WMS

MA134
Geometry & Motion

Revision Guide

Written by Jack Betteridge

WMS
ii MA134 Geometry & Motion

Contents
1 Curves & Parameterisations 1

2 Derivatives 4

3 Area & Volume 5

4 Polar & Spherical Coordinates 6

5 Transformations & Generalised Coordinates 7

6 Line Integrals 8

7 Surface Integrals 10

8 Critical Points 13

Introduction
Geometry & Motion is an extension of differentiation and integration learned at A-level to higher di-
mensions. The focus of the course is on calculation and applying the concepts taught. As such there are
a great many definitions to commit to memory, but not many proofs. It is also very important to learn
to correctly parameterise curves, surfaces and solids. Not only does it mean you will end up with the
correct answer at the end of the question, but once you have a parameterisation more often than not the
rest of the question is just a case of carefully integrating or differentiating.
Dwight also records all of his lectures as screen casts, which are available through the undergraduate
handbook and moodle site. This should be the first place to look for course material as it is the most
relevant and up to date source for the current course, as well as further examples.
I would recommend trying to make your own summary notes from the lectured course, as this will
help you learn the material better than just looking through these, which are in fact a typeset copy of
my efforts in first year. But, these will serve as a quick reference for those who do not find summarising
notes a useful form of revision.
Note 0.1 (A note on notation). Throughout I have tried to be consistent with notation, but along the
way I may have slipped up. Where you see r 0 (t) this means the derivative of r(t) with respect to t,
elsewhere, and more usually this is denoted ṙ(t), but I’m sure you will know what I mean.
If you don’t like my notation feel free to pick your own, but be sure to explain in your work precisely
what you mean with your notation, especially if it is non-standard.

Disclaimer: Use at your own risk. No guarantee is made that this revision guide is accurate or
complete, or that it will improve your exam performance. Use of this guide will increase entropy,
contributing to the heat death of the universe. Contains no GM ingredients. Your mileage may vary.
All your base are belong to us.

Authors
This revision guide for MA134 Geometry & Motion has been designed as an aid to revision, not a
substitute for it. Written by Jack Betteridge.
Based upon lectures given by Prof. Dwight Barkley at the University of Warwick in 2011 and updated
for the 2015 course
Any corrections or improvements should be entered into our feedback form at http://tinyurl.com/WMSGuides
(alternatively email [email protected]).
MA134 Geometry & Motion 1

1 Curves & Parameterisations

r : I → R3 , I ⊂ R s.t. t 7→ r(t) = (x(t), y(t), z(t)) is a vector valued function of time

Definition 1.1. C ⊂ R is a curve (or path) if ∃ r : I → Rn continuous s.t. C = {r(t) : t ∈ I}, where I
is an interval.

Definition 1.2. The mapping t 7→ r(t), r : I → Rn is called a parameterisation of C if it consists of n

continuous functions of one variable t; x1 (t), . . . , xn (t), called the components of r (a parameterisation
is not unique!)

Example 1.3 (Simple parameterisation). Parameterise the triangle (0, 0), (1, 0), (0, 1) using the unit
interval.
Step 1: Draw a picture (I leave space here for you to practise this essential skill)

Step 2: Split up the interval into [0, 1/3], [1/3, 2/3], [2/3, 1] then
(0, 0) to (1, 0) is just (3t, 0)
(1, 0) to (0, 1) is (1 − 3(t − 1/3), 3(t − 1/3)) = (2 − 3t, 3t − 1)
(0, 1) to (0, 0) is (0, 1 − 3(t − 2/3)) = (0, 3 − 3t)
Step 3: Put it all together:



 (3t, 0) on [0, 1/3]

r(t) = (2 − 3t, 3t − 1) on [1/3, 2/3]


(0, 3 − 3t)

on [2/3, 1]

Example 1.4 (Parameterisation). Parameterise a helix starting at (a, 0, 0), ending at (−a, 0, 1) with
radius a and 1.5 turns anticlockwise around the z-axis.
Step 1: Draw a picture (I leave space here for you to practise this essential skill)
2 MA134 Geometry & Motion

Step 2: We know a circle with radius a has parameterisation

(a cos t, a sin t), t ∈ [0, 2π]

This is traversed anticlockwise. To go around 1.5 times t ∈ [0, 3π]. We also need the curve to climb to a
height of 1 using the same t in the z-component i.e:

t
3π

Step 3: Put it all together and we have the parameterisation:

r(t) = (a cos t, a sin t, t/3π)

