What is Rn ?
• R × R × · · · × R (n times)
• points represented as “ordered n-tuples” (x1 , x2 , . . . , xn ) where each xi is a real number
Vectors
Vectors are oriented line segments. It is convenient to think of them as arrows connnecting pairs of points in
Rn . Consequently, they have a length and a direction. If u and v are two vectors they can be added together
by concatenating them “nose to tail”. That is, slide the vector v so that its tail (the end without the arrow)
is located at the same point as the head of u (the end with the arrow). Arranged like this, u and v appear to
form two sides of a triangle. The “third side” (the vector connecting the tail of u to the head of v) is u + v.
To do calculus, this needs to be made more concrete. For reference, we’ll slide v so that its tail sits at the
origin (0, 0, 0). If v connects the origin to a point (x, y, z) in 3d-space (for example), we’ll write
v = (x, y, z).
That is, a vector is written down in terms of its coordinates. General vector addition is done by coordinates:
u + v = (u1 , u2 , . . . , un ) + (v1 , v2 , . . . , vn ) = (u1 + v1 , u2 + v2 , . . . , un + vn ).
Note: Vectors v are written in boldface. By hand, distinguish vector quantities by writing v or ~v .
i
In R3 , the special vectors
i = (1, 0, 0)
j = (0, 1, 0)
k = (0, 0, 1)
are coordinate vectors and all other vectors in R3 can be built from them.
Example 2. Suppose that v = (1, 1, 2). Then
u · v := u1 v1 + u2 v2 + · · · + un vn .
• If v 6= 0 then
v
v̂ :=
kvk
is a unit vector, and v = kvk v̂.
• If u · v = 0 then u and v are orthogonal (or perpendicular) and write: u ⊥ v (the symbol ⊥ is called
“perp”).
√
Example 1. Let v = (2, 1, −2). Then v · v = 22 + 12 + (−2)2 = 9 so that kvk = 9 = 3 and
v 2 1 −2 v 2 1 −2
= , , so that v = kvk =3 , , .
kvk 3 3 3 kvk 3 3 3
ii
Remarks
π
• u ⊥ v ⇐⇒ cos θ = 0 ⇐⇒ θ = 2
u × v = (u2 v3 − u3 v2 , u3 v1 − u1 v3 , u1 v2 − u2 v1 ).
Remarks
Recall cofactor
expansion
+ − +
1. A useful trick for remembering the formula is that in 1st row · · ·
· · ·
i j k
det u1 u2 u3 = u × v.
v1 v2 v3
u×v = −v × u (0.1)
u×u = 0 (0.2)
(u × v) ⊥ u (0.3)
(u × v) ⊥ v. (0.4)
(Proved via the scalar triple below: see “More vector geometry”.)
3. Let θ be the angle between u and v. Then
We can think of the vector u × v as an “oriented area” of the parallelogram with sides u and v.
Explanation: First,
iii
(0.4) Lines and planes
In multivariable calculus, we frequently make linear approximations to nonlinear objects like curves and
surfaces; these linearisations are lines and planes, and we review them now. Throughout, points are repre-
sented by position vectors: r = (x, y, z).
Lines in the plane or in space: these are defined by a point, and a direction. In parametric form:
L = {r0 + t · v : t ∈ R}
where
- r0 is a point on line
• independent of dimension
• Example: The “vertical” line x = 2 can be written as r = (2, 0) + t(0, 1), t ∈ R, but not with y as a
function of x.
{(0, 3) + t(1, 2) : t ∈ R}
| {z }
ordered pairs (x, y)
We can use vector notation to elaborate the parametric form. Then,
x 0 1
= +t
y 3 2
| {z } | {z }
= r v
0
0+t
= .
3 + 2t
We write this as
x = t
y = 3 + 2t
Note that by substituting the first equation into the second one we can recover y = 3 + 2t = 3 + 2x.
