You are on page 1of 7

CALC III

1 Vectors : For two vectors


a = a
1

i + a
2

j + a
3

k ;

b = b
1

i + b
2

j + b
3

k
we dene the dot and cross product as follows,
a

b := a
1
b
1
+ a
2
b
2
+ a
3
b
3
a

b := (a
2
b
3
a
3
b
2
)

i + (a
3
b
1
a
1
b
3
)

j + (a
1
b
2
a
2
b
1
)

k .
Notice that the dot product is a number, while the cross product is a vector. The basic geometric property
of these two products is contained in the formulas below,
a

b = a

b cos()
a

b = a

b sin()
a

b a ; a

b ,
where is the angle between the two vectors. You can use these two formulas in order to compute the sine
and cosine of the angle between the two vectors. The projection of vector

b onto vector a is,
Pr
a
_

b
_
:=
_

b a
_
a
a
2
which is a vector parallel to a.
For given orthogonal (Cartesian) axes labeled x, y and z we use the coordinates of the point P(x, y, z)
to create a vector r := x

i + y

j + z

k which we will call the position vector of the point P.


2 Lines and Planes : a)For some xed vectors
r
0
:= x
0

i + y
0

j + z
0

k ; v := a

i + b

j + c

k
indicating initial position and direction respectively, we can write the equation of a straight line r(t) = r
0
+tv,
where t is a free parameter (variable). If we eliminate t from the equations we obtain the nonparametric
form,
x x
0
a
=
y y
0
b
=
z z
0
c
.
Notice that (a, b, c) is the direction vector of the line. If r
1
= x
1

i + y
1

j + z
1

k is the position vector of some


point, the distance of this point from the line is given by
d :=
(r
1
r
0
) v
v
.
b) The equation of a plane with normal (orthogonal) vector n = a

i +b

j +c

k and passing through some xed


point r
0
= x
0

i + y
0

j + z
0

k is n
_
r r
0
_
= 0, or in coordinates,
a(x x
0
) + b(y y
0
) + c(z z
0
) = 0 or ax + by + cz = d .
Notice that (a, b, c) is the normal vector to the plane. If r
1
is the position vector of some point in space,
then the distance of this point from the plane is
d :=
|(r
1
r
0
) n|
n
.
1
3 Curves : A general curve in space is described by an equation of the form,
r(t) = x(t)

i + y(t)

j + z(t)

k ; a t b
where x(t), y(t) and z(t) are three given functions of the parameter (variable) t. We dene the velocity and
acceleration as follows,
v(t) :=
dr
dt
=
dx
dt

i +
dy
dt

j +
dz
dt

k
a(t) :=
d
2
r
dt
2
=
d
2
x
dt
2

i +
d
2
y
dt
2

j +
d
2
z
dt
2

k .
Smooth parametrization means v(t) > 0. The length and curvature of the curve are computed via,
L :=
_
b
a
v(t)dt and (t) :=
v a
v
3
.
We write ds := v(t)dt and for a given function f(r) we dene the geometric integral,
_
C
f(r)ds :=
_
b
a
f(r(t)v(t)dt =
_
b
a
f(r(t))
_
_
_
_
dr
dt
_
_
_
_
dt ,
where the function f(r) is evaluated on the curve.
Functions and Derivatives : For a given function f(x, y) or f(x, y, z) of two or three variables, we can
think of r = x

i+y

j +z

k (the position vector) as input and the numerical value f(r) as output, thus assigning
a numerical value to the point r in space. The gradient of the function is the vector

f = f
x

i + f
y

j + f
z

k .
The directional derivative in the direction v and at the point r is,
D
v
f(r) =

f(r) v
provided that the partial derivatives are continuous. If the second derivatives are continuous we have the
equality of mixed partial derivatives,
f
xy
= f
yx
; f
yz
= f
zy
; f
zx
= f
xz
.
Chain Rules : a) For a given function f and curve r(t) we have,
d
dt
f(r(t)) =