Alternatively we can have the unit interval parameterisation:

r(t) = (a cos(3πt), a sin(3πt), t)

Definition 1.5. If C is parameterised by r(t) & t is actually time then velocity is the vector:

dr r(t + h) − r(t)
v(t) = (t)= lim
dt h→0 h

Component by component:  
dx1 dxn
v(t) = (t), . . . , (t)
dt dt
Speed is the magnitude of velocity, a scalar, i.e:

dr
speed = kv(t)k = (t)

dt

Acceleration is the rate of change of velocity (a vector):

dv
a(t) := (t)
dt

Definition 1.6. A closed curve or loop is where r : [a, b] → Rn & r(a) = r(b). To rule out intersections
r(t1 ) 6= r(t2 ) given t1 6= t2 unless t1 , t2 ∈ {a, b}.
dr
Definition 1.7. A curve is regular if there exists a parameterisation s.t. dt is defined and non-zero at
all points, so has no corners or cusps.

Definition 1.8. Let C be a curve parameterised by r : [a, b] → Rn the length of C :

Z b
dr
`(C) := (t) .dt
dt
a

Note 1.9. The length is independent of parameterisation.

Definition 1.10. Given a parameterisation of C we know r 0 (t) is tangent to C at r(t).

The tangent vector τ is given by:
r 0 (t)
τ (t) :=
kr 0 (t)k
MA134 Geometry & Motion 3

Principle normal n is:

τ 0 (t)
n(t) :=
kτ 0 (t)k

Curvature:
Define κ as curvature:
kτ 0 (t)k
κ(t) :=
kr 0 (t)k

Define ρ as radius of curvature:

1
ρ(t) :=
κ(t)

The binormal vector b is:

b(t) := τ (t) × n(t)

Note 1.11. κ, ρ are scalars & independent of parameterisation.

Note 1.12. τ , n, b form the Frenet basis.

Example 1.13 (Properties of curves). Find the speed, acceleration and length of the curve

C: r(t) = (a cos(3πt), a sin(3πt), t3 ) t ∈ [0, 1]

Furthermore find an expression for the Frenet basis and radius of curvature.
Step 1: Draw a picture (I leave space here for you to practise this essential skill)

Velocity:
v(t) = r 0 (t) = (−3πa sin(3πt), 3πa cos(3πt), 3t2 )

Speed:
p
kv(t)k = kr 0 (t)k = (−3πa sin(3πt))2 + (3πa cos(3πt))2 + (3t2 )2
q
= 9π 2 a2 sin2 (3πt) + 9π 2 a2 cos2 (3πt) + 9t4
p
= 3 π 2 a2 + t4

Acceleration:
a(t) = v 0 (t) = r 00 (t) = (−9π 2 a cos(3πt), −9π 2 a sin(3πt), 6t)
4 MA134 Geometry & Motion

Magnitude of acceleration:
p
ka(t)k = kv 0 (t)k = kr 00 (t)k = (−9π 2 a cos(3πt))2 + (−9π 2 a sin(3πt))2 + (6t)2
q
= 81π 4 a2 cos2 (3πt) + 81π 4 a2 sin2 (3πt) + 36t2
p
= 3 9π 4 a2 + 4t2

Hence:

r 0 (t) 1
τ (t) = = √ (−πa sin(3πt), πa cos(3πt), t2 )
kr 0 (t)k π 2 a2 + t4
τ 0 (t) 1
n(t) = = √ (−3π 2 a cos(3πt), −3π 2 a sin(3πt), 2t)
kτ 0 (t)k 9π 4 a2 + 4t2
1
b(t) = τ (t) × n(t) = √ √ (−πa sin(3πt), πa cos(3πt), t2 )
π a + t 9π 4 a2 + 4t2
2 2 4

×(−3π 2 a cos(3πt), −3π 2 a sin(3πt), 2t)

1
= √ √ (2tπa cos(3πt) + 3t2 πa sin(3πt),
π 2 a2 + t4 9π 4 a2 + 4t2
−3t2 πa cos(3πt) − 2tπasin(3πt), 3π 3 a2 cos(6πt))
√ r
1 kr 0 (t)k 3 π 2 a2 + t4 π 2 a2 + t4
ρ(t) = = = √ =
κ(t) kτ 0 (t)k 3 9π 4 a2 + 4t2 9π 4 a2 + 4t2

