Вы находитесь на странице: 1из 7

Calculus III (part 1): Vector-Valued Functions

(by Evan Dummit, 2013, v. 1.40)

Contents
1

Vector-Valued Functions

1.1

Vectors and Scalars . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

1.2

Dot Product and Cross Product . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

1.2.1

Norm and Dot Product

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

1.2.2

Cross Product

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

1.2.3

Equations of Lines and Planes

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

1.3

Vector-Valued Functions, Derivatives, Tangent Vectors . . . . . . . . . . . . . . . . . . . . . . . . . .

1.4

Curves in 3-Space: Arclength; Tangent, Normal, and Binormal Vectors; Curvature and Torsion

. . .

Vector-Valued Functions

1.1

Vectors and Scalars


A vector is a quantity which has both a magnitude and a direction.

This is in contrast to a scalar, which carries only a magnitude.

We denote the
where the

ai

n-dimensional

vector from the origin to the point

(a1 , a2 , , an )

as

v = ha1 , a2 , , an i,

are scalars.

We use the angle brackets

hi

rather than parentheses

a vector and the coordinates of a point in space.

()

so as to underscore the dierence between

We will, however, view coordinates of vectors and

coordinates of points as essentially interchangeable.

We also write vectors in boldface (v, not

v ),

so that we can tell them apart from scalars. When writing

by hand, it is hard to dierentiate boldface, so the notation

~v

is also sometimes used.

The typical way to think of vectors is as directed line segments: the length of the line segment gives the
magnitude of the vector, and the direction the segment is pointing gives the direction of the vector.

Note/Warning: Vectors are a little bit dierent from directed line segments, because we don't care where
a vector starts: we only care about the dierence between the starting and ending positions. Thus: the
directed segment whose start is

(2, 2) represent the

same vector,

(0, 0) and end is (1, 1) and the segment starting at (1, 1) and ending at
h1, 1i. This distinction is rarely necessary in most applications, however.

We can add vectors (provided they are of the same dimension!) in the obvious way, one component at a time:
if

v = ha1 , , an i

and

w = hb1 , , bn i

then

v + w = ha1 + b1 , , an + bn i.

Similarly we can 'scale' a vector by a scalar, one component at a time: if

is a scalar, then we have

r v = hra1 , , ran i.

Scaling a vector by a factor of

1/2,

for example, produces a new vector in the same direction, but with

half the length as the original.

Scaling a vector by

1 produces a new vector with the same length but pointing in the opposite direction

from the original vector.

Example: If

v = h1, 2, 2i and w = h3, 0, 4i then 2w = h6, 0, 8i

2w = h7, 2, 10i

, and

v+w = h2, 2, 2i

. Furthermore,v

1.2

Dot Product and Cross Product

1.2.1

Norm and Dot Product


Denition: We dene the norm (length, magnitude) of the vector

If

32

02

and

(0, . . . , 0)

(a1 )2 + + (an )2

ha1 , . . . , an i is just the length


(a1 , . . . , an ).
p
||v|| = (1)2 + 22 + 22 = 3 , and ||w|| =

to the point

w = h3, 0, 4i,

we have

is a scalar, we can see immediately from the denition that

out a

v = h1, 2, 2i
+ (4)2 = 5 .

Example: For

This is just an application of the distance formula: the norm of the vector
of the line segment joining the origin

v = ha1 , . . . , an i as ||v|| =

r2 = |r|

||r v|| = |r| ||v||,

since we can just factor

from each term under the square root.