f(r(t))
dr
dt
.
b) If we make a change of variables i.e. express (x, y) as functions of new variables (u, v) via some formulae
x := x(u, v) and y := y(u, v) then a given function f(x, y) turns into a new function (via composition)
g(u, v) := f(x(u, v), y(u, v)) .
The partial derivatives of the new function are computed via the formulae,
g
u
=
f
x
x
u
+
f
y
y
u
g
v
=
f
x
x
v
+
f
y
y
v
2
which can be writen (using Linear Algebra)
(g
u
, g
v
) = (f
x
, f
y
)
_
x
u
x
v
y
u
y
v
_
.
If a function z(x, y) is dened implicitly via the equation f(x, y, z(x, y)) = c for some constant c, then we
can compute (in principle) the partial derivatives of z(x, y) using the formulae
z
x
(x, y) =
f
x
(x, y, z)
f
z
(x, y, z)
; z
y
(x, y) :=
f
y
(x, y, z)
f
z
(x, y, z)
,
provided that f
z
(x, y, z) = 0 and (x, y, z) satisfy the equation f(x, y, z) = c. If {(x, y, z)

f(x, y, z) = c} is a
level surface for the function f, then the gradient

f(r
0
) is orthogonal (normal) to the surface at the point
r
0
on the surface (provided that the gradient does not vanish). This means that the unit normal to the level
surface at the point r
0
is
n =

f(r
0
)

f(r
0
)
.
Critical points : Critical points means solving the equation

f =

0. For a function of two variables you
want to solve
f
x
(x, y) = 0 and f
y
(x, y) = 0
for x and y. To characterize the critical point, say (x
0
, y
0
) you compute the Hessian
_
f
xx
(x
0
, y
0
) f
xy
(x
0
, y
0
)
f
yx
(x
0
, y
0
) f
yy
(x
0
, y
0
)
_
and
det
_
f
xx
(x
0
, y
0
) f
xy
(x
0
, y
0
)
f
yx
(x
0
, y
0
) f
yy
(x
0
, y
0
)
_
> 0 , f
xx
(x
0
, y
0
) > 0 ; local min
det
_
f
xx
(x
0
, y
0
) f
xy
(x
0
, y
0
)
f
yx
(x
0
, y
0
) f
yy
(x
0
, y
0
)
_
> 0 , f
xx
(x
0
, y
0
) < 0 ; local max
det
_
f
xx
(x
0
, y
0
) f
xy
(x
0
, y
0
)
f
yx
(x
0
, y
0
) f
yy
(x
0
, y
0
)
_
< 0 ; saddle point
If the determinant is zero then we cannot decide what type of critical point we have without computing
higher order derivatives.
If f(x, y) is a continuous function dened on some closed bounded region of the plane then this function
attains both its minimum and its maximum. This means that if f is continuously dierentiable you serach
for critical points inside the region and you do a separate search on the boundary in order to determine the
overall maximum and the minimum.
Lagrange multipliers : If we have to nd extrema (maxima or minima) of some function f subject to
some restriction g = c where c is a constant, then the extrema, say at r
0
satisfy the equation

f(r
0
) =

g(r
0
) and g(r
0
) = c
provided that

g(r
0
) =

0. is called the Lagrange multiplier and it is an extra unknown together with
(x
0
, y
0
, z
0
). For function f(x, y) and g(x, y) you are supposed to solve the system
f
x
(x, y) = g
x
(x, y) ; f
y
(x, y) = g
y
(x, y) ; g(x, y) = c
in order to determine the extrema.
3
Double and Triple integrals : We indicate double and triple integrals as
_ _
D
f(x, y)dA ;
_ _ _
D
f(x, y, z)dV
where D is the domain of integration. Sometimes it is convenient to write dA = dxdy and dV = dxdydz in
order to remember which variables we are integrating.
Fubini : For double integrals, if the domain of integration is described by,
D := {(x, y)