Note: The parametric form of a line in Rn is obtained by solving a system of equations: usually Ax = b
where A is a (n − 1) × n matrix. The solution will have the parametric form r0 + t v where r0 is a particular
solution and v is a homogeneous solution. It is easy to check that Av = 0, so the “direction vector” of the
line lives in the null space of A. (The null space of a matrix is null(A) = {x : Ax = 0}. Sometimes this is
called the homogeneous solution space — does this sound familiar from differential equations?)
iv
Planes in R3
To build our intuition we start with the parametric form. Then
P = {r0 + su + tv | s, t ∈ R},
where
r = r0 + su + tv
where s, t are parameters and u, v are parallel to P . Then if n is any normal vector to P ,
r · n = (r0 + su + tv) · n
= r0 · n + s(u · n) + t(v · n)
= r0 · n + s(0) + t(0)
Note: If we let r = x the the plane equation can be made to look like Ax = b where A = nT and b = r0 · n.
It is therefore no surprise that one can move from the implicit to parametric form of a plane by solving a
suitable equation. To pass from the parametric to implicit form, you need to find a normal vector (n = u×v
is a good choice) and proceed as above.
v
Summary
Vectors are a convenient notation for describing objects in nd-space. The dot and cross products are impor-
tant for defining length, angle and orthogonality. These concepts will recur on many occasions in our study
of nd-calculus. The simplest objects of study in nd-space are the “flat” ones: lines and planes. It is essential
to gain a proper understanding of these objects via parametric descriptionsa and implicit desciptionsb . In
particular, you should learn to recognise the parametric and implicit representations of lines and planes in
R3 , and be familiar with how to obtain the various forms. You can review your first year notes, or continue
reading. Think about this: to what extent are lines and planes the “graph of a function”?
a
Explicit formulae for all the points on the object, swept out by parameters.
b
Equations that all the points on the object satisfy.
We can also work out the angle between u and v. On the one hand, u · v = 6, on the other
Thus, r
u·v 6 6
cos θ = √ √ = √ √ =
6 11 6 11 11
so
θ = 0.739 . . . (radians)
= 42.39 . . . (degrees).
vi
5. Let u = (1, 1, 2), v = (1, −1, 3). Then
i j k
1 2 1 2 1 1
• u×v = 1 1 2 =
i−
j+ k = 5 i−j−2 k = (5, −1, −2).
1 −1 3 −1 3 1 3 1 −1
p √ √
• ku × vk = k(5, −1, −2)} = 52 + (−1)2 + (−2)2 = 25 + 1 + 4 = 30.
i j k
1 2 1 2 1 1
• u×(u×v) = 1 1 2 = i− j+
5 −1 k = 0 i+12 j−6 k =
5 −1 −2 −1 −2 5 −2
1. Compute u × v , v × w.
2. Compute w × u , u × w , v × u.
3. Sketch u, v and u × v on the same axes.
u · (v × w) = (u × v) · w
h = kwk cos φ
vii
But n · w = knk kwk cos φ, so
h = kwk cos φ = n · w/knk. (0.6)
Finally, since n = u × v is a suitable normal vector, we can put together (0.5) and (0.6) to get:
(u × v) · w
vol = ku × vk = (u × v) · w.
ku × vk
Thus
v2 v3 v1 v3 v1 v2
u · (v × w) = u1 − u2
w1 w3 + u3 w1 w2
w2 w3
u1 u2 u3
= v1 v2 v3 .
w1 w2 w3
Proof of scalar triple product identity: Every time a row is exchanged in a matrix, the determinant changes
sign. Thus
u1 u2 u3
u · (v × w) =
v1 v2 v3 [Exchange rows 1 and 3]
w1 w2 w3
w1 w2 w3
= − v1 v2 v3 [Exchange rows 2 and 3]
u1 u2 u3
w1 w2 w3
= + u1 u2 u3 = w · (u × v) = (u × v) · w.
v1 v2 v3
viii
Proof of (0.3): (u × v) ⊥ u. We calculate
Thus, u is orthogonal to u × v.
2. Find a parametric representation of the line L which is the solution set of the equations:
x + y + z = 2
x + y − z = 0.
Solution: L is the set of (x, y, z) which satisfy both equations. We must solve
x
1 1 1 y = 2 .
