constant, and does therefore not depend on x, so we can factor it out of the x-
integral:

1
0
f(x)
_
2
0
g(y)dy
_
dx =

1
0
f(x) dx

2
0
g(y) dy,
which is what we had to show.
(VI3.3c) This is false, and there is no simple way of fixing it. To see that this fails evaluate both sides with f(x) = 1
and g(y). On the le you get the area of the disc D, which is , and on the right you get 2 2 = 4.
(VI3.4) The volume under the graph is
1
3
ba
3
+
1
3
ab
3
=
1
3
ab(a
2
+ b
2
). The volume of the surrounding block is
a b (a
2
+ b
2
), so the region beneath the graph occupies one third of the surrounding block, no maer
which a or b you choose.
(VI3.5a) 16
(VI3.5b) 4
(VI3.5c) 15/8
(VI3.5d) 1/2
(VI3.5e) 5/6
(VI3.5f) 12 65/(2e).
(VI3.5g) 1/2
(VI3.5h) (2/9)2
3/2
(2/9)
(VI3.5i) (1 cos(1))/4
(VI3.5j) (2

2 1)/6
(VI3.5k) 2
(VI3.6a) 8
(VI3.6b) 2
(VI3.6c) 5/3
(VI3.6d) 81/2
MATH 234 ANSWERS AND HlNTS 193
(VI3.6e) 2a
3
/3
(VI3.6f) 8
(VI3.6g) /32
(VI3.8a) A
(VI3.8b) B/2
(VI3.9a)

1
0


1x
2
0
2xy
x
2
+y
2
dy dx.
(VI3.9b) In P.C. the function simplifies to F(r, ) = 2 sin cos , so the volume is
V =

1
0

/2
0
2 sin cos r d dr =

1
0
_
sin
2

_
/2
0
r dr =
1
2
.
(VI7.1a) A cone around the positive z axis, with opening angle /6.
(VI7.1b) The negative half of the z axis.
(VI7.1c) The xy plane.
(VI7.1d) The half of the yz plane which contains the positive y axis, and which ends at the z-axis.
(VI7.2a) 0 /2, 0 a, 0 /2.
(VI7.2b) 0 /2, 0 r a, 0 z

a
2
r
2
, or:
0 /2, 0 z a, 0 r

a
2
z
2
.
(VI7.3) Figures 15 and 16.
(VI7.4a) Large circle has radius 1, the smaller has radius

1 z
2
.
(VI7.4b) x =

1 y
2
z
2
for the point in front, and x =

1 y
2
z
2
for the point in the back (furthest away
from you, the viewer).
(VI7.5a) The potential energy is massheightg. The mass of the small piece of honey is m = V , where V
is the volume occupied by the small piece of honey. This is not an exact formula, but only an approximation,
since not all particles in the small piece of honey have exactly the same height. However, as one considers
smaller and smaller pieces the approximation gets beer.
(VI7.5b) The total potential energy is
P.E. =

D
gz dV.
Interpretation: this is the total energy that would be released if you put all the honey at height zero (e.g. by
pouring it out of the jar onto the floor.)
(VI7.5c) The iterated integral is
P.E. =

A
x=0

B
y=0

f(x,z)
z=0
gz dz dy dx =
1
2
g

A
x=0

B
y=0
f(x, y)
2
dy dx.
(VI7.6a) The kinetic energy in a small region of the airmass is
1
2
mv
2
, where mis the mass of the air in the small
region. This mass is V , with V the volume of the small region, so the kinetic energy of the small region
is
1
2
v
2
V . Partitioning the whole airmass, and adding the kinetic energies of all the small pieces leads
to this integral:
K.E. =

D
1
2
v(r)
2
dV =
1
2

D
v(r)
2
dV.
194 MATH 234 ANSWERS AND HlNTS
(VI7.6b) In cylindrical coordinates the domain is defined by 0 r R and 0 < z H, so the integral is
K.E. =
1
2

2
=0

H
z=0

R
r=0
r
1 +r
2
dr dz d =

2
H ln
_
1 +R
2
_
.
(VI7.7a) 623/60
(VI7.7b) 3e
2
/4 + 2e 3/4
(VI7.7c) 1/20
(VI7.7d) /48
(VI7.7e) 11/84
(VI7.7f) 151/60
(VI7.8) 32
(VI7.9) 64/3
(VI7.10) x = y = 0, z = 16/15
(VI7.11) x = y = 0, z = 1/3
(VI7.12a) I = V
+
, J = V