g
1
(x) y g
2
(x) , a x b}
we can turn the double integral into repeated (nested) integral via Fubinis formula,
_ _
D
f(x, y)dxdy =
_
b
a
_
_
g2(x)
g1(x)
f(x, y)dy
_
dx
where the point in the above formula is that in the innermost integration x is frozen. In a similar spirit, if
for a triple integral the domain is described by
D := {(x, y, z)

g
1
(x, y) z g
2
(x, y) , (x, y) R}
then we can write
_ _ _
D
fdV =
_ _
R
_
_
g2(x,y)
g1(x,y)
f(x, y, z)dz
_
dxdy
where again the point is that in the innermost integration (x, y) are frozen and R is the projection of D onto
the x y plane.
Change of variables : If we make a change of variables from (x, y) to new variables (u, v) or from (x, y, z)
to (u, v, w) then the integrals change as follows,
_ _
D
f(x, y)dxdy =
_ _
K
f(x(u, v), y(u, v))

(x, y)
(u, v)

dudv ,
_ _
D
f(x, y, z)dxdydz =
_ _ _
K
f(x(u, v, w), y(u, v, w), z(u, v, w))

(x, y, z)
(u, v, w)
,

dudvdw
where
(x, y)
(u, v)
= det
_
x
u
x
v
y
u
y
v
_
;
(x, y, z)
(u, v, w)
= det
_
_
x
u
x
v
x
w
y
u
y
v
y
w
z
u
z
v
z
w
_
_
are called the Jacobians of the transformation and K is the new domain in the new variables of integration
(u, v) or (u, v, w).
Polar coordinates : Polar coordinates are dened via the formulae x = r cos() y = r sin() where
0 r < + and 0 < 2. We can compute the Jacobian easily,
(x, y)
(r, )
= r
and the integral trasforms according to,
_ _
D
f(x, y)dxdy =
_ _
K
f(r cos(), r sin())rdrd
where K is the new integration domain in the (r, ) variables.
4
Spherical coordinates : Spherical coordinates are described via the formulae,
x = sin() cos()
y = sin() sin()
z = cos()
where, 0 < +, 0 , 0 < 2. We can compute the Jacobian again,
(x, y, z)
(, , )
=
2
sin()
and the triple integral transforms according to,
_ _ _
D
f(x, y, z)dxdydz =
_ _ _
K
f( sin() cos(), sin() sin(), cos())
_

2
sin()
_
ddd .
Notice that
_
x
2
+ y
2
z
= tan()
thus for a cone described by z = a
_
x
2
+ y
2
the slope is determined by tan() = 1/a.
Surface area : For a given surface described by r(u, v) with (u, v) K the surface area is dened via the
integral,
A(S) :=
_ _
K
r
u
r
v
dudv ,
which is independed of the parametrization of the surface as long as the parametrization is smooth i.e.
r
u
r
v
> 0. In the important special case that the surface is the graph of some function f(x, y) i.e.
S := {(x, y, z)

z = f(x, y)}
we can write r(x, y) = x

i +y

j +f(x, y)

k for the position vector on the surface and subsequently the surface
area is given by the integral,
A(S) :=
_ _
R
_
1 + f
2
x
+ f
2
y
dxdy
where R is the domain of the function f(x, y) i.e. the projection of the surface onto the x y plane. We
write
dS := r
u
r
v
dudv
or for a graph
dS :=
_
1 + f
2
x
+ f
2
y
dxdy
and for some given function g(x, y, z) the surface integral is
_ _
S
gdS =
_ _
K
g(r(u, v))r
u
r
v
dudv
or for a graph,
_ _
S
gdS =
_ _
R
g(x, y, f(x, y))
_
1 + f
2
x
+ f
2
y
dxdy .
Stokes-Green-Gauss : A vector eld assigns a vector to every point in space. We can write

F(r) to
indicate this, which means that for every r = x

i + y

j + z

k in some domain D we have three functions, say


M, N and P of (x, y, z) as folows,

F := M

i + N

j + P

k .
5
We call

F(r) a vector eld and f(r) a scalar eld. There are two types of derivative of a vector eld that
are the most important in applications, the divergence (which is a scalar eld) dened by,