Finally:
Z b
`(C) = kr 0 (t)k.dt
a
Z 1 p
= 3 π 2 a2 + t4 .dt
0
Z arsinh(1/πa)
= 3π 2 a2 cosh2 (θ).dθ Using t = πa sinh(θ)
0
  r !
3 1 1
= πa πa arsinh + 1+
2 πa πa

2 Derivatives
Definition 2.1. Let f : Rn → R, then the partial derivative of f with respect to xk is:
 
∂f f (x1 , . . . , xk + h, . . . , xn ) − f (x1 , . . . , xk , . . . , xn )
(x1 , . . . , xn ) := lim
∂xk h→0 h

Definition 2.2. The directional derivative of f : Rn → R at a point x in a direction v, (x, v ∈ Rn ) is:

f (x + hv) − f (x)
(Dv f )(x) := lim
h→0 h

Definition 2.3.  
∂f ∂f
∇f (x) := (x), . . . , (x)
∂x1 ∂xn
A useful approximation is: f (x + h) ≈ f (x) + h · ∇f (x)
So (Dv f )(x) = v · ∇f (x) = kvkk∇f (x)k cos θ
MA134 Geometry & Motion 5

π
θ = 0 gives maximum gradient, θ = π gives minimum gradient, θ = π ± 2 gives the direction for level
curve.

Definition 2.4. Tangent plane:

π(x) : (r − x) · ∇f (x) = 0

Definition 2.5. The chain rule:

One dimension:
d
f (h(t)) = f 0 (h(t)) · h0 (t)
dt
Multivariable: If r : R → Rn , f : Rn → R, let g = f ◦ r, then for n = 3

d d dr ∂f dx ∂f dy ∂f dz
g(t) = f (r(t)) = ∇f (r(t)) · (t) = + +
dt dt dt ∂x dt ∂y dt ∂z dt

or more generally:
n
dg X ∂f dxi
=
dt i=1
∂xi dt

3 Area & Volume

Definition 3.1. Area under curve f :

n
X Z b
Area = lim f (xi )∆xi = f (x).dx
n→∞ a
i=1

Definition 3.2. Volume under surface f :

n
!
X ZZ Z d Z b
Volume = lim f (xi )∆Ai = f.dA = f (x, y).dx .dy
n→∞ Ω c a
i=1

Integrating f over a 2D area with limits:

Z bZ h(x) Z dZ η(y)
V = f (x, y).dy.dx V = f (x, y).dx.dy
a g(x) c ξ(y)
   
a ≤ x ≤ b, ξ(y) ≤ x ≤ η(y),
where: Ω = (x, y) : where: Ω = (x, y) :
g(x) ≤ y ≤ h(x) c≤y≤d

RR
Note 3.3. Area = Ω
.dA

We also have:
n
X ZZZ Z bZ dZ f
lim f (xi )∆Vi = f.dV = f (x, y, z).dz.dy.dx
n→∞ Ω a c e
i=1

Integrating f over a 3D volume with limits:

ZZZ Z bZ h(x)Z η(x,y)
f.dV = f (x, y, z).dz.dy.dx
Ω a g(x) ξ(x,y)
6 MA134 Geometry & Motion

 

 a ≤ x ≤ b, 

where: Ω = (x, y, z) : g(x) ≤ y ≤ h(x)
 
ξ(x, y) ≤ z ≤ η(x, y)
 
RRR
Note 3.4. Volume = Ω
.dV

Definition 3.5. Centre of Mass: For a body with density ρ(x, y, z), let
ZZZ
1
x̄ = xρ.dV
M Ω
ZZZ
1
ȳ = yρ.dV
M Ω
ZZZ
1
z̄ = zρ.dV
M Ω

where ZZZ
M= ρ.dV
Ω

Then the centre of mass is

r COM := (x̄, ȳ, z̄)

4 Polar & Spherical Coordinates

Definition 4.1. Polar Coordinates (r, θ):

∆A = ∆r∆θ r

so ZZ
V = f (r, θ)r.dr.dθ
θ r

Definition 4.2. Cylindrical Coordinates(r, θ, z):

