Calculus III

Chapter 13 - Vector Functions

1. Vector Functions and Space Curves

A function r : R → Vn is called a vector-valued function. That is, the input is a real number
and the output is a vector. For V3 , we write

r(t) = hf (t), g(t), h(t)i = f (t)i + g(t)j + h(t)k.

We call f (t), g(t), h(t) the component functions of r. The domain of r is the intersection of
the domains of its component functions.

We find the limit of r as t → a by taking the limit of its component functions. Hence,

lim r(t) = hlim f (t), lim g(t), lim h(t)i.

t→a t→a t→a t→a

Then r is continuous at a if
lim r(t) = r(a).
t→a

Example 1. Consider the vector valued function

t2 − t √ sin πt
r(t) = i + t + 8j + k.
t−1 ln t
We identify the component functions as the coefficients of i, j, k. The domain of r(t) is then
 2
t −t
 √  
sin πt

D ∩D t+8 ∩D = {t : t 6= 1} ∩ {t : t ≥ −8} ∩ {t : t > 0 and t 6= 1}
t−1 ln t
= (0, 1) ∪ (1, ∞).

The function r(t) is not continuous at t = 1, however

t2 − t √
   
  sin πt
lim r(t) = lim i + lim t + 8 j + lim k
t→1 t→1 t − 1 t→1 t→1 ln t

= i + 3j − πk.

For the last limit, we applied L’Hospital’s rule to get,

sin πt LR π cos πt
lim = lim = −π.
t→1 ln t t→1 1/t
1
Given a curve C in three-dimensional space, we can think of tracing that curve out over
time (think of a particle moving in some direction along the curve). If we restrict to only
recording the x coordinate of the particle at various intervals, we obtain a function f (t) of
time. We can then do this for y and z. This is the basic idea of space curves.

Let f, g, h be continuous, real-valued functions on an interval I. The set C of all points

(x, y, z) with x = f (t), y = g(t), and z = h(t) as t varies over I, is called a space curve. We
call these functions the parametric equations of C and call t the parameter.

Example 2. The curve whose vector equation is r(t) = cos ti + sin tj + tk is a helix. Note
that all of the points lie on the cylinder x2 + y 2 = 1.

Example 3. Consider the intersection of the paraboloid z = 4x2 + y 2 and the parabolic
cylinder y = x2 . Set x = t, then the equation for the parabolic cylinder gives y = t2 . The
paraboloid’s equation then says that z = 4t2 + t4 . Thus, the intersection is the space curve,

r(t) = ht, t2 , 4t2 + t4 i.

2
 2. Derivatives and integrals of vector functions

Let C be a space curve with equation r(t). Let P and Q be points on C corresponding to
−→
r(t) (for some t) and r(t + h) (for some h > 0), respectively. We call P Q a secant vector of
the curve and its equation is r(t + h) − r(t). The scalar multiple (1/h)(r(t + h) − r(t)) has
−→
the same direction as P Q. We let h → 0 to obtain the tangent vector to the curve C defined
by r(t) at the point P . That is, the derivative r0 of a vector function r is
dr r(t + h) − r(t)
= r0 (t) = lim ,
dt h→0 h
if the limit exists and r0 (t) 6= 0. The tangent line to C at P is the line through P parallel to
r0 (t). The unit tangent vector is defined as
r0 (t)
T(t) = .
|r0 (t)|

The next theorem just says that, in nice cases, we can compute the derivative componentwise.

Theorem 4. If r(t) = hf (t), g(t), h(t)i where f , g, and h are differentiable functions, then
r0 (t) = hf 0 (t), g 0 (t), h0 (t)i.

Example 5. Consider the vector function r(t) = he−t , sin2 (t), ln ti. All three components
are differentiable functions, and
1
r0 (t) = h−e−t , 2 sin(t) cos(t), i.
t

Most of the differentiation rules you know from Calc I still apply to vector functions.