F :=
_

i +
y

j +
z

k
_

_
M

i + N

j + P

k
_
= M
x
+ N
y
+ P
z
,
and the curl (which is a vector eld) dened by


F =
_

i +
y

j +
z

k
_

_
M

i + N

j + P

k
_
= (P
y
N
z
)

i + (M
z
P
x
)

j + (N
x
M
y
)

k .
We have two identities,

f
_
=

0 and


F
_
= 0 .
A vector eld

F can be integrated over a curve or a surface. We call the rst circulation and the second
ux. Given a curve r(t) with a t b we dene the circulation to be the integral,
_
C

F dr :=
_
b
a

F(r(t))
dr
dt
dt =
_
b
a
_
M
dx
dt
+ N
dy
dt
+ P
dz
dt
_
dt
which is an ordinary integral in the variable t and the (x, y, z) are replaced by (x(t), y(t), z(t)). The important
point here is that the integral does not depend on the parametrization i.e. any smooth parametrization gives
the same numerical value for the integral. The ux through a surface S described by r(u, v) with (u, v) K
is dened as follows
_ _
S
_

F n
_
dS =
_ _
K
_

F (r
u
r
v
)
_
dudv
because
n =
r
u
r
v
r
u
r
v

and dS = r
u
r
v
dudv .
Again the point in the above formula is that the integral is independent of the parametrization of the surface
as long as the parametrization is smooth. In the important special case that r(x, y) = x

i + y

j + f(x, y)

k
with (x, y) R some region in the x y plane, the ux turns into the following double integral,
_ _
S
_

F n
_
dS =
_ _
R
_
Mf
x
Nf
y
+ P
_
dxdy =
_ _
R
_
Mz
x
Nz
y
+ P
_
dxdy
if we write z = f(x, y) and think of z as a function of (x, y). In the formula above the unit normal vector n
is chosen to point in the positive z direction.
The three fundamental formulae are : 1) If

F =

f i.e. the vector eld is the gradient of some function
(scalar eld) f then,
_
C

f(r) dr = f(r(b)) f(r(a)) .


If C is a closed curve i.e. it has empty boundary, the integral is zero Notice that

f
_
=

0 always and
if for some vector eld

F it happens that



F =

0 then we can nd some scalar function f such that

F =

f.
2) Stokes : If we take the curl of a vector eld

F and compute its ux through some surface S this is equal
to the circulation of the vector eld around the boundary.
_ _
S
_
(


F) n
_
dS =
_
C

F dr .
Notice that if


F =

0 then the ux of the curl is zero and this means that the circulation is also zero.
This is to be expected since the vanishing of the curl implies that

F =

f for some scalar eld and the
6
circulation around a closed curve is zero. In the special case that

F = M(x, y)

i + N(x, y)

j + 0

P, Stokes
formula reduces to Greens formula,
_ _
R
_
N
x
M
y
_
dxdy =
_
C
_
Mdx + Ndy
_
.
In N
x
M
y
= 0 then we can nd a function f(x, y) such that f
x
= M and f
y
= N. The integrals above are
oriented in the following way. Imagine that you walk around the boundary of S with your head pointing in
the direction of n. If S is on your left hand as you walk around the boundary then you have the positive
orientation.
3) Gauss : Finally, for a vector eld

F the volume integral of the divergence over some region D in space
is equal to the ux of

F through the boundary i.e.
_ _ _
D
_


F
_
dV =
_ _
S
_

F n
_
dS
where the choice of n is such that it points outwards from D. These formulas are valid provided that

F
has continuous partial derivatives in the region of integration. All three formulae are manifestations of the
fundamental theorem of calculus. Notice that if


F = 0 in some region in space D then the ux through
the boundary of D is zero. For example we know that


F
_
= 0 which means that for a closed surface
S we have _ _
S
_
(


F) n
_
dS = 0 .
This is to be expected since S has no boundary (being closed) thus Stokes formula says that the ux is zero.
7

You might also like