∆V = ∆r∆θ∆z r

so Z Z Z
V = f (r, θ, z)r.dz.dr.dθ
θ r z

Definition 4.3. Spherical Coordinates (ρ, θ, φ):

x = ρ sin φ cos θ
y = ρ sin φ sin θ
z = ρ cos θ

so
∆V = ∆ρ∆φ∆θρ2 sin φ

so ZZZ Z Z Z
f.dV = f (ρ, θ, φ)ρ2 sin φ.dρ.dθ.dφ
Ω φ θ ρ
MA134 Geometry & Motion 7

Example 4.4 (Triple integrals and spherical coordinates). Find the mass of the following region

Ω = (x, y, z) ∈ R : x2 + y 2 + z 2 ≤ 5, x2 + y 2 ≤ z 2 , z ≥ 0


given that its density is given by

f (x, y, z) = x2 + y 2 + z 2 + 1

Step 1: Draw a picture (I leave space here for you to practise this essential skill)

Step 2: Notice from your picture that this is just a portion of a sphere and hence it may be best to
use spherical coordinates.
Step 3: Reparameterise the domain using spherical coordinates
n π o
Ω = (θ, ϕ, ρ) ∈ R : 0 ≤ θ ≤ 2π, 0 ≤ ϕ ≤ , 0 ≤ ρ ≤ 5
4

Step 4: Reparameterise the density function

f (θ, ϕ, ρ) = f (ρ) = ρ2 + 1

Step 5: Integrate
ZZZ Z 2π Z π/4 Z 5
f.dV = (ρ2 + 1)ρ2 sin(ϕ).dρ.dϕ.dθ
Ω 0 0 0
Notice we can split the integral
Z 2π Z π/4 Z 5
= .dθ sin(ϕ).dϕ ρ4 + ρ2 .dρ
0 0 0
iπ/4  ρ5 3 5

h ρ
= 2π − cos ϕ +
0 5 3 0
√

2000 2 − 2 π
=
3

5 Transformations & Generalised Coordinates

2D Case: ψ : R2 → R2 , ϕ : Γ → Ω, (u, v) 7→ (x(u, v), y(u, v)) now

 !
 ∂x ∂x 
∂u ∂v
∆A = det  ∆u∆v
 
 ∂y ∂y 
∂u ∂v
8 MA134 Geometry & Motion

so:  !
ZZ Z Z  ∂x ∂x 
∂u ∂v
f.dA = f (ψ(u, v)) det  .dv.du
 
∂y ∂y
Ω
| u{z v}
 ∂u ∂v

Γ

Denote !
∂x ∂x
∂u ∂v ∂(x, y)
∂y ∂y
as
∂u ∂v
∂(u, v)

which is called the Jacobian Matrix

3D Case: ψ : R3 → R3 , ϕ : Γ → Ω, (u, v, w) 7→ (x(u, v, w), y(u, v, w), z(u, v, w)) now
  
 ∂(x, y, z) 
∆V = det
 ∆u∆v∆w
∂(u, v, w) 

so: ZZZ Z Z Z   
 ∂(x, y, z) 
f.dV = f (ψ(u, v, w)) det .dw.dv.du
Ω u v w ∂(u, v, w) 
| {z }
Γ

6 Line Integrals
Definition 6.1. The arclength parameterisation (or natural parameterisation) is a parameterisation
r(t) of a curve C s.t. ∀s ∈ [0, T ] where `(C) = T we have
Z s
d d
kr(t)k.dt = s=1
ds 0 ds

=⇒ kr 0 (s)k = 1

Note 6.2. When integrating, we often denote integrating with respect to an arc length parameterisation
by ds = kr̃ 0 (t)k.dt, where r̃ : [a, b] → Rn is any parameterisation of a curve C and r : [0, T ] → Rn is the
arclength parameterisation of the curve C i.e:
Z Z T Z b
f.ds = f (r(s)).ds = f (r(t))kr 0 (t)k.dt
C 0 a

Example 6.3. Recall for a curve C with parameterisation r : [a, b] → Rn :

Z b
`(C) = kr 0 (t)k.dt
a

So now we are free to denote Z

`(C) = .ds
C

where
ds = kr 0 (t)k.dt

denotes integrating with respect to the arc length parameterisation.