Differentiation Rules: Suppose u and v are differentiable vector functions, c is a scalar,

and f is a real-valued function.
d d
(i) [u(t) + v(t)] = u0 (t) + v0 (t) (iv) [u(t) · v(t)] = u0 (t) · v(t) + u(t) · v0 (t)
dt dt
d d
(ii) [cu(t)] = cu0 (t) (v) [u(t) × v(t)] = u0 (t) × v(t) + u(t) × v0 (t)
dt dt
d d
(iii) [f (t)u(t)] = f (t)u0 (t) + f 0 (t)u(t) (vii) [u(f (t))] = f 0 (t)u0 (f (t)).
dt dt
Example 6. Suppose |r(t)| = c where c is a constant. Then r(t) · r(t) = |r(t)|2 = c2 is a
constant. Hence, by property (4),
d
0= [r(t) · r(t)] = r0 (t) · r(t) + r(t) · r0 (t) = 2r0 (t) · r(t).
dt
Thus, r0 (t) · r(t) = 0, so r(t) and r0 (t) are orthogonal.
3
Integrals of vector functions can be defined analogously. Suppose r(t) is a continuos vector
function. Let [a, b] be an interval in the domain of r(t) and consider a (regular) partition
of [a, b] into n subintervals, each of length ∆t. In each subinterval, choose a representative
point t∗i . The definite integral of r(t) = hf (t), g(t), h(t)i is then defined as
Z b Xn
r(t) dt = lim r(t∗i )∆t
a n→∞
i=1
" n
! n
! n
! #
X X X
= lim f (t∗i )∆t i + g(t∗i )∆t j + h(t∗i )∆t k .
n→∞
i=1 i=1 i=1

Hence, Z b Z b  Z b  Z b 
r(t) dt =
f (t) dt i + g(t) dt i + h(t) dt i.
a a a a
R R
As with real-valued functions, we use r(t) dt for the indefinite integral. Hence, r(t) dt =
R(t) + C where R(t) is a vector function such that R0 (t) = r(t) (an antiderivative of r(t))
and C is a constant vector.

Example 7. Using standard integration techniques, we have

Z
1 1 1
te2t i + j + t cos(πt)k dt = (2t − 1)e2t i + ln |t + 1|j + 2 (cos(πt) + tπ sin(πt))k + C.
t+1 4 π

Fundamental Theorem of Calculus for vector functions If r(t) be a continuous vector

function and R(t) is an antiderivative of r(t), then
Z b
r(t) dt = [R(t)]ba = R(b) − R(a).
a

4
 3. Arc length and curvature

The formula for arc length of a space curve is analogous to that of a real valued function.
If r(t) = hf (t), g(t), h(t)i is a vector equation of a space curve on the interval [a, b] with
f 0 (t), g 0 (t), h0 (t) continuous, and if the curve is traversed exactly once as t increases from a
to b, then the length of the curve is
Z b Z bp
0
L= |r (t)| dt = [f 0 (t)]2 + [g 0 (t)]2 + [h0 (t)]2 dt.
a a

Example 8. Let r(t) = cos ti + sin tj + ln cos tk be the equation of the space curve C on the
interval [0, π/4]. Then the length of C is
Z π/4 Z π/4 p
0
L= |r (t)| dt = [f 0 (t)]2 + [g 0 (t)]2 + [h0 (t)]2 dt
0 0
s 2
Z π/4 
− sin t
= [− sin t]2 + [cos t]2 + dt
0 cos t
Z π/4 p Z π/4 √ Z π/4
2 2
= 1 + tan t dt = sec t dt = sec t dt
0 0 0
 √  √
= [ln | sec t + tan t|]π/4
0 = ln( 2 + 1) − (ln(1 + 0)) = ln( 2 + 1).

One comment about this argument. We should really have put | sec t| when we took the
integral. However, because sec t is positive on this interval.