1 1 −1 0
z
The 2nd row immediately yields z = 1, the lack of a leading 1 in the y column lets us put y = t so
the 1st row reads x + t = 1. Thus
x 1 −1
y = 0 + t 1 .
z 1 0
ix
3. Sketch the three vectors (1, 1, 0) , (0, 1, 1) and (1, 0, −1). Do these three vectors lie in a plane through
(0, 0, 0)?
Solution: Write the vectors as
a = (1, 1, 0) = i + j
b = (0, 1, 1) = j + k
c = (1, 0, −1) = i − k
x
6. Find a parametric form of the plane x + 2 z = 3.
Solution: What one has to do is find the general solution to the matrix equation
x
1 0 2 y =3 which has augmented matrix 1 0 2 3 .
z
Since the matrix is in echelon form, we can see immediately that there are free parameters corre-
sponding to the second and third columns of the augmented matrix; namely, the variables y and z.
We thus set y = s, z = t and solve x = 3 − 0 y − 2 z = 3 − 2 t. Writing out in matrix notation:
x 3 0 −2
y = 0 + s 1 + t 0 , s, t ∈ R.
z 0 0 1
(x, y) = 12 t, t + 3 t ∈ R
1
= (0, 3) + t · ,1
2
| {z }
parametric form of a line
Each t gives you a point on L. Note: the role of the point (0, 3), and
(tangent/direction) vector (1/2, 1) on adjacent diagram.
xi
Example 2. Let r(t) = (t, t2 ) for t ∈ R.
| {z }
vector in R2
So “(x, y) is a function of t”.
t (x, y)
-2 (-2,4)
-1 (-1,1)
0 (0,0)
1 (1,1)
2 (2,4)
3 (3,9)
This describes a parabola as a parametrised curve.
Example 3. Let (x, y) = (2 cos t, 2 sin t) for −π2
≤ t ≤ π2 . We could make up a table of t values and sketch
the points on some axes, but it is faster to use M APLE.
plot([2*cos(t),2*sin(t),t=-Pi/2..Pi/2]);
You can even animate in real time:
with(plots): animatecurve([2*cos(t),2*sin(t),t=-Pi/2..Pi/2]);
Because the axes are a bit skewed, close inspection is needed to see that this is really a semi-circle. In fact,
a bit of algebra confirms: Since x = 2 cos t and y = 2 sin t,
x2 + y 2 = 4 cos2 t + 4 sin2 t
= 4(cos2 t + sin2 t) = 4.
Thus, this parametrised curve is part of the circle of radius 2.
Remarks
xii
since a curve needn’t be the graph of a function. (Eg. a circle requires two functions—one for the top
and one for the bottom). We will also see below that parametric descriptions do not give preference
to any variables when differentiating or integrating.
We’d like to visualise, and compute the tangent line at the point (0, 1, π/2). The first is easily accomplished
in M APLE:
with(plots): spacecurve([cos(t),sin(t),t],t=0..100);
(obviously we need to specify a finite parameter range). A more dynamic version is:
with(plots): animate(spacecurve,[[cos(t),sin(t),t],t=0..T],T=0..100);
This is an appealling approach. To figure out the tangent line, we are going to have think a bit more about
the direction of a curve.
Fill in the table, sketch each point in the r column as a dot. Use
each r point as the tail point of the corresponding drdt
. Join the
dots to form a curve. Observe apparent tangents.
xiii
Interpretation as tangent vectors
If {r(t) : t ∈ [a, b]} is a parametrised curve, then
dr
v(t) =
dt
is the tangent vector to the curve at t. The tangent line
to the curve at t = t0 is
Sketch justification: Let r(t) = (f (t), g(t)); the extension to space curves is obvious.
df dg
dr
r(t0 + ∆t) = (f (t0 + ∆t), g(t0 + ∆t)) ≈ (f (t0 ), g(t0 )) + ∆t ,
dt dt
= r(t0 ) + ∆t (t0 ). (0.7)
dt
The right-hand side of (0.7) is the parametric form of a line, with parameter ∆t. The line proceeds “tangen-
tially” for all values of the parameter: we replace ∆t with s and v = drdt
.
Note: The key estimate in the above is
dr
∆r = r(t0 + ∆t) − r(t0 ) ≈ ∆t.
dt
The ≈ means when ∆t is small, the error is very small! This is non-rigorous, but we can fix that later when
we’ve got a proper a definition of differentiability!
Example 1 revisited. To compute the tangent line at (0, 1, π/2) requires several steps:
Step 1: Check that the point is on the curve! We need to find a value of t0 such that
cos t0 0
sin t0 = 1 . Clearly, t0 = π/2 .
π
t0 2
xiv
To get the tangent vector, evaluate at t = t0 :
− sin π2
− sin t0 −1
v = cos t0 = cos π2 = 0
1 1 1
0 −1
L = 1 + s. 0 : s ∈ R .
π
1
2
d
v = r = ṙ = r0
dt
d
a = v = r̈ = r00 .
dt
Exercises
1. Prove that if a ∈ R and v ∈ Rn then ka vk = |a| kvk. Deduce that kv̂k = 1.
xv
3. Derive a formula for decomposing a vector v into a component parallel to u and a component orthog-
onal to u: v = vk + v⊥ .
(a) u · v = v · u;
(b) (u + v) · w = u · w + v · w
(c) a(u · v) = (au) · v
7. Compute (1, 3, 0) × (0, −1, 2) and as many similar ones as necessary to get the right answer!
10. Show that (except in degenerate cases), (u × v) × w lies in the plane of u and v, whereas u × (v × w)
lies in the plane of v and w. What are the degenerate cases?
12. Forensic example. A car window was broken after being hit by a bullet. Measurements of the
bullet reveal that it struck the window with an angle θ, measured from the normal to the car window.
Assume that the car window is a plane, aligned parallel to the y–axis and making an angle φ with
the z–axis. The forensic problem is to find the line along which the shot was fired. It is known that
the shot was fired in the xy–plane, at an unknown angle α from the x–axis. Show that α obeys the
formula
−1 sin(π/2 − θ)
α = π/2 − sin .
sin(π/2 − φ)
[Hint: work out formulae for the normal vector to the window and the tangent vector to the bullet’s
path, then compute the angle between them.]
13. A water main is to be constructed with a 20% grade in the north direction and a 10% grade in the east
direction. Determine the angle θ in the pipe as it turns from north to east.
Find the exact time(s) where the particles collide, and the directions of their motion at collision.
15. Two helicopters set off from Hamilton airport at time t = 0. They follow the paths r1 (t) =
(−140 t, −140 t, 5t) and r2 (t) = (200 t, 0, 4 t) – t is measured in hours, and distances in kilome-
tres. After a short flight, the first helicopter encounters problems and lands immediately on Pirongia
at coordinates (−28, −28, 1). Three minutes later, the second helicopter alters its course to rescue the
occupants of the first helicopter. It flies in a straight line towards the landing site at 250km/hr. How
long does it take for the second helicopter to get there?
xvi
16. Circular motion. The parameterisation
describes motion along a circle of radius r. Use differentiation to compute the velocity v = dr dt
and
acceleration a = dv
dt
. Try using M APLE to make a plot of the paramtrised curve (you’ll need to specify
a radius r for the circle). Add a few vectors v and a (at appropriate points on the curve). What do
you notice about the direction of a? Calculate v · a and interpret your answer.
18. Write the plane 2 x + 3 y + z = 4 in parametric form. That is, (x, y, z) as functions of real parameters
s, t.
19. Let r0 ∈ R3 , and let n1 , n2 be two non-zero vectors. Let two planes be described by:
P1 = r ∈ R3 : (r − r0 ) · n1 = 0 ,
P2 = r ∈ R3 : (r − r0 ) · n2 = 0 .
Explain why L = P1 ∩ P2 .
(b) If n1 × n2 = 0, describe the set P1 ∩ P2 .
(a) x + 5 y = 6
(b) x2 + 2 x + 4 y − y 2 = 2
(c) x + x y = 2
(d) x + x y + y 2 = 3
dy
21. For each of the formulae in the previous question calculate dx at the indicated point by (i) differenti-
ating the formula for y as a function of x; (ii) implicit differentiation