From any nonzero vector we can nd a unit vector (that is, a vector of norm 1) in the same direction of
by scaling
as

v.

by 1 over its norm. In other words, the vector

This vector

~u

is sometimes called the normalization of

v
~u =
||v||
v.

just

is a unit vector in the same direction


1 2 2
is a unit vector in the same direction
Example: For v = h1, 2, 2i, we see that ~u1 = , ,
3 3 3


3
4
and for w = h3, 0, 4i we see that ~
u2 =
, 0,
is a unit vector in the direction of w.
5
5

v,

If we have two vectors, we now know how to nd their lengths. But another thing we might want to know
about two vectors is the angle

as

between them. This motivates the denition of the dot product:

Denition: The dot product of two vectors

v1 v2 = a1 b1 + a2 b2 + + an bn

v1 = ha1 , . . . , an i

and

v2 = hb1 , . . . , bn i

is dened to be the scalar

Example: The dot product

h1, 2i h3, 4i

Example: The dot product

h1, 2, 2i h3, 0, 4i

Remark:

is

(1)(2) + (3)(4) = 14
is

(1)(3) + (2)(0) + (2)(4) = 11

The dot product obeys several very nice properties reminiscent of standard multiplication.

For any vectors

v, v1 , v2 , w,

r,

and any scalar

we can verify the following properties directly from the

denition:

Dot product is commutative:

Dot product distributes over addition:

Dot product is sort of  associative with scalar multiplication:

Dot product of a vector with itself is the square of the norm:

Theorem: For vectors

~v1

and

~v2

vw =wv

(v1 + v2 ) w = (v1 w) + (v2 w)

forming an angle

(r v) w = r (v w)
2

v v = ||v||

between them, we have

v1 v2 = ||v1 || ||v2 || cos()

Proof: To prove this statement, we use the Law of Cosines in the triangle formed by
which states that

||v2 v1 || = ||v1 || + ||v2 || 2 ||v1 || ||v2 || cos().

v1 , v2 ,

and

v2 v1 ,

Now we know that the square of the

norm is the dot product of a vector with itself so we can apply this and the other dot product properties
to see that

||v2 v1 ||

(v2 v1 ) (v2 v1 )

(v2 v2 ) (v1 v2 ) (v2 v1 ) + (v1 v1 )


2

= ||v2 || 2(v1 v2 ) + ||v1 || .


Now by comparing to the Law of Cosines expression, we see that everything cancels and leaves us precisely
with the result we wanted.

Denition: We say two vectors are orthogonal if their dot product is zero.



= 0, we see that two nonzero vectors are orthogonal if


2

the angle between them is


, which is to say, if they are perpendicular.
2
Example: The vectors h2, 1, 4i and h3, 2, 1i are orthogonal, since their dot product is (2)(3)+(1)(2)+
(4)(1) = 0.
Example: The vectors h2, 2, 1i and h3, 0, 4i have a dot product of 10, as we computed earlier, and
norms of 3 and 5 respectively. Therefore we see that the angle between them satises 10 = 3 5 cos(),
 
2
1
.
hence = cos
3

1.2.2

From the Dot Product Theorem, since

cos

Cross Product

v1 = hx1 , y1 , z1 i and v2 = hx2 , y2 , z2 i is dened to be the


v1 v2 = hy1 z2 y2 z1 , z1 x2 z2 x1 , x1 y2 x2 y2 i . It is orthogonal to both v1 and v2 .

Denition: The cross product of

vector

Important Note: The cross product is only dened for vectors with 3 components, and outputs another
vector with 3 components. Contrast with the dot product, which is dened for vectors of any length,
and outputs a scalar.

A way to remember the cross product formula (aside from memorization) is the determinant formula


i

v1 v2 = det x1
x2



y1
=
y2

k
z1
z2



x
z1
i 1
x2
z2

i = h1, 0, 0i, j = h0, 1, 0i,

dard unit vectors:

j
y1
y2

and



x
z1
j + 1
x2
z2


y1
k
y2

i, j, k

, where

are the stan-

k = h0, 0, 1i.

It's a little unusual to have vectors inside a determinant, but it works out to the correct answer.
Don't forget the minus sign on the middle term.

v1 v2 is orthogonal to v1 and to v2 . To verify this, we can just evaluate the


v1 (v1 v2 ) and v2 (v1 v2 ) and check that they are both zero. For example we have
v1 w = x1 (y1 z2 y2 z1 ) + y1 (z1 x2 z2 x1 ) + z1 (x1 y2 x2 y2 ), which is zero because each term appears
once with a +and once with a .
We claim that this vector
dot products

Unlike the dot product, the cross product is NOT commutative! Indeed, we can see from the denition
that

v1 v2 = (v2 v1 )

(v1 + v2 ) w = (v1 w) + (v2 w)

for any vector

v.

We also have the same scalar multiplication sort of associativity:

Theorem: If

is the angle between

of the parallelogram formed by

vv =0

We still do have a distributivity property, like with the dot product: it is fairly easy to check from the
denition that

. In particular, we see that

v1

v1

Proof: We just need to show that

||v1 || ||v2 || cos()

and

and

v2 ,

then

(r v) w = r (v w)

||v1 v2 || = ||v1 || ||v2 || sin() = A

, where

is the area

v2 .
2

||v1 v2 || + (v1 v2 ) = ||v1 || ||v2 ||

, because we know that

v1 v2 =

from the Dot Product Theorem.

(y1 z2  y2 z1 ) + (z1 x2 z2 x1 )  +



(x1 y2 x2 y2 ) + (x1 x2 + y1 y2 + z1 z2 ) is equal to (x1 )2 + (y1 )2 + (z1 )2 (x2 )2 + (y2 )2 + (z2 )2 .
When we expand the rst thing, we get each of the 9 possible square terms (1 42 )2 where  and
4 are each one of x, y , or z , and the cross terms like 2x1 x2 y1 y2 will all cancel out.
We get exactly the same sum of 9 square terms when we expand the second thing. So they are equal
To check this we multiply everything out. So we need to see that

and we're done.

For the statement about the area, we can just use geometry to see that the area of the triangle with
sides

~v1

and

~v2

is

1
||v1 || ||v2 || sin().
2

The parallelogram's area is twice this.

Remark: This quite nice property is one reason we chose the denition we did for the cross product.

1.2.3

Equations of Lines and Planes

and

P1 = hx1 , y1 , z1 i and P2 = hx2 , y2 , z2 i, the points hx, y, zi on the line l


P1 and P2 are given parametrically by hx, y, zi = P1 + t v , as t varies through the real numbers,
v = P2 P1 = hx2 x1 , y2 y1 , z2 z1 i.

Proof: There is a unique line between two points, by the axioms of geometry. So we just need to check

Proposition: Given distinct points


through

that this is a line, and that it goes through

P1

and

P2 .

l explicitly is tells us that x = x1 + t(x2 x1 ), y = y1 + t(y2 y1 ), and


z = z1 + t(z2 z1 ), and these are all linear equations. So it's a line.
We see l goes through P1 because at t = 0 we get P1 . Similarly, at t = 1 we get P2 . So we're done.

The parametric equation for

Note: This procedure works to nd the parametrization of a line in any space, not just 3-space.
Remark: We call the vector

v = P2 P1 the direction vector for the line l: it tells us in which direction


P1 in the sum P1 + t v species which, of all possible lines in that direction,

the line is moving. The term


is the line we want.

(1, 2, 3) and (1, 2, 1) we just need to nd the direction


v = h(1) 1, 2 2, (1) 3i = h2, 0, 4i. Then the line is given parametrically by
hx, y, zi = h1 2t, 2, 3 4ti .

Example: To nd the line through the points


vector, which is

Proposition: The plane dened by

ax+by +cz = d

words, every line lying in this plane is orthogonal to

n = ha, b, ci. In other


n = ha, b, ci,
(x0 , y0 , z0 ), and its equation is

is orthogonal to its normal vector

ha, b, ci.

Conversely, given a nonzero vector

there is a unique plane normal to that vector passing through a given point

a(x x0 ) + b(y y0 ) + c(z z0 ) = 0.