Note: A curve may have multiple parameterizations, but all parameterizations will give the
same arc length. Another way of saying this is that arc length is independent of parameter-
ization of the curve.

If a curve C is given by a vector function r(t) = f (t)i + g(t)j + h(t)k on the interval [a, b]
where r0 (t) is continuous and C is traversed exactly once as t increases from a to b. The arc
length function s is defined as
Z t
s(t) = |r0 (u)| du.
a
Note that we have replaced the variable of integration by u since t is being used by the
bound of integration. Differentiating both sides of this equation give
ds
= |r0 (t)|.
dt
Example 9. Consider the helix r(t) = cos ti + sin tj + tk. We choose the point (1, 0, 0)
(corresponding to t = 0) as our initial point from which to measure arc length and we move
5
in the direction of increasing t. The arc length function is then
Z t Z t√ √
2 2
s(t) = |(− sin t) + cos t + 1| du = 2 du = 2t.
0 0
√
We can now reparameterize our curve by setting s = s(t) and solving for t, so t = s/ 2, and
√ √ √
r(t(s)) = cos(s/ 2)i + sin(s/ 2)j + (s/ 2)k.

This is called parameterization of the curve with respect to arc length. If we set s = 1, the
r(t(1)) is the positive vector of the point 1 unit from (1, 0, 0).

A parameterization r(t) is called smooth on an interval I if r0 is continuous and r0 (t) 6= 0 on

I. A curve C is called smooth if it has a smooth parameterization r(t). For such a curve,
the unit tangent vector T(t) is given by
r0 (t)
T(t) = 0 .
|r (t)|
Recall this indicates the direction of the curve but it can also tell us how “sharp” C is as it
bends. This is the idea behind curvature. That is, curvature is the measure of how quickly
the curve changes direction at a point. Formally, curvature is defined as the magnitude of
the rate of change of the unit tangent vector with respect to arc length, so
 
 dT 
κ =  
ds
where T is the unit tangent and s is arc length. By the chain rule,
dT dT ds
=
dt ds dt
and so
 dT   dT/dt  |T0 (t)|
   
κ(t) =   = 
   = .
ds ds/dt  |r0 (t)|

Example 10. Consider the curve with vector equation r(t) = ht, 21 t2 , t2 i. Note that r0 (t) =
h1, t, 2ti. The unit tangent vector is
r0 (t) h1, t, 2ti
T(t) = 0
=√ .
|r (t)| 1 + 5t2
By the product rule,
5t h0, 1, 2i
T0 (t) = − 2 3/2
(h1, t, 2ti) + √
(1 + 5t ) 1 + 5t2
1 2 2 2 2

h−5t, 1, 2i
= 2 3/2
−5ht, t , 2t i + h0, 1 + 5t , 2 + 10t i = .
(1 + 5t ) (1 + 5t2 )3/2
6
Thus,
√ √ √ √
|T0 (t)| |h−5t, 1, 2i|/(1 + 5t2 )3/2 25t2 + 5 5 1 + 5t2 5
κ= 0 = √ = 2 2
= 2 2
= .
|r (t)| 1 + 5t2 (1 + 5t ) (1 + 5t ) (1 + 5t2 )3/2
√
Thus, at the point (2, 1, 4), corresponding to t = 2, the curvature is 5/(21)3/2 .

Theorem 11. The curvature of a curve given by the vector function r is

|r0 (t) × r00 (t)|

κ(t) = .
|r0 (t)|3

Example 12. Consider the previous example. We have r00 (t) = h0, 1, 2i and a computation
shows that
 
i j k 
 
r0 (t) × r00 (t) = 1 t 2t = 0i − 2j + k.
 
 
0 1 2 

√
The magnitude of this vector is 5. Hence, by the theorem,
√
|r0 (t) × r00 (t)| 5
κ(t) = = .
|r0 (t)|3 (1 + 5t2 )3/2

A common (and useful) technique in Calc I/II is linear approximation. In this, we construct
a line that approximates a curve near a point using tangents. In 3d space, given a curve (say
a helix), we can ask what plane best approximates the curve near a point. We do this using
tangent vectors.

Let r(t) be a smooth space curve with unit tangent vector T(t). Since |T(t)| = 1 for all t,
then T(t) · T0 (t) = 0. Hence, T0 (t) is orthogonal to T(t), but it need not be a unit vector.
At any point where κ 6= 0, we define the (principal) unit normal vector as

T0 (t)
N(t) = .
|T0 (t)|

This vector tells us the direction the curve is turning at each point. The binormal vector is
B(t) = T(t) × N(t) and it is orthogonal to both T(t) and N(t).

The vectors B and N at a point P on a curve C determine a plane, called the normal plane
of C at P . It consists of all lines that are orthogonal to the tangent vector T.
7
Example 13. Consider the circular helix with equation r(t) = cos ti + sin tj + tk. We will
determine the normal plane of C at the point P (1, 0, 0). Standard computations show
r0 (t) h− sin t, cos t, 1i
T(t) = = √
|r0 (t)| 2
1
T0 (t) = √ h− cos t, sin t, 0i
2
T0 (t) 1 h− cos t, sin t, 0i
N(t) = 0 =√ √ = h− cos t, sin t, 0i
|T (t)| 2 1/ 2
1
B(t) = T(t) × N(t) = √ hsin t, cos t, 1i.
2
Now the normal plane to C at P has normal vector r0 (0) = √1 h0, 1, 1i.
2
Hence, an equation
of the plane is
0(x − 1) + 1(y − 0) + 1(z − 0) = 0,
or just y + z = 0.

The vectors T and N also determine a plane, called the osculating plane of C at P . It is the
plane that comes closes to containing the part of the curve near P . Hence if the curve lies
in a plane (is a planar curve) then this is just the plane that contains the curve.

An easy computation shows that the curvature κ of a circle of radius a is 1/a (see Example
3 on page 864). This implies that the radius of a circle is 1/κ. The circe that lies in the
osculating plane of C at P , has the same tangent as C at P , lies on the concave side of C
(toward which N points), and has radius ρ = 1/κ is called the osculating circle of C at P .

8
 4. Motion in space

Recall that we can think of a vector function r(t) as tracing the path (or position) of a particle
moving along a space curve. With that interpretation, it follows that r0 (t) represents the
velocity vector v(t). The speed of the particle is the magnitude of the velocity vector, so
|v(t)| = |r0 (t)|. The acceleration is a(t) = r00 (t).

Example 14. Suppose a particle’s acceleration on a curve is given by the vector function
a(t) = sin ti + 2 cos tj + 6tk. Further suppose that v(0) = −k and r(0) = j − 4k (that is,
these are the velocity and position vectors at time 0). We wish to find the functions for
velocity and position. To do this, we use integration:
Z
v(t) = a(t) dt = − cos ti + 2 sin tj + 3t2 k + C.

Thus, −k = v(0) = −i + C, so C = i − k and hence,

v(t) = (1 − cos t)i + 2 sin tj + (3t2 − 1)k.

Now Z
r(t) = v(t) dt = (1 − sin t)i − 2 cos tj + (t3 − t)k + C.
Thus,
j − 4k = r(0) = i − 2j + D.
Thus, D = −i + 3j − 4k and so

r(t) = (− sin t)i + (3 − 2 cos t)j + (t3 − t − 4)k.

Newton’s Second Law of Motion: If, at any time t, a force F(t) acts on an object of
mass m produces an acceleration a(t), then F(t) = ma(t).

Example 15. Suppose an object with mass m moves in a circular path with constant angular
speed ω and has position vector r(t) = a cos ωti + a sin ωtj. The velocity and acceleration
are then given by

v(t) = −aω sin ωti + aω cos ωtj

a(t) = −aω 2 cos ωti − aω 2 sin ωtj.

Thus, Newton’s Second Law gives,

F(t) = ma(t) = −mω 2 r(t).

Hence, the foce acts in the direction opposite to the radius vector r(t) and so points in the
direction of the origin (centripetal force).
9
Suppose a projectile is fired with angle of elevation α and initial velocity v0 .

There is no loss in assuming that the projectile starts at the origin. The force of gravity acts
downward, and hence,
F = ma = −mgj

where g = |a| ≈ 9.8m/s2 . Thus, a = −gj and since v0 (t) = a, we have

v(t) = −gtj + C

where C = v(0) = v0 . Hence, r0 (t) = v(t) = −gtj + v0 . Integrating again we find

1
r(t) = − gt2 j + tv0 + D
2
and D = r(0) = 0. Hence, r(t) = − 21 gt2 j + tv0 . Write |v0 | = v0 for initial speed of the
projectile, then by the Pythagorean Theorem,

v0 = v0 cos αi + v0 sin αj

and so
1
r(t) = (v0 cos α)ti + [(v0 sin α)t − gt2 ]j.
2
The parametric equations of trajectory are then
1
x = (v0 cos α)t y = (v0 sin α)t − gt2 .
2
Example 16. A batter hits a baseball 3 ft above the ground toward the center field fence,
which is 10 ft high and 400 ft from home plate. The ball leaves the bat with speed 115 ft/s
at an angle 50◦ above the horizontal. Is it a home run? (Does the ball clear the fence?)
10
We denote by x the distance from home plate and by y the height above the ground. Thus,
r0 = 3j. Initial velocity is determined by the above,

v(0) = v0 = v0 cos αi + v0 sin αj = 115 cos 50◦ i + 115 sin 50◦ j.

Acceleration is a = −gj where g ≈ 32.174 ft/s2 . Using the above, but adjusting for r0 , we
have
1
r(t) = r0 − gt2 j + tv0
2
1 2
= 3j − gt j + t(115 cos 50◦ i + 115 sin 50◦ j)
2
1
= 115 cos 50◦ ti + (3 − gt2 + 115 sin 50◦ t)j.
2
Hence,
1
x = 115 cos 50◦ t and y = 3 − gt2 + 115 sin 50◦ t.
2
The ball reaches the fence when x = 400. To find the time that this happens, we set

400 = 115 cos 50◦ t ⇒ t ≈ 5.4.

Hence, the ball reaches the fence 5.4 seconds after it leaves the bat. At this time, y(500) ≈
8.654 ft, which isn’t high enough to clear the fence. Hence, it’s not a homerun1.

Wrote v = |v| for the speed of a particle, then

r0 (t) v(t) v
T(t) = 0
= = .
|r (t)| |v(t)| v
Hence, v = vT. Differentiating both sides gives

a = v0 = v 0 T + vT0 .

Recall that curvature is given by

|T0 | |T0 |
κ= = .
|r0 | v
Hence, |T0 | = κv. The unit normal is N = T0 /|T0 | so T0 = |T0 |N = κvN and hence,

a = v 0 T + κv 2 N.

We set the coefficients of T and N as aT = v 0 and aN = κv 2 , respectively, and call these the
tangential component and normal component of acceleration.

Note that
v · a = vT · (v 0 T + κv 2 N) = vv 0 T · T + κv 3 T · N = vv 0 .
1Unless a fan reaches over and grabs it and the umpire doesn’t see it.
11
Some comment s regarding this computation are in order. Since |T| = 1, then T·T = |T|2 =
1. Moreover, as T is a vector function with constant magnitude, then an earlier computation
(Example 6) shows that T · N = 0. Hence,
r0 (t) · r00 (t)
aT = .
|r0 (t)|
Using Theorem 11, we also have
|r0 (t) × r00 (t)|
aN = .
|r0 (t)|

12