A1 2015 1 / 1

A1 Vector Algebra and Calculus

Prof David Murray

[email protected]
www.robots.ox.ac.uk/∼dwm/Courses/2VA

8 lectures, MT 2015
A1 2015 2 / 1

Vector Algebra and Calculus

1 Revision of vector algebra, scalar product, vector product

2 Triple products, multiple products, applications to geometry

3 Differentiation of vector functions, applications to mechanics

4 Scalar and vector fields. Line, surface and volume integrals,

curvilinear co-ordinates

5 Vector operators — grad, div and curl

6 Vector Identities, curvilinear co-ordinate systems

7 Gauss’ and Stokes’ Theorems and extensions

8 Engineering Applications
A1 2015 3 / 1

More Algebra & Geometry using Vectors

In which we discuss ...

Vector products:
Scalar Triple Product, Vector Triple Product, Vector Quadruple
Product

Geometry of Lines and Planes

Solving vector equations

Angular velocity and moments

A1 2015 4 / 1

Triple and multiple products

Using mixtures of scalar products and vector products, it is possible to
derive

“triple products” between three vectors

n-products between n vectors.

Nothing new about these

— but some have nice geometric interpretations ...
We will look at the

Scalar triple product

Vector triple product

Vector quadruple product

A1 2015 5 / 1

Scalar triple product a · (b × c)

Scalar triple product given by the true determinant

a1 a2 a3
a · (b × c) = b1 b2 b3
c1 c2 c3

Your knowledge of determinants tells you that if you

swap one pair of rows of a determinant, sign changes;
swap two pairs of rows, its sign stays the same.
Hence
(i) a · (b × c) = c · (a × b) = b · (c × a) (Cyclic permutation.)
(ii) a · (b × c) = −b · (a × c) and so on. (Anti-cyclic permutation)
(iii) The fact that a · (b × c) = (a × b) · c allows the scalar triple product
to be written as [a, b, c].
This notation is not very helpful, and we will try to avoid it below.
A1 2015 6 / 1

Geometrical interpretation of scalar triple product

The scalar triple product gives the volume of the parallelopiped whose
sides are represented by the vectors a, b, and c.

c cos β
c
Vector product (a × b) has magnitude
β equal to the area of the base × height
b in direction perpendicular to the base.
a

The component of c in this direction is equal to the height of the

parallelopiped, hence

volume of parallelopiped = |(a × b) · c|

A1 2015 7 / 1

Linearly dependent vectors

If the scalar triple product of

three vectors n

a · (b × c) = 0 b
c
then the vectors are linearly
dependent. a

a = λb + µc

You can see this immediately either using the determinant

— The determinant would have one row that was a linear combination
of the others
or geometrically for a 3-dimensional vector.
— the parallelopiped would have zero volume if squashed flat.
A1 2015 8 / 1

Vector triple product a × (b × c)

a
a × (b × c) is perpendicular to (b × c) bx c
In arbitrary direction
but (b × c) is perpendicular to b and c. c
So a × (b × c) must be coplanar with b and
b
c.
⇒a × (b × c) = λb + µc a x (bx c)

(a × (b × c))1 = a2 (b × c)3 − a3 (b × c)2

= a2 (b1 c2 − b2 c1 ) + a3 (b1 c3 − b3 c1 )
= (a2 c2 + a3 c3 )b1 − (a2 b2 + a3 b3 )c1
= (a1 c1 + a2 c2 + a3 c3 )b1 − (a1 b1 + a2 b2 + a3 b3 )c1
= (a · c)b1 − (a · b)c1

Similarly for components 2 and 3, so

a × (b × c) = (a · c)b − (a · b)c
A1 2015 9 / 1

Projection using vector triple product

Books say that the vector projection of any v
vector v into a plane with normal ^
n is n
n × (v × ^
vIN PLANE = ^ n)
We would say that the component of v in the ^ n
direction is v · ^
n, so the vector projection is v INPLANE

vIN PLANE = v − (v · ^n)^n

Can we reconcile the two expressions? (Yes we can.)
n ← a, v ← b, ^
Subst. ^ n ← c, into our earlier formula

a × (b × c) = (a · c)b − (a · b)c
n × (v × ^
^ n) n ·^
= (^ n · v)^
n)v − (^ n
= v − (v · ^
n)^
n

Fantastico! But v − (v · ^ n is much easier to understand ...

n)^
... and cheaper to compute!
A1 2015 10 / 1

Vector Quadruple Product (a × b) × (c × d)

We have just learned that

p × (q × r) = (p · r)q − (p · q)r
⇒ (a × b) × (c × d) = ??

Regarding a × b as a single vector

⇒ vqp must be a linear combination of c and d

Regarding c × d as a single vector

⇒ vqp must be a linear combination of a and b.

Substituting in carefully (you check ...)

(a × b) × (c × d) = [(a × b) · d]c − [(a × b) · c]d

and also = [(c × d) · a]b − [(c × d) · b]a
A1 2015 11 / 1

Vector Quadruple Product /ctd

Using just the R-H sides of what we just wrote ...

[(a × b) · c] d = [(b × c) · d] a + [(c × a) · d] b + [(a × b) · d] c

So
[(b × c) · d] a + [(c × a) · d] b + [(a × b) · d] c
d =
[(a × b) · c]
= αa + βb + γc .

a
Oh, we saw this yesterday ... ... the d

projection of a 3D vector d onto a c

basis set of 3 non-coplanar vectors is
UNIQUE. b
A1 2015 12 / 1

♣ Example
Q:
Use the quadruple vector product to express the vector d = [3, 2, 1] in
terms of the vectors a = [1, 2, 3], b = [2, 3, 1] and c = [3, 1, 2].
A:

[(b × c) · d] a + [(c × a) · d] b + [(a × b) · d] c

d=
[(a × b) · c]
So, grinding away at the determinants, we find
(a × b) · c = −18 and (b × c) · d = 6
(c × a) · d = −12 and (a × b) · d = −12.
So
1
d = (6a − 12b − 12c)
−18
1
= (−a + 2b + 2c)
3
A1 2015 13 / 1

Geometry using vectors: Lines

^
Equation of line passing through λb
point a1 and lying in the direction of Point r traces
a out line.
vector b is
r

r = a + βb
NB! Only when you make a unit vector in the dirn of b does the
parameter take on the length units defined by a:

r = a + λ^
b

For a line defined by two points a1 and a2

r = a1 + β(a2 − a1 )

or the unit version ...

r = a1 + λ(a2 − a1 )/|a2 − a1 |
A1 2015 14 / 1

The shortest distance from a point to a line

Vector p from c to ANY line point r is

λb
p = (r − c) = a + λ^
b − c = (a − c) + λ^
b
a
r−c
r
which has length squared
c
p 2 = (a − c)2 + λ2 + 2λ(a − c) · ^
b.

Easier to minimize p 2 rather than p itself.

d 2
p =0 when λ = −(a − c) · ^
b.
dλ

So the minimum length vector is p = (a − c) − [(a − c) · ^

b]^
b.

No surprise! It’s the component of (a − c) perpendicular to ^

b.
A1 2015 15 / 1

Shortest distance between two straight lines

Shortest distance from point to line is along the perp line
⇒ shortest distance between 2 lines is along mutual perpendicular.
The lines are:
r = a + λ^b r = c + µ^
d c
µd
λb
The unit vector along the mutual perp is P
a
Q
^
b ×^d
^
p= .
^ ^
|b × d|
(Yes! Don’t forget that ^
b ×^d is NOT a unit vector.)
The minimum length is therefore the component of (a − c) in this
direction !
^
b ×^
d
pmin = (a − c) · .
|^
b ×^
d|
A1 2015 16 / 1

♣ Example

Q: for civil engineers who like pipes

Two long straight pipes are specified using
Cartesian co-ordinates as follows:
Pipe A: diameter 0.8; axis through points
(2, 5, 3) and (7, 10, 8).
Pipe B: diameter 1.0; axis through points
(0, 6, 3) and (−12, 0, 9).
Do the pipes need re-aligning to avoid
intersection?
A1 2015 17 / 1

♣ /ctd
A: Pipes A and B have axes:
√
rA = [2, 5, 3] + λ0 [5, 5, 5] = [2, 5, 3] + λ[1, 1, 1]/ 3
√
rB = [0, 6, 3] + µ0 [−12, −6, 6] = [0, 6, 3] + µ[−2, −1, 1]/ 6

[0,6,3] [7,10,8]
Non-unit perpendicular to both axes is

ı̂ ̂ k̂ b p
p= 1 1 1 = [2, −3, 1] [−12,0,9]
−2 −1 1
a [2,5,3]

The length of the mutual perpendicular is

[2, −3, 1] [2, −1, 0] · [2, −3, 1]

(a − b) · √ = √ = 1.87 .
14 14
Sum of the radii of the pipes is 0.4 + 0.5 = 0.9. ⇒no collision
A1 2015 18 / 1

Three ways of describing a plane. Number 1

1. Point + 2 non-parallel vectors

If b and c non-parallel, and a is a point on the plane, then

r = a + λb + µc

where λ, µ are scalar parameters.

a
c

NB that these are

O b parallel to the plane, not
necessarily in the plane
A1 2015 19 / 1

Three ways of describing a plane. Number 2

r c
2. Three points b
Points a, b and c in the plane.
a
r = a + λ(b − a) + µ(c − a)

O
Vectors (b − a) and (c − a) are said to span the plane.
A1 2015 20 / 1

Three ways of describing a plane. Number 3

^
n

3. Unit normal Unit normal to the

r
plane is ^
n, and a point in the plane
is a a
r ·^
n = a ·^
n=D

O
Notice that |D| is the perpendicular distance to the plane from the origin.
A1 2015 21 / 1

The shortest distance from a point to a plane

The plane is r · ^
n = a ·^
n=D
Now, the shortest distance from point d to the plane ... ?
1 Must be along the ^
n
perpendicular d
2 n must be a point
d + λ^
on plane
r
⇒ (d + λ^
n) · ^
n=D
⇒ λ = D − d ·^ n a
O
⇒ dmin = |λ| = |D − d · ^
n|
A1 2015 22 / 1

Rotation, angular velocity and acceleration

A rotation can represented by a
vector whose
ω v
direction is along the axis of
rotation in the sense of a
right-handed screw,
magnitude is proportional to
r
the size of the rotation.
The same idea can be extended to the derivatives

angular velocity ω

angular acceleration ω̇.

The instantaneous velocity v(r) of any point P at r on a rigid body

undergoing pure rotation can be defined by a vector product

v = ω × r.
A1 2015 23 / 1

Vector Moments

Angular accelerations arise because

of moments.
The vector equation for the moment M
M of a force F about a point Q is

M=r×F Q r
α
where r is a vector from Q to any F
point on the line of action L of force
F.

The resulting angular acceleration ω̇ is in the same direction as the

moment vector M. (How are they related?)
A1 2015 24 / 1

Solution of vector equations

Find the most general vector x satisfying a given vector relationship.
Eg
x=x×a+b

General Method (assuming 3 dimensions)

1 Set up a system of three basis vectors using two non-parallel

vectors appearing in the original vector relationship. For example
a, b, (a × b)

2 Write
x = λa + µb + νa × b
where λ, µ, ν are scalars to be found.

3 Substitute expression for x into the vector relationship to determine

the set of constraints on λ,µ, and ν.
A1 2015 25 / 1

♣ Example 1: Solve x = (x × a) + b.
Step 1: Basis vectors a, b and vector product a × b.
Step 2: x = λa + µb + νa × b.
Step 3: Stick x back into the equation ...


λa + µb + νa × b = λa + µb + νa × b × a + b
= 0 + µ(b × a) + ν(a × b) × a + b

But (a × b) × a = a2 b − (a · b)a

λa + µb + νa × b = −ν(a · b)a + (νa2 + 1)b − µ(a × b)

Equating coefficients of a, b and a × b in the equation gives

λ = −ν(a · b) µ = νa2 + 1 ν = −µ

⇒ µ = 1/(1 + a2 ) ν = −1/(1 + a2 ) λ = (a · b)(1 + a2 ) .

So finally the solution is the single point:
1
x= [(a · b)a + b − (a × b)]
1 + a2
A1 2015 26 / 1

♣ Example 2: Solve x · a = K

This is in 2A1A, but we want to think around it ...

First note that there are not two fixed vectors in the expression ...
A:
Step 1 Use a, and introduce an arbitrary vector b, then find a × b
Step 2: x = λa + µb + νa × b.
Step 3: Bung x back into the equation!

... GRIND AWAY ...

and, recalling λ and ν are free parameters, we find

K − λa2
 
x = λa + b + νa × b
b·a
A1 2015 27 / 1

♣ Example #2: x · a = K

K − λa2
 
x = λa + b + νa × b
b·a

This is certainly correct ... but it looks very odd, given that the geometry
is very obvious in this case!

x must lie on the plane x · ^

a = K /a ...
... a plane with unit normal ^a and perpendicular distance |K /a| from the
origin.

So why does it look so complicated?

It is because b has been chosen arbitrarily and is one of the basis vectors.
A1 2015 28 / 1

♣ Example 2: x · a = K
As we can see upfront that this must be a plane,
here is a cunning plan ...

Choose b arbitrarily, but don’t use b as the second vector

Instead use it to find a second vector that is

perpendicular to BOTH a AND (a × b).

We can write down without further thought

K
x= a + µ(a × (a × b)) + ν(a × b) . µ, ν are free
a2

Can you see why?

A1 2015 29 / 1

A comment about solving vector identities

Suppose you are faced with

µa + λb = c

and you want to find µ.

What is the fast way of getting rid of b?

Use (b × b) = 0 ...

µ(a × b) = c×b
⇒µ(a × b) · (a × b) = (c × b) · (a × b)
(c × b) · (a × b)
⇒µ =
(a × b) · (a × b)
A1 2015 30 / 1

A comment about solving vector identities

µa + λb = c

An alternative is to construct two simultaneous equations

µa · b + λb 2 = c·b
2
µa + λa · b = a · c

and eliminate λ
(a · b)(b · c) − (a · c)b 2
µ=
(a · b)2 − a2 b 2
Compare with previous

(c × b) · (a × b)
µ=
(a × b) · (a × b)
A1 2015 31 / 1

Summary
We’ve discussed ...
• Vector products • Line & Plane geometry
• Angular velocity/moments • Solving vector equations
Key point from Lectures 1 and 2:
* Use vectors and their algebra “constructively” to solve problems.
(The elastic collision was a good example.)
* Don’t be afraid to produce solutions that involve vector operations.
Eg: µ = a · b/|c × a|. Working out detail could be left to a
computer program.
* Run with natural coordinate systems.
* If you are constantly breaking vectors into their components, you are
(probably) not using their power.
* Apply checks that equations are vector or scalar on both sides.
(Underline vectors.)