Proof: Suppose

is a line in the plane. All we need to show is that its direction vector is orthogonal to

n.

v = P2 P1 , where both of the points P2 = hx2 , y2 , z2 i and P1 =


hx1 , y1 , z1 i lie in the plane.
Then P1 n = a x1 + b y1 + c z1 = d since P1 lies in the plane, and similarly P2 n = d.
But then we have v n = P2 n P1 n = d d = 0, which is exactly what we wanted.
For the converse statement, clearly if n = ha, b, ci then the equation of the plane must be a x + b y +
c z =  for some value of , by the previous proposition.
But if we are given a point that lies in the plane, we can plug in to see that  = ax0 + by0 + cz0 ,

So say the direction vector is

and so we have uniquely determined the equation of the plane, and hence the plane. We can rewrite
the equation as

a(x x0 ) + b(y y0 ) + c(z z0 ) = 0,

so this is the desired plane.

Remark: This proposition says that a plane is specied by its normal vector along with a point that the
plane passes through.

Example:

The plane with normal vector

2(x 2) + 3(y 2) (z 7) = 0

h2, 3, 1i

We need the normal vector

n.

take their cross product to get

We have

Then

has equation

P1 = (3, 0, 1), P2 = (1, 2, 2)

and

P3 = (2, 1, 4).

To nd it, we just need to nd two vectors lying in the plane, and then

n.

We get the two vectors as

v 1 = P2 P1

and

v 2 = P3 P1 .

v1 = P2 P1 = h2, 2, 3i and v2 = P3 P1 = h5, 1, 5i.








2 3




i 2 3 j + 2 2 k = h7, 5, 8i.
n = v1 v2 =




1 5
5 5
5 1

Sanity check:

n v1 = (7)(5) + (5)(1) + (8)(5) = 0

Thus the equation of the plane is

(2, 2, 7)

Example: Find an equation for the plane passing through

passing through the point

and

n v2 = (7)(2) + (5)(2) + (8)(3) = 0.

7(x 3) 5(y 0) + 8(z + 1) = 0,

or

7x 5y + 8z = 13

For an extra error check, we verify that all three points do lie in this plane: we have

8(1) = 13, 7(1) 5(2) + 8(2) = 13,


equation.

and

7(2) 5(1) + 8(4) = 13.

7(3) 5(0) +

Hence this is the correct

1.3

Vector-Valued Functions, Derivatives, Tangent Vectors


Denition: A vector-valued function
a function of the parameter

Example:

r(t) = t2 , 2t

r(t)

is a function whose output is a vector, each of whose components is

t.

.

We can add and scalar-multiply vector-valued functions in the same manner as normal vectors.

and


2
2
t

r1 (t) = e2 , cos(t), t 1 and r2 (t) = t, 0, t we have r1 (t) + r2 (t) = he + t, cos(t), 1i


2r2 (t) = 2t, 0, 2t .

Example: For

r(t) = hx(t), y(t)i and r(t) = hx(t), y(t), z(t)i,


t and output a vector with 2 or 3 coordinates. These functions trace

We will primarily be interested in vector functions of the form


which have a single input parameter

out parametric curves in 2 or 3-dimensional space (respectively).

Later, we will also be interested in vector functions of the form


input parameters

and

t.

r(s, t) = hx(s, t), y(s, t), z(s, t)i,

for two

These functions, in general, will describe surfaces in 3-dimensional space.

The next question to ask is: can we take derivatives of vector-valued functions? The answer is yes:

Denition: The derivative of the vector function

r(t)

is given by

r(t + h) r(t)
h0
h

r0 (t) = lim

, provided the

limit exists.

Note the extreme similarity of this denition with the denition of the derivative of a (scalar) function
of one variable

f (t),

which reads

r(t) = hx(t), y(t), z(t)i,


hx (t), y 0 (t), z 0 (t)i.
For

f 0 (t) = lim

h0

f (t + h) f (t)
.
h

then by applying the denition of the derivative, we see that

r0 (t) =

More generally, we can see that taking the derivative of a vector function is the same thing as
dierentiating each component.

Example: For


r(t) = et , cos(t), t2 1 ,

we have


r0 (t) = et , sin(t), 2t

By the limit part of the denition, we see that the derivative of a vector function tells us in which direction

t.

the curve is moving at time

More specically,

r0 (t)

is a vector tangent to the graph of

Compare to the situation of the derivative of a (scalar) function of one variable

f (t):

r(t).
the derivative

f 0 (t)

gives the slope of the tangent line.

We can also take higher derivatives: for example,

and so forth.


r(t) = et , cos(t), t2 1 , we have r0 (t) = het , sin(t), 2ti, so r00 (t) = et , cos(t), 2


r000 (t) = et , sin(t), 0 , and so on.

Example: For

and

The derivatives of vector-valued functions have the same physical interpretations as we are used to:

r0 (t + h) r0 (t)
,
h0
h

r00 (t) = lim

The rst derivative

r0 (t)

The second derivative

The magnitude

00

gives the velocity of the particle at time

r (t)

||r0 (t)||

t.

gives the acceleration of the particle at time

t.

of the velocity vector gives the speed of the particle.

Derivatives of vector functions satisfy rules strongly reminiscent of the product rule with regard to the dot
and cross products:

d
[r1 r2 ] = r1 (r02 ) + (r01 ) r2
dt

and

d
[r1 r2 ] = r1 (r02 ) + (r01 ) r2
dt

These properties can be veried by expanding out the dot and cross products of

[r2 (t + h) r2 (t)],

applying the limit denition of derivative, and simplifying.

[r1 (t + h) r1 (t)]

with

1.4

Curves in 3-Space: Arclength; Tangent, Normal, and Binormal Vectors; Curvature and Torsion

A particle traveling on a parametric curve

r(t) = hx(t), y(t), z(t)i

in 3-space possesses various related vectors

and quantities:

v(t) = r0 (t)

The velocity

The arclength of the curve between

and acceleration

t = t1

a(t) = v0 (t) = r00 (t).


t = t2 ,

and

given by

t2 q

s=

t1

t2

[x0 (t)] + [y 0 (t)] + [z 0 (t)] dt =

||v(t)|| dt.
t1

v(t)
||v(t)||

measures the direction in which the particle is moving.

By the denition, the unit tangent vector satises

r0 (t).

T(t) =

The unit tangent vector

||T(t)|| = 1,

and

T(t)

is in the same direction as

Thus, it is a unit vector, in the same direction as the tangent vector. (Whence the name.)

The unit normal vector

N(t) =

T0 (t)
||T0 (t)||

, orthogonal to the unit tangent vector, measures roughly the

direction that forces are pulling the particle.

||N(t)|| = 1, as long as T0 (t) 6= 0.


T(t) T(t) = 1, taking the derivative with

By the denition, the unit normal vector satises

N(t) is orthogonal to T(t), since


respect to t
T(t) T0 (t) + T0 (t) T(t) = 0. But since the dot product is symmetric, this says T (the unit
0
tangent vector) and T (the normal vector) have dot product 0. In other words, the (unit) normal

To see that
gives

vector is orthogonal to the tangent vector.

B(t) = T(t) N(t) =

The unit binormal vector

v(t) a(t)
||v(t) a(t)||

, orthogonal to the tangent and normal

vectors, gives the direction perpendicular to the particle's plane of motion.

By its denition via the cross product, the binormal vector is orthogonal to both the (unit) normal
and (unit) tangent vectors, and since those vectors have length 1, so does

B.

Thus, we see that the unit tangent, unit normal, and unit binormal vectors are all orthogonal to each
other  provided that the normal and binormal vectors are dened (which is not always guaranteed
because of the division by

||T0 (t)||

in the denition of

N).

There are several dierent planes associated to these vectors:

The osculating plane is the plane containing

T(t)

and

N(t),

passing through the point

r(t).

It is

sometimes called the tangent plane to the curve.

The rectifying plane is the plane

The curvature

N(t) and B(t), passing through the point r(t).


containing T(t) and B(t), passing through the point r(t).

The normal plane is the plane containing

||T0 (t)||
||v(t) a(t)||
=
3
||v(t)||
||v(t)||

A line has curvature 0, since for a line

measures how much the path of the particle is curving.

T(t)

is always constant hence

||T0 (t)|| = 0.

Conversely, from

essentially the same observation, the only curves whose curvature is zero everywhere are lines, since
a curve with curvature zero must have

For a circle of radius

For a general curve, the quantity

T(t)

be constant.

(e.g., parametrized by

1
(t)

hx, yi = hr cos t, r sin ti),

the radius of curvature is

1
.
r

is called the radius of curvature of the curve: it gives the

radius of the tangent circle to the curve.

The tangential and normal components of acceleration:

aN = ||v||

a = aT T + aN N,

where

aT =

d
||v||
dt

and

aN , observe that v = ||v|| T, and then dierdT


d ||v||
d ||v||
entiate both sides using the Product Rule to get a =
T + ||v||
=
T + ||v|| ||T0 (t)|| N;
dt
dt
dt
d
2
||v|| and aN = ||v|| ||T0 (t)|| = ||v|| .
then aT =
dt
Note that although aT can be positive, negative, or zero, aN is always nonnegative, because both
and ||v|| are nonnegative.
From a = aT T+aN N, one has va = ||v|| T(aT T + aN N) = ||v|| aT (T T)+||v|| aN (T N) =
va
and ||v a|| = ||v|| aN =
||v|| aN B. Then one can immediately derive the formulas B =
||v a||
2
||v|| .

0
0
0
x
00 y00 z00

x
000 y000 z000
0

x
y
z
B (t) N(t)
The torsion =
=
measures how sharply the path of the particle is
2
||v(t)||
||v a||
To see that

a = aT T + aN N for these values of aT

and

twisting out of its plane of motion.

Another relation satised by the torsion, which may be easier to use, is

Remark: By observing that

ds
= ||v||,
dt

we can rewrite the above as

dT
= N
ds

three Frenet-Serret formulas: the other two say

r(t) = hsin(3t), cos(3t), 4ti,

T, N, B, , , aT , aN .
q
||v|| = 9 cos2 (3t) + 9 sin2 (3t) + 16 = 5.

Hence

N = h sin(3t), cos(3t), 0i

so that

Curvature is given by

dB
=
dt

We have

Finally, we have

||dT/dt||
9/5
9
=
=
||v||
5
25


12 sin(3t) 12 cos(3t)

,
,0 .
5
5

aT = 0

As a sanity check, we



dT 9


dt = 5 .



i

3 cos(3t)
We compute the cross product B = TN =

5
sin(3t)

is one of the

nd

v(t) = h3 cos(3t), 3 sin(3t), 4i, so that




3 cos(3t) 3 sin(3t) 4
Hence T =
,
,
.
5
5
5


9 sin(3t) 9 cos(3t)
dT
=
,
,0
Next we compute
dt
5
5

dB
= N . This
ds
dN
= B T .
ds

dN
dT
formula follows from the denition of , while the
formula can be proven by writing
ds
ds
dN
dB
dT
N = B T and then taking the derivative with respect to s: one obtains
=
T+B
=
ds
ds
ds
(N T) + B (N) = B T.

We have

1 dB
.
||v|| dt

The

Example: For the curve

and

N =

j
3 sin(3t)

5
cos(3t)

k
4
5
0






4 cos(3t) 4 sin(3t) 3

,
,
=

5
5
5

Thus, torsion is given by

aN = 9 .
compute a(t) = h9 sin(3t), 9 cos(3t), 0i,

dB
N
12
= dt
=
||v||
25

and

and indeed

a(t) = 0 T + 9 N.

Well, you're at the end of my handout. Hope it was helpful.


Copyright notice: This material is copyright Evan Dummit, 2013. You may not reproduce or distribute this material
without my express permission.

Вам также может понравиться