Definition 6.4. Let f : Rn → R & C ⊂ Rn be a curve with a parameterisation r : [a, b] → Rn , the

MA134 Geometry & Motion 9

line integral of f over C is

Z Z b
f.ds = f (r(t))kr 0 (t)k.dt
C a

Example 6.5 (Scalar line integral). Evaluate the line integral for a function

f (x, y) = x2 + y 2

along the path

C: r(t) = (t, t) t ∈ [0, 1]

Step 1: Calculate r 0 (t) and hence kr 0 (t)k

r 0 (t) = (1, 1)
√
kr 0 (t)k = 2

Step 2: Use this to calculate the line integral

Z Z 1
f.ds = f (r(t))kr 0 (t)k.dt
C 0
Z 1 √
= f (t, t) 2.dt
0
√ Z 1
= 2 2t2 .dt
0
√ t3 1
 
= 2 2
3 0
√
2 2
=
3

Definition 6.6. Let v : Rn → Rm & C ⊂ Rn be a curve with a parameterisation r : [a, b] → Rn . The

line integral of the vector field v over C is
Z Z b
v.d` = v(r(t)) · r 0 (t).dt
C a

The line integral differs only by the direction r traverses C i.e: by a factor of −1.

Definition 6.7. Gradient Fields are vector fields with the property that v = ∇f for some f : Rn → R

Theorem 6.8. Fundamental Theorem for Line Integrals of Vector Fields: Let C be a curve in Rn , r :
[a, b] → Rn with endpoints a = r(a), b = r(b) & let v be a gradient field i.e: v = ∇f . Then
Z
v.d` = f (b) − f (a)
C

i.e: The line integral of a gradient field depends only upon the end points and not the path taken.

Corly 6.9. I
v.d` = 0 ⇐⇒ v is a gradient field.
C
H
Where denotes integrating over a closed curve.
10 MA134 Geometry & Motion

Example 6.10 (Vector line integral). Evaluate the line integral for a vector field

v(x, y) = (2x, 2y)

along the path

C: r(t) = (cos t, sin t) t ∈ [0, 2π]

Step 1: Calculate r 0 (t), remember we don’t have to calculate kr 0 (t)k here

r 0 (t) = (− sin t, cos t)

Step 2: Use this to calculate the line integral

Z Z 2π
v.d` = f (v(t)) · r 0 (t).dt
C 0
Z 2π
= v(cos t, sin t) · (− sin t, cos t).dt
0
Z 2π
= (2 cos t, 2 sin t) · (− sin t, cos t).dt
0
Z 2π
= −2 cos t sin t + 2 sin t cos t.dt
0
= 0

Note 6.11. v = ∇f where f (x, y) = x2 + y 2 and C is a closed curve, so we could have arrived at this
result far quicker if we had used the fundamental theorem of line integrals. This is very useful to notice
in an exam!

7 Surface Integrals
Note 7.1. For curves we have: r : I → Rn , I = [a, b]
For surfaces: r : Ω → Rn Ω = [a, b] × [c, d]

Definition 7.2. The tangent plane to a surface r at the point (u0 , v0 ) is given by: p : R2 → R3 ,
(h, k) 7→ R3 , where
∂r ∂r
p(h, k) = r(u0 , v0 ) + h (u0 , v0 ) + k (u0 , v0 )
∂u ∂v
Definition 7.3. The normal vector to the tangent plane is given by

∂r ∂r n
n= × & n̂ =
∂u ∂v knk

Definition 7.4. For f : R3 → R, the surface integral for a surface S, parameterised by r, is

ZZ ZZ
∂r ∂r
f.dS = f (r(u, v)) ×
∂u ∂v .du.dv

S Ω

Example 7.5 (Surface integral). Find the surface integral of

f (x, y, z) = z
MA134 Geometry & Motion 11

over the upper unit hemisphere.

Step 1: Draw a picture (I leave space here for you to practise this essential skill)

Step 2: Notice from your picture that this is just a portion of a sphere and hence it may be best to
parameterise using trigonometric functions.