(note the minus sign), K = V


+
V

, L = V
+
+V

.
(VI7.13) /12
(VI7.14) 5/4
(VI7.15) 0
(VI7.16) 5/4
(VI7.17) 4/5
(VI7.18) 256/15
(VI7.19) 4
2
(VI7.20) kh
2
a
2
/12
(VI7.21) kha
3
/6
(VI7.22)
2
/4
(VI7.23) 4/5
(VI7.24) 15
(VII4.1a) The answer is 1. You could compute that, but you dont have to. The distance is 1 everywhere, so its average
should also be 1.
(VII4.1b)

/2
0
d
/2
= /4
(VII4.3) The average x coordinate is zero, and the average y coordinate is 2/.
(VII4.4)
#
x(t) =
_
t
t
2
_
is a parametrization, so the integral becomes

C
x ds =

1
t=0
t
.
x=t

1 + 4t
2
.

#
x

(t)
dt =
_
2
3
1
8
_
1 + 4t
2
_
3/2
_
1
0
=
5

5 1
12
.
MATH 234 ANSWERS AND HlNTS 19
(VII4.5a) a, H, L are lengths; T
0
is a temperature.
(VII4.5b) a = 1 is the radius of the cylinder on which the helix lies, and H = /2 is the height of one turn of the helix.
(VII4.5c) The average is
average temp. =

C
T ds

C
ds
.
With the given parametrization ds =
#
x

(t) dt =

a
2
+H
2
/4
2
dt an ugly expression, but its
constant, which is good for integrating. You get

C
ds =

2
0

a
2
+H
2
/4
2
dt = 2

a
2
+H
2
/4
2
=

4
2
a
2
+H
2
.
and

C
T ds =

2
0
T
0
e
Ht/2L

a
2
+H
2
/4
2
dt
= T
0

a
2
+H
2
/4
2
_

2L
H
e
Ht/2L
_
2
t=0
= T
0

a
2
+H
2
/4
2
2L
H
_
1 e
H/L
_
.
Therefore the average temperature is
average temp. =
L
H
_
1 e
H/L
_
T
0
.
(VII8.1) Yes. It is the gradient of f(x, y) = gy.
(VII8.3) By Clairauts theorem, if
#
F is a gradient, then Py = Qx.
By the fundamental theorem for line integrals, if
#
F is a gradient, then

C
#
F d
#
x = 0 (or, equivalently,

C
Pdx +Qdy = 0) for every closed curve C.
(VII8.4a)

C
#
F d
#
x =

C
xdx = 0.

C
#
G d
#
x =

C
xdy = .
(VII8.4b) Since

C
#
G d
#
x = 0 the vector field
#
Gcannot be a gradient.
(VII8.4c) The integral

C
#
f d
#
x vanishes, but to check that
#
F is conservative one has to check

C
#
f d
#
x = 0 for all
closed curves C, and not just the unit circle. So our integral computation does not imply that
#
F is a gradient.
You can use dierent arguments to show directly that
#
F is a gradient, for instance, by noting that
#
(
1
2
x
2
) = (
x
0
), or if youre not that lucky, by using the methods of IV.14.
(VII12.2b) The answer in both cases is the same (because they are two dierent ways of computing the same integral).
The second approach, using Greens theorem leads to

C
2y dx + 3xdy =

R
_
3x
x

2y
y
_
dA =

R
(3 2) dA,
so the answer is the area of the square, i.e. 1
(VII12.3) Using Greens theorem we get zero. But here we do not need Greens theorem: the Fundamental Theorem for
line integrals (see sec:integral-over-closed-curve-of-gradient-vanishes) tells us that this integral must be zero.
(VII12.5a) 0
(VII12.5b) 1/(2e) 1/(2e
7
) +e/2 e
7
/2
(VII12.5c) 1/2
(VII12.5d) 0
19 MATH 234 ANSWERS AND HlNTS
(VII12.5e) 1/6
(VII12.5f) (2

3 10

5 + 8

6)/3 2

2/5 + 1/5
(VII12.5g) 11/2 ln(2)
(VII12.5h) 2 /2
(VII12.5i) 17/12
(VII12.5j) 0
(VII12.5k) /2
(VII12.5l) /2
(VII12.5m) 12
(VII17.1) The distance to the central axis is r
2
= y
2
+z
2
, so
#
v(x, y, z) = vc
_
1
y
2
+z
2
R
2
_
#

(VII17.2) The inverse square law holds:

#
F =
_
_
_
_
C
#
x

#
x
3
_
_
_
_
=
C

#
x
3

#
x =
C

#
x
2
.
(VII17.3) n = 2 and C =
0
I/2.
(VII17.5a)
_
e
#
m
#
x
_
x
1
= m
1
e
#
m
#
x
, and the same for the x
2
and x
3
derivatives. Therefore
#

_
e
#
m
#
x
_
=
_
_
m
1
e
#
m
#
x
m
2
e
#
m
#
x
m
3
e
#
m
#
x
_
_
= e
#
m
#
x
_
m
1
m
2
m
3
_
.
(VII17.5b) Aer simplifying you get
#

#
v =
#
m
#
ae
#
m
#
x
.
(VII17.5c)
#

#
v =
#
m
#
ae
#
m
#
x
.
(VII17.5d)
#
a and
#
mmust be perpendicular.
(VII17.5e) If
#
v is the gradient of some function, then its curl must vanish. Therefore
#
a
#
m =
#
0 in view of part 3 of
this problem. The conclusion is that
#
a and
#
mmust be parallel.
(VII17.6)
#
v
#
f = Pfx +Qfy +Rfz.
(VII17.7a) By definition,
#
(f
#
v) =
#

_
_
fP
fQ
fR
_
_
=
fP
x
+
fQ
y
+
fR
z
= fxP +fPx +fyQ+fQy +fzR +fRz
= fxP +fyQ+fzR +f
_
Px +Qy +Rz
_
=
_
_
fx
fy
fz
_
_

_
_
P
Q
R
_
_
+f
#

#
v
=
#
f
#
v +f
#

#
v,
as claimed.
MATH 234 ANSWERS AND HlNTS 19
(VII17.7b)
#
(f
#
v) = (
#
f)
#
v + f
#

#
v is the rule. The derivation goes along the same lines as in the previous
product rule.
(VII17.8) This is example 16.1.
(VII17.9a) 5
2
.
(VII17.9b)
#
x
#
x

=
#
x
2
/ =
2
/ = .
(VII17.9c) Note that
#
x = , so you have to compute
#
(
#
x/
3
). The answer is zero.
It says that the divergence of the gravitational field of the Earth is zero.
(VII17.10b) Since
#
x is the gradient of some function its curl must vanish.
(VII17.10c)
#
(
#
x) = (
#
)
#
x +
#

#
x =
#
0
(VII17.11)
#
v(x, y, z) =
_
y
x
0
_
so
#

#
v =
_
0
0
2
_
= 2
#
k.
(VII17.12a)
#
v(x, y, z) =
_
_
x(x
2
+y
2
+z
2
)
n/2
y(x
2
+y
2
+z
2
)
n/2
z(x
2
+y
2
+z
2
)
n/2
_
_
.
(VII17.12b) Using the product rule, you get
#
(
n #
x) = (
#

n
)
#
x +
n
#

#
x = n
n1
(
#
)
#
x +
n
#

#
x.
Now recall (or compute again):
#
=
#
x

, and
#

#
x = 3.
This leads to
#
(
n #
x) = n
n1
#
x

#
x + 3
n
= n
n2

#
x
2
+ 3
n
= (n + 3)
n
)
(VII17.12c) n = 3.
(VII17.13) There are a long and a short answer. The long(er) computation goes likes this:
#
F() =
_
_
F()x
F()y
F()z
_
_
=
_
_
F

()x
F

()y
F

()z
_
_
= F

()
_
_
x
y
z
_
_
.
Now recall (185), and you find
#
F() = F

()
_
_
x/
y/
z/
_
_
=
1

()
#
x.
The short computation is essentially the same, but you never write the components of the vectors:
#
F() = F

()
#
=
1

()
#
x.
(VII17.13a) If f(x, y, z) = F(), then by the previous problem we have
#
f =
1
F

()
#
x. We want this to be equal
to
n #
x, so F() must satisfy

1
F

(rho) =
n
F

() =
1+n
F() =

2+n
2 +n
+C
for some constant C. We are only asked to find on function f, so we find that the given vector field is indeed
the gradient of a radially symmetric function:
#
v =
n #
x =
#

_
2+n
2 +n
_
.
The exceptional case is when n = 2, in which case you get F() = ln .