Step 3: Parameterise the surface

S: r(θ, ϕ) = (cos θ sin ϕ, sin θ sin ϕ, cos ϕ) θ ∈ [0, 2π], ϕ ∈ [0, π/2]

Step 4: Evaluate the partial derivatives with respect to each of your parameters

∂r
= (− sin θ sin ϕ, cos θ sin ϕ, 0)
∂θ

∂r
= (cos θ, cos ϕ, sin θ cos ϕ, − sin ϕ)
∂ϕ
12 MA134 Geometry & Motion

Step 5: Calculate ∂r
∂θ ×
∂r

∂ϕ

∂r ∂r
∂θ × ∂ϕ = k(− sin θ sin ϕ, cos θ sin ϕ, 0) × (cos θ, cos ϕ, sin θ cos ϕ, − sin ϕ)k

= (− cos θ sin2 ϕ, − sin θ sin2 ϕ, − sin2 θ sin ϕ cos ϕ − cos2 θ sin ϕ cos ϕ)

= (− cos θ sin2 ϕ, − sin θ sin2 ϕ, − sin ϕ cos ϕ)

q
= (− cos θ sin2 ϕ)2 + (− sin θ sin2 ϕ)2 + (− sin ϕ cos ϕ)2
q
= cos2 θ sin4 ϕ + sin2 θ sin4 ϕ + sin2 ϕ cos2 ϕ
q
= sin4 ϕ + sin2 ϕ cos2 ϕ
= sin ϕ since sin is positive for ϕ ∈ [0, π/2]

Step 6: Use this to find the surface integral

ZZ Z 2π Z π/2
∂r

∂r
f.dS = f (r(θ, ϕ))
× .dϕ.dθ
S 0 0 ∂θ ∂ϕ
Z 2π Z π/2
= f (cos θ sin ϕ, sin θ sin ϕ, cos ϕ) sin ϕ.dϕ.dθ
0 0
Z 2π Z π/2
= cos ϕ sin ϕ.dϕ.dθ
0 0
Notice we can split the integral
1 2π
Z Z π/2
= .dθ sin 2ϕ.dϕ
2 0 0
 π/2
1 1
= 2π − cos 2ϕ
2 2 0
= π

Note 7.6. Area of S is given by

ZZ ZZ
∂r ∂r
.dS = × .du.dv
Ω ∂u ∂v

S

Definition 7.7. Let v : R3 → R3 & S ⊂ R3 be a curve with a parameterisation r : Ω → R3 . The

flux integral is given by: ZZ
v · n̂.dS
S

where
∂r ∂r
× ∂r ∂r
n̂ = ∂u ∂v
, dS =
× .du.dv
∂r × ∂r ∂u ∂v
∂u ∂v

hence ZZ ZZ  
∂r ∂r
v · n̂.dS = v· × .du.dv
S Ω ∂u ∂v
Example 7.8 (Flux integral). Find the flux of the vector field

v(x, y) = (−4y, x, 0)

through the upper unit hemisphere.

MA134 Geometry & Motion 13

Step 1: Notice that steps 1–5 will be identical to the previous example, but we don’t need to find the
magnitude of the cross product. Again, it will save time in an exam if you spot things like this.
Step 2 (or 6): Use previous work to calculate the flux integral
ZZ Z 2π Z π/2  
∂r ∂r
v · n̂.dS = v(r(θ, ϕ)) · × .dϕ.dθ
S 0 0 ∂θ ∂ϕ
Z 2π Z π/2
= v(cos θ sin ϕ, sin θ sin ϕ, cos ϕ)
0 0
·(− cos θ sin2 ϕ, − sin θ sin2 ϕ, − sin ϕ cos ϕ).dϕ.dθ
Z 2π Z π/2
= (−4 sin θ sin2 ϕ, cos θ sin2 ϕ, 0)
0 0
·(− cos θ sin2 ϕ, − sin θ sin2 ϕ, − sin ϕ cos ϕ).dϕ.dθ
Z 2π Z π/2
= 4 sin θ cos θ sin3 ϕ − sin θ cos θ sin3 ϕ.dϕ.dθ
0 0
Notice we can split the integral
Z 2π Z π/2
= 3 sin θ cos θ.dθ sin3 ϕ.dϕ
0 0
3 2π
Z Z π/2
= sin 2θ.dθ sin ϕ − cos2 ϕ sin ϕ.dϕ
2 0 0
3h i2π h 1 iπ/2
= cos 2θ cos ϕ + cos3 ϕ
2 | {z 0 } 3 0
=0
= 0

8 Critical Points
When ∇f = 0, let

∂2f ∂2f ∂2f

A= (x0 ), B= (x0 ), C= (x0 ) & D = AC − B 2
∂x2 ∂x∂y ∂y 2

f is a maximum at x0 if D > 0 & A < 0

f is a minimum at x0 if D > 0 & A > 0
f is a saddle point at x0 if D < 0
If D = 0 at x0 no conclusion can be made about f (x0 )

Here is a duck: