Publications of the Astronomical Society of Australia (2022), 1–12

doi:%

REVIEW

A Pedagogical Review of the Vacuum Retarded Dipole Model of Pulsar

Spin Down
J. C. Satherley and C. Gordon
School of Physical and Chemical Sciences, University of Canterbury, Christchurch, New Zealand
Author for correspondence: J. C. Satherley, Email: jsa113@uclive.ac.nz.
arXiv:2103.10617v5 [astro-ph.HE] 18 Aug 2022

(Received dd Mmm YYYY; revised dd Mmm YYYY; accepted dd Mmm YYYY; first published online 19 March 2021)

Abstract
Pulsars are rapidly spinning highly magnetised neutron stars. Their spin period is observed to decrease with time. An early analytical model for
this process was the vacuum retarded dipole (VRD) by Deutsch (1955). This model assumes an idealised star and it finds that the rotational
energy is radiated away by the electromagnetic fields. This model has been superseded by more realistic numerical simulations that account for
the non-vacuum like surroundings of the neutron star. However, the VRD still provides a reasonable approximation and is a useful limiting case
that can provide some qualitative understanding. We provide detailed derivations of the spin down and related electromagnetic field equations
of the VRD solution. We also correct typographical errors in the general field equations and boundary conditions used by Deutsch (1955).
Keywords: stars: pulsars – stars: magnetic field – stars: neutron

1. INTRODUCTION for standard electrodynamics theory.

Pulsars are highly magnetised rapidly rotating neutron stars In Sec. 2, we begin by outlying the definition of an ide-
that emit electromagnetic radiation almost across the entire alised star, along with fundamental equations that apply due
spectrum, from radio to γ -rays (Travelle, 2011). Because of to this assumption, which include Maxwell’s equations. We
the extreme magnetic and electric fields, the environment also note down the Alfvén theorem and a brief description of
around a pulsar proves to be interesting and is believed to be a vector field. In Sec. 3, we define the magnetic field symmetry
the main cause of a pulsar’s energy emission. This energy axis and its associated coordinate system. The magnetic field
emission results in a decreasing spin-rate for the neutron star. is assumed to take the form of a dipole about this axis. The
An analytical approximation for this process is given by the transformation between the two coordinate systems is shown
Vacuum Retarded Dipole (VRD) model which was originally and used to derive the electromagnetic field boundary condi-
derived by Deutsch (1955) (hereafter referred to as D55). tions in the rotation symmetry axis coordinates. In Sec. 4, we
show the derivations for the time-dependent and static compo-
The derivations of the results in D55 are very concise, with
nents of the electromagnetic fields using series expansions and
many details missing. The purpose of this review is to rederive
matching coefficients to the boundary conditions. In Sec. 5,
the main results of D55 and in doing so provide details on
we simplify the form of the electromagnetic fields for a large
how this derivation takes place mathematically and physically.
radius, then the power and spin down equations are derived.
We also uncover three typographical errors in D55, the most
important of which is in our Eqn. (31), the polar component
of the electric field. It is worth noting the first papers we found
that mentioned the error in the general field equations are
Melatos (1997) and Michel & Li (1999). However, they do
2. The Internal Fields of An Idealised Star
not show the derivations for the general field equations and
how the typographical error is found. A reasonably detailed
For the VRD solution, we consider the electric and magnetic
derivation of the VRD solution, without the typos, is shown
fields of an idealised star. An idealised star is a star that is a
in the textbook by Michel (1991). However, the method used
sharply bounded sphere, is isotropic, is perfectly conducting,
by Michel is different from D55. While D55 uses methods
and has permittivity and permeability of a vacuum everywhere
outlined by Stratton (1941), Michel just considers multipole
inside. The boundary is not continuous but rather an immedi-
expansions at an inclination and does not derive the normalisa-
ate change between the star’s medium and the surrounding
tions. Also the following working carried out by us goes into
vacuum. Which, in the case of a real star is not true. However,
greater detail for the derivations. In this review we use meth-
it greatly simplifies the situation.
ods outlined in Stratton (1941) (hereafter referred to as S41).
In particular, we refer frequently to Chapters VII and VIII,
which consider spherical waves and radiation respectively. We
use Griffiths (2017) (hereafter referred to as G17) as a reference
2 J. C. Satherley et al.

2.1 Maxwell’s Equation Fig. 1a. Hence, (5) in the notation used for the rest of the
The Maxwell equations in SI units are, report becomes,

ρ H (r , θ, ϕ, t) = Hr (r , θ, λ) r̂ + Hθ (r , θ, λ) θ̂ + Hϕ (r , θ, λ) ϕ̂.
∇·E= , (1)
0
∇ · H = 0, (2)
∂H z, ω
∇ × E = –µ0 , (3)
∂t
∂E
∇ × H = J + 0 , (4)
∂t
r̂
where E is the electric field, H is the magnetic field (related ϕ̂
to B by B = µ0 H), 0 is the permittivity of free space, and
µ0 the permeability of free space. H will typically be referred θ̂
to as the H field. In D55, there is a missing minus sign in his
equivalent of equation (3). This is clearly a typo as he takes
into account the minus sign in later parts of his article.

2.2 The H field

Firstly, D55 describes the typical spherical polar coordinates
around the star’s axis of rotation, (r , θ, ϕ), with rotation axis
(a)
ω aligned with the z-axis and origin at the centre of the star. z, ω
Where r is the radius, θ is the angle from the pole and ϕ is the
azimuth angle measured counterclockwise from the x-axis.
Let H (r , θ, ϕ, t) be an arbitrary solenoidal function within
the star. A solenoidal function is one for which the divergence e
is zero at all points in the field (∇ · H = 0). D55 states that a r̂
necessary and sufficient condition that H be “frozen into” the ν̂
star is that ψ̂

H (r , θ, ϕ, t) = r0 Hr (r , θ, λ) + θ0 Hθ (r , θ, λ) + ϕ0 Hϕ (r , θ, λ), (5)

where λ is an azimuthal coordinate measured from a meridian

fixed in the star, and r0 , θ0 and φ0 are the unit vectors in
the respective coordinate directions. That is the time and
azimuthal angle ϕ can be replaced by λ as the H field and
the meridian move with the star. The variable λ is defined
(b)
more clearly in eqn. (6) and Fig. 2b. Hr , Hθ and Hϕ give the
magnitude of the field in each basis vector’s direction. The Figure 1. (a) The unit vectors of a vector field at a point on a sphere in
concept of the magnetic field being “frozen into” a medium (r , θ, ϕ). (b) The unit vectors of a vector field at a point on a sphere in
comes from Alfvén’s Theorem (Roberts, 2007). (r , ψ, ν).

2.2.1 Vector Fields

As H is a vector field, at a particular point (r , θ, ϕ) in space, the
field can be represented as a vector with some magnitude in
each unit vector’s direction. In D55, the directions are given
by the subscript 0 terms, and the magnitudes are given by the
functions proceeding the unit vector terms. In polar coor-
dinates, r0 is the radial component, pointing away from the
centre of the sphere. The polar component is θ0 , its direc-
tion is tangent to the polar angle. Lastly, ϕ0 is the azimuthal
component which has a direction tangent to the azimuthal
angle. This notation continues throughout D55. However,
we will here on forth elect to use r̂, θ̂ and ϕ̂ to represent
these components of the vector field. A visual representation
of the unit vectors of a field in polar coordinates is shown in
3. The External Field in Vacuum
It is convenient to define a new set of polar coordinates (r , ψ, ν)
with axis e such that the magnetic field is symmetrical about e
(known as the magnetic field symmetry axis). The axis e will be
in fixed rotation with the star due to Alfvén’s theorem. This
new coordinate system is shown in Fig. 2a. The magnetic field
symmetry here refers to, for a fixed r and ψ , the magnitude
of the field does not change as ν is changed and the direction
does not change relative to the local unit vectors. The angle
between the rotation axis ω and e is χ. This relationship is
shown in Fig. 2b. The axis e has polar angle χ and azimuthal
angle ω t in (r , θ, ϕ), e = (r , χ, ω t), at some time t during the
rotation of the star. The azimuthal angle λ mentioned earlier
Publications of the Astronomical Society of Australia 3

is measured from e along ϕ, so that ditions for the fields in both coordinate systems. It is conve-
nient to figure out the transformation now, then approach
λ = ϕ – ω t. (6) the boundary conditions. Define the transformation matrix
R : ( r̂, θ̂, ϕ̂) → ( r̂, ψ̂, ν̂) which rotates between the two
z unit vector systems at a point in space. Figs. 1a and 1b show
the two sets of basis vectors at the same point. It is clear that
r̂ points radially out, normal to the sphere’s surface, in both
e
figures. Hence, at the same point in space r̂ is the same in
ν both coordinate systems. However, the angular components
undergo a rotation as θ̂, ϕ̂, ψ̂ and ν̂ all lie in a plane tangent
ψ to the sphere. The rotation between these unit vectors will
be a two dimensional rotation about r̂. Let the angle of this
x rotation be ξ , the transformation matrix R is then
 
y 1 0 0
R = 0 cos ξ – sin ξ  . (8)
0 sin ξ cos ξ
This matrix transformation takes a basis of vectors at some
point in space from ( r̂, θ̂, ϕ̂) and rotates them anticlockwise
(a) by an angle ξ with r̂ as axis to the unit vectors ( r̂, ψ̂, ν̂). To
z ω
reverse this transformation the angle of rotation is replaced by
–ξ .
e The angle ξ can be found in terms of already defined co-
ordinate parameters and values. This is done by considering
χ
triangles on the surface of a sphere. The spherical triangle for
this situation can be found in Fig. 4b in Appendix 6. Then
from solutions of spherical triangles (50) and (51) the following
ωt λ relationships can be found,
x
y sin χ sin λ
sin ξ = ,
sin ψ
(9)
cos χ – cos ψ cos θ
cos ξ = .
sin ψ sin θ
An interesting relationship that is obtained from the standard
(b) spherical triangle relation (49) is
Figure 2. The relation between the two coordinate systems (r , θ, ϕ) and cos ψ = cos χ cos θ + sin χ sin θ cos λ, (10)
(r , ψ, ν). x, y, and z are the Cartesian coordinates. (a) The coordinate sys-
tem (r , ψ, ν) portrayed on a sphere. (b) The star’s rotation axis ω and the which will be required momentarily.
magnetic field’s symmetry axis e. χ is the angle of inclination between ω
and e, and λ is measured from the projection of e onto the xy-plane.
3.2 Boundary Conditions of the Fields
As the surface of the star is a rigid boundary between the
We shall now consider a magnetic field which within the
interior of the star and free space, the necessary components of
star has the form,
the external electromagnetic fields must satisfy the boundary
H = R1 (r )S1 (ψ) r̂ + R2 (r )S2 (ψ) ψ̂ + R3 (r )S3 (ψ) ν̂, (7) conditions at r = a (see G17). These boundary conditions
are set by the internal electromagnetic fields. From equations
where Ri and Si are arbitrary functions of r and ψ respectively, (7.61) and (7.62) of G17, the magnetic fields orthogonal to
with i = 1, 2, or 3. Note that H has no dependence on ν the surface must be equal and the electric fields tangent to the
for the magnitude of the vectors and hence the magnitude is surface must be equal. Hence, using (7), the first condition at
symmetrical about axis e as stated earlier. In this situation D55 the surface of the star that must be satisfied is,
assumes the interior magnetic field to be a dipole about e by
Hrext = Hrint = R1 (a)S1 (ψ).
taking S1 (ψ) = cos ψ .
Next we look at the internal electric field. To do this we
3.1 Matrix Transformation between Vector Field Basis Vec- use Ohm’s law from G17, Eqn. 7.2, that relates the electric
tors field and current density, J and is given by,
A vector transformation between the two coordinate systems J
mentioned is required whilst working out the boundary con- E + µ0 V × H = (11)
σ
4 J. C. Satherley et al.
where σ is the conductivity and V is the local velocity of
rotation of the star. Assuming σ → ∞ gives
E = –µ0 V × H. (12)
The velocity in ( r̂, θ̂, ϕ̂) is
V = ω × r = r ω sin θ ϕ̂. (13)
However, the vector field (7) must be converted from
( r̂, ψ̂, ν̂) to ( r̂, θ̂, ϕ̂) so that V and H are defined on the
same basis. This is done by rotating (7) using (8). However,
as the reverse transformation is needed, the signs of R2,3 and
R3,2 elements are switched.
R(–ξ)H =R1 (r )S1 (ψ) r̂
+ [R2 (r )S2 (ψ) cos ξ + R3 (r )S3 (ψ) sin ξ ] θ̂
+ [–R2 (r )S2 (ψ) sin ξ + R3 (r )S3 (ψ) cos ξ ] ϕ̂.
Applying the previous result to (12) with (13) and using Fig. 4b
of Appendix 6 we can find the E field inside the star,
E = – µ0 V × (R(–ξ)H )
=µ0 r ω sin θ (R2 (r )S2 (ψ) cos ξ + R3 (r )S3 (ψ) sin ξ ) r̂
– ωµ0 rR1 (r )S1 (ψ) sin θ θ̂
+ 0 ϕ̂,
and evaluating this at r = a gives the tangential components as
Eθ = –ωµ0 aR1 (a)S1 (ψ) sin θ,
Eϕ = 0.
All but the third boundary condition match what D55
obtained. There is a typographical error in D55’s boundary
conditions (Deutsch’s Eqn. (10)), the third condition should
be written as Eext int ext int
ϕ = Eϕ = 0 rather than Er = Eϕ = 0.
A more useful form of the boundary conditions can be
found. In particular, the field equations are solved for when
S1 = cos ψ . So by recalling (10), the boundary conditions
then results in,
Hrext = Hrint = R1 (a)(cos χ cos θ + sin χ sin θ cos λ),
Eext int
θ = Eθ = –ωµ0 aR1 (a)(cos χ cos θ + sin χ sin θ cos λ) sin θ
1 following section on the wave solution. Then use a multipole
= – ωµ0 aR1 (a)(sin 2θ cos χ + (1 – cos 2θ) sin χ cos λ),
2 expansion to provide the static component.
Eext int
ϕ = Eϕ = 0.
(14)
From the boundary conditions in the (r , θ, φ) coordinate sys-
tem, there is a static (time-independent) component and a
time-dependent component for the fields. The terms in (14)
which relate to the time-dependent component have a depen-
dency on λ, which depends on time from (6).
4. Derivation of General Field Equations
The general field equations for the VRD solution external
to the star are provided in the Appendix of D55. However,
with minimal explanation given in D55 about the derivation
of the field equations, we wish to show how they are found.
We use results found in Chapters VII and VIII of S41 for this
section. Chapter VII proves beneficial in supplying general
solutions to the vector wave equation which, must be satisfied
by the time evolving electromagnetic fields around the neutron
star. We apply these solutions to the neutron star situation by
firstly considering the propagation factor of the waves outside
the star. In S41, on page 392, the propagation factor (complex
wavenumber) is given as,
k2 = µω 2 + iσµω, (15)
and as we are now considering the fields external of the star, in
√
a vacuum,  = 0 , µ = µ0 , σ = 0 and c = 1/ µ0 0 . Hence,
k = ω /c. (16)
Then for convenience define,
ρ = kr = (ω /c)r ,
(17)
α = ka = (ω /c)a,
where ω is the rotational velocity of the star, c the speed of
light, and a is the radius of the star. Here ρ and α are unitless
numbers. A trivial relationship that will be required is
∂ ω ∂
= .
∂ρ c ∂r
Next consider the EM fields surrounding the star in two
parts, a static time-independent component and a time-dependent
component, where the complete field will be the linear combi-
nation of the two components (G17). The time-independent
component will be axially symmetric about the axis ω and
hence have no ϕ dependence (m = 0, m will be defined clearer
shortly). Whilst the time-dependent component will be non-
axially symmetric and hence have a dependence on ϕ (m 6= 0).
This can be expressed as
H = H static (r , θ) + H time-dependent (r , θ, λ),
E = Estatic (r , θ) + Etime-dependent (r , θ, λ).
We first consider the time-dependent component in the
4.1 Wave Solution
Due to the rotating nature of the magnetic field symmetry
axis around the star’s axis of rotation, the vector fields E and
H satisfy the same vector differential wave equation from S41,
Sec. 7.1, Eqn. (1), which is given as
∂2C ∂C
∇2 C – µ – µσ = 0, (18)
∂ t2 ∂t
where C represents either vector E or H. We can express E
and H as a series of solutions Mn and Nn which satisfy the
Publications of the Astronomical Society of Australia 5

vector differential wave equation (18) for C. From S41, Sec. eration (22) essentially selects cos(mϕ) or sin(mϕ) depending
7.1, Eqn. (12) this representation is, on the complex-valued nature of f (x).
X We wish to solve for the field equations in polar coordi-
E=– (an Mn + bn Nn ) , nates around rotation axis ω . Due to the dependence of ϕ
n in our electromagnetic boundary conditions (14) only being
(19)
k X cos(ϕ) and sin(ϕ), we are limited to m = 1 for our equations
H=– (an Nn + bn Mn )
iωµ0 n (20) and (21). Otherwise terms which contain cos(mϕ) and
sin(mϕ), where m 6= 1, will appear in the solutions of (19)
where Mn and Nn have more recently been referred to as and cause disagreement of the external and internal electro-
the Magnetic and Electric harmonics respectfully. In par- magnetic fields at the surface of the star. By also comparing
ticular, the functions are expressed as Mn = ∇ × aψ and the series expression (19) and the boundary conditions (14)
Nn = (1/k)∇ × Mn , where a is a constant vector and ψ is a we see that if a trigonometric function of θ does not appear
scalar function satisfying ∇2 ψ + k2 ψ = 0, which is known in the boundary conditions, its weighting coefficient will be
as the Helmholtz equation. The derivation for the general zero (as will be noticed in the derivations shortly). Hence, it is
solutions of Mn and Nn is provided in S41 and the results apparent from inspection of the boundary conditions and the
are quoted below. These solutions will also contain a time Associated Legendre Polynomials that only n = 1 and n = 2
dependence. Due to the linearity of the wave equation, the are required. So evaluating (20) and (21) for m = 1, n = 1 and
time dependency can be split off such that M = me–iωt and n = 2 and including the new notation from (22) gives,
N = ne–iωt without any loss of generality. Then from S41,
Sec. 7.11, Eqns. (11) and (12), and rewriting them in a slightly m1,1 =ih1 (ρ)eiϕ θ̂ – h1 (ρ) cos θeiϕ ϕ̂,
more useful form with ρ instead of kr,
m1,2 =3ih2 (ρ) cos θeiϕ θ̂ – 3h2 (ρ) cos 2θeiϕ ϕ̂,
m (1) sin h1 (ρ) 1
memn = ∓ hn (ρ)Pnm (cos θ) (mϕ) θ̂ n1,1 =2 sin θeiϕ r̂ + (ρh01 (ρ) + h1 (ρ)) cos θeiϕ θ̂
o sin θ cos (20) ρ ρ
(1) ∂ Pm cos
– hn (ρ) n (mϕ) ϕ̂, i
+ (ρh01 (ρ) + h1 (ρ))eiϕ ϕ̂,
∂θ sin ρ
n(n + 1) (1) cos h2 (ρ) 3
nemn = hn (ρ)Pnm (cos θ) (mϕ) r̂ n1,2 =9 sin 2θeiϕ r̂ + (ρh02 (ρ) + h2 (ρ)) cos 2θeiϕ θ̂
o ρ sin ρ ρ
1 ∂ h (1) i ∂ m cos 3i
+ ρ hn (ρ) P (cos θ) (mϕ) θ̂ + (ρh02 (ρ) + h2 (ρ)) cos θeiϕ ϕ̂,
ρ ∂ρ ∂θ n sin ρ
m ∂ h (1) i m sin (23)
∓ ρ hn (ρ) Pn (cos θ) (mϕ) ϕ̂, where we have neglected the operator that extracts the real
ρ sin θ ∂ρ cos
(21) part of the result. This will be added during the calculation
(1) for the coefficients in (19).
where hn is the spherical Bessel function of the third kind
(refer to Appendix 3), Pnm (cos θ) are the associated Legendre
polynomials (refer to Appendix 4) , and m and n are associated 4.2 Multipole Expansion
to a particular solution to the Helmholtz equation. In particular, A static field symmetric about the rotation axis can be expressed
m defines the azimuthal dependence of the E and H fields. using a multipole expansion. Upon inspection of the static
Whilst n controls the polar dependence. These functions (20) terms in the boundary conditions (14) and the multipole ex-
and (21) are now known as vector spherical harmonics. The pansions given in G17 it can be seen, with some working, that
subscript e or o define the even and odd solution, this chooses only a magnetic dipole and electric quadrupole are present in
the sign of some terms and whether the trigonometric function this situation based on the conditions that multipole terms are
(1) linearly independent of each other and that the boundary con-
for ϕ is sin or cos. From here on hn will be referred to
as hn . However, D55 does not use the even-odd notation ditions will uniquely determine the fields (electromagnetism
from S41 but rather D55 multiplies the function by a complex uniqueness theorem, see G17).
exponential then takes the real part of the result to give the
wave solutions. This can be expressed as an operator that first 4.2.1 Magnetic Dipole
multiplies a function f (x) by eimϕ , then takes the real part of
We can begin by taking a magnetic dipole and check whether
the result,
it satisfies the H field boundary condition from (14). From
cos equation (5.88) of G17, a magnetic dipole is given as,
f (x ) (mϕ) = <(f (x)eimϕ ),
sin
(22) m  
sin (24)
∓f ( x ) (mϕ) = –<(f (x)ieimϕ ), H dip (r) =
4π r 3
2 cos θ r̂ + sin θ θ̂ ,
cos
where f (x) may be complex-valued function and < denotes the where m is the magnetic dipole moment of the field. Now
operator which returns the real part of an expression. The op- compare the radial component of (24) to the static term of the
6 J. C. Satherley et al.

radial H field boundary condition from (14). terms in a similar form as D55,
m 1 a4
2 cos θ = R1 (a) cos χ cos θ. Equad, r = – ωµ0 a 4 R1 (a) cos χ (3 cos 2θ + 1) ,
4π a3 4 r
We see that the function depending on θ in the boundary 1 a4 (26)
Equad, θ = – ωµ0 aR1 (a) 4 cos χ sin 2θ,
condition is satisfied fully by the dipole term. Hence, a mag- 2 r
netic dipole is the only static term present due to the boundary Equad, ϕ = 0.
conditions. This implies the ratio m/(4π) will satisfy,

m 1 4.3 Derivation of General Field Equations

= a3 R1 (a) cos χ.
4π 2 We are now in a position to derive the equations in the Ap-
pendix of D55. The general field equations will be the sum of
Therefore, the static component of the magnetic field will be
the multipoles derivations (25) and (26), and the series expres-
sions (19),
1 a3  
H static (r) = 3
R1 (a) cos χ 2 cos θ r̂ + sin θ θ̂ . (25)
2r 1
H = H dip – (a1 n1,1 + a2 n1,2 + b1 m1,1 + b2 m1,2 )e–iωt ,
icµ0 (27)
4.2.2 Electric Quadrupole E = Equad – (a1 m1,1 + a2 m1,2 + b1 n1,1 + b2 n1,2 )e–iωt ,
We begin by assuming the solution for the static electric field
to be a quadrupole, after which we show that it is the only com- where a1 , a2 , b1 and b2 are constants to be determined by the
ponent required to satisfy the boundary conditions due to the boundary conditions (14). These constants may not necessarily
dependency on θ. From G17, equation (3.65), the quadrupole be real valued.
potential term is

B2 1  
4.3.1 H Field Radial Component
Vquad (r ) = 3 3 cos2 θ – 1 .
r 2 By substituting the radial components from (23) and (25) into
where B2 is the weighting coefficient from the series expansion (27) we have,
in G17.
From equation (2.23) of G17, the electric field is the nega- a3
Hr = 3 R1 (a) cos χ cos θ
tive gradient of the potential. Hence, the polar component of r  
the E field is 1 h1 (ρ) h2 (ρ)
– 2a1 sin θeiϕ + 9a2 sin 2θeiϕ e–iωt ,
icµ0 ρ ρ
1 ∂V B2
Equad, θ = – = 4 3 cos θ sin θ.
r ∂θ r which, when r = a (ρ = α), will be equal to the boundary
The functions depending on the variable θ match those condition on Hr from (14). We can use eiϕ e–iωt = ei(ϕ–ωt) =
of the boundary condition (14). Therefore, the boundary eiλ to write the exponent in terms of λ and that eiλ = cos λ +
condition only requires the static component of the electric i sin λ. The constants can then be determined by matching
field to be a quadrupole. Comparing the above result to (14) trigonometric functions. For sin θ,
we can uniquely determine B2 for the situation,  
1 h1 (α) iλ
< – a1 e = R1 (a) sin χ cos λ,
B2 icµ0 α
3 cos θ sin θ = –ωµ0 aR1 (a) cos χ cos θ sin θ,
a4
1 where the < operator has come from (22). Hence, a1 must be
=⇒ B2 = – ωµ0 a5 R1 (a) cos χ.
3 purely imaginary to cancel out i in the bottom of the fraction
so that cos λ is obtained instead of sin λ.
and on applying the gradient function to the potential in spher-
ical coordinates (see Appendix 5), 2 h1 (α)
– a1 cos λ = R1 (a) sin χ cos λ,
icµ0 α
∂V 1 a4  
R1 (a)icµ0 sin χ α
Equad, r = – = – ωµ0 a 4 R1 (a) cos χ 3 cos2 θ – 1 , =⇒ a1 = – . (28)
∂r 2 r 2 h1 (α)
1 ∂V a4
Equad, θ = – = –ωµ0 aR1 (a) 4 cos χ cos θ sin θ,
r ∂θ r As there is no sin 2θ term in the boundary condition, we
1 ∂V will have a2 = 0. The gives the radial component of the H
Equad, ϕ = – = 0.
r sin θ ∂ϕ field as the real part of

The trigonometric identities 3 cos2 θ – 1 = 12 (3 cos 2θ + 1)

 3 
a α h1 (ρ) iλ
and cos θ sin θ = 21 sin 2θ, can be used to write the quadrupole Hr = R1 (a) 3 cos χ cos θ + sin χ sin θe .
r h1 (α) ρ
Publications of the Astronomical Society of Australia 7

4.3.2 E Field Polar Component 4.3.3 E Field Azimuthal Component

Now considering the Eθ component which can be found by The final boundary condition is on the Eϕ component. We
combing the θ̂ components of results (23) and (26) into (27), combine the results from above with azimuthal component of
(27), which yields
1 a4
Eθ = – ωµ0 aR1 (a) 4 cos χ sin 2θ – a1 ih1 (ρ)eiϕ e–iωt
r
 
2 3i
Eϕ = – –a1 h1 (ρ) cos θ + b2 (ρh02 (ρ) + h2 (ρ)) cos θ eiλ
1 0 ρ
– b1 (ρh1 (ρ) + h1 (ρ)) cos θeiϕ e–iωt 1

α ρh02 (ρ) + h2 (ρ) h1 (ρ)

ρ = ωµ0 aR1 (a) 0 – i sin χ cos θeiλ ,
2 αh2 (α) + h2 (α) ρ h1 (α)
3
– b2 (ρh02 (ρ) + h2 (ρ)) cos 2θeiϕ e–iωt ,
ρ and clearly at r = a (ρ = α) this reduces to zero and matches the
and when r = a this is equal to the Eθ component from (14). required boundary condition Eϕ (a, θ, ϕ) = 0, implying that
Once again, matching functions of θ and using eiϕ e–iωt = eiλ , the values for the constants are consist with all the boundary
we find that for the term excluding any function of θ, conditions in this situation.
 
cµ0 R1 (a) α iλ
< – h1 (α) sin χe 4.3.4 Remaining Vector Field Components
2 h1 (α)
Noting that the expressions for the constants a1 and b2 can be
1 rewritten using α = (ω /c)a, the field components are given
= – ωµ0 aR1 (a) sin χ cos λ,
2 by the real part of the following,
then it is convenient to change α to (ω /c)a,
1 a3 ρh0 (ρ) + h1 (ρ)
 
1
Hθ = R1 (a) cos χ sin θ – a1 1 cos θ + b2 (3ih2 (ρ) cos θ) eiλ
ωµ0 aR1 (a) 1 2 r3 icµ0 ρ
– sin χ cos λ = – ωµ0 aR1 (a) sin χ cos λ, 1
 3
a

α ρh01 (ρ) + h1 (ρ)
2 2 = R1 (a) 3 cos χ sin θ +
2 r h1 (α) ρ
 
1 α
the LHS and RHS match, which validates the result for a1 + ωa 0 h2 (ρ) sin χ cos θeiλ
c αh2 (α) + h2 (α)
found in (28). As the boundary condition does not contain any 1
 3
a

α ρh01 (ρ) + h1 (ρ) α2 h2 (ρ)
 
= R1 (a) 3 cos χ sin θ + + 0 sin χ cos θeiλ ,
cos θ terms, 2 r h1 (α) ρ αh2 (α) + h2 (α)

1
–b1 (ρh01 (ρ) + h1 (ρ)) cos θeiλ = 0,
 
1 i
Hϕ = – a1 (ρh01 (ρ) + h1 (ρ)) – 3b2 h2 (ρ) cos 2θ eiλ
ρ (29) icµ0 ρ
α ρh01 (ρ) + h1 (ρ) α2
 
1 1
=⇒ b1 = 0. = R1 (a) i
2 h1 (α) ρ
sin χ –
i αh02 (α) + h2 (α)
h2 (ρ) sin χ cos 2θ eiλ

α ρh01 (ρ) + h1 (ρ) α2

 
1
Lastly, the cos 2θ terms, = R1 (a)
2 h1 (α) ρ
+ 0
αh2 (α) + h2 (α)
h2 (ρ) cos 2θ i sin χeiλ ,
 
3
< –b2 (αh02 (α) + h2 (α)) cos 2θeiλ 1 a4
Er = – ωµ0 a 4 R1 (a) cos χ (3 cos 2θ + 1) – 9b2
h2 (ρ)
sin 2θeiλ
α 4 r ρ
1

1a 4 α h2 (ρ)

1 = ωµ0 aR1 (a) – 4 cos χ(3 cos 2θ + 1) + 3 0 sin χ sin 2θeiλ .
= – ωµ0 aR1 (a) cos 2θ sin χ cos λ, 2 2r αh2 (α) + h2 (α) ρ
2
again we want cos λ instead of sin λ so b2 must be purely real. The components of the general field equations are gathered in
Hence, Appendix 1 for easier reference. A plota of the magnetic field
is shown in Fig. 3.
3
–b2 (αh02 (α) + h2 (α)) cos 2θ cos λ
α 5. The Radiation Fields of the Star
1 Far from a star, the electromagnetic fields take on their radi-
= ωµ0 aR1 (a) cos 2θ sin χ cos λ,
2 ation form, allowing us to find the fields which ‘leak’ out of
1 1 α the magnetosphere and propagate away from the electromag-
=⇒ b2 = – ωµ0 aR1 (a) 0 sin χ. (30)
2 3 αh2 (α) + h2 (α) netic source. To find the radiation form of the electromagnetic
fields we consider the limit in which r  c/ω (Griffiths, 2017).
All the results for the constants thus far lead to,
 4 Hence, from (17) if r  c/ω then ρ  1. Then the limit
1 a of the spherical Bessel functions when ρ is large (44) can be
Eθ = ωµ0 aR1 (a) – 4 cos χ sin 2θ
2 r substituted into (38). However, we find that α = (ω /c)a  1,
ρh02 (ρ) + h2 (ρ)
  
+
α
cos 2θ –
h1 (ρ)
sin χ e iλ
.
which means h1 (α) can be approximated using (43). Using
αh02 (α) + h2 (α) ρ h1 (α) c ' 3 × 108 m s–1 and the typical star values given by D55,
(31)
D55 has a typographical error in his Appendix for eqn. a ' 109 m, ω ' 10–5 rad/s,
α ρh02 (ρ)+h2 (ρ)
(31). The αh0 (α)+ h2 (α) ρ term is instead written as
2 a For animations of the solution see https://github.com/jsa113/General_
αh02 (α)+h2 (α) ρ
α ρh02 (ρ)+h2 (ρ)
. Field_Equation_gifs
8 J. C. Satherley et al.

where λ + ρ = ϕ – ω t + ω r/c = ω(r/c – t) + ϕ. Repeating the

same limit and substituting the approximations in for the polar
component yields,
" ! # !
ieiρ
α2 α3 –ieiρ eiρ

1 iλ
Hθ ' < R1 (a) + – 2 sin χ cos θe
2 α α94i – α33i
ρ –i ρ ρ
 5
α3 α3
  
1 α
=< R1 (a) + + 2 sin χ cos θei(λ+ρ) .
2 6ρ ρ iρ

The term with the largest magnitude in the line above is the
α3 /ρ term as α  1 and ρ  1. So,

α3
 
1
Hθ ' < R1 (a) sin χ cos θei(λ+ρ)
2 ρ
α3
 
Figure 3. The magnetic field from (38) using typical star parameters is plot- 1
ted as streamlines showing the direction and the log magnitude of the mag- =< R1 (a) sin χ cos θ [cos(λ + ρ) + i sin(λ + ρ)]
2 ρ
netic field strength represented by the colour. Directly vertical is the star’s
rotation axis, and the magnetic field axis is set as χ = π /4. Note the dashed 1 ω2 3 1 h r  i
red circle is the light cylinder, where ρ = 1 or r = c/ω. = a R 1 (a ) sin χ cos θ cos ω – t + ϕ .
2 c2 r c

The azimuthal component has similar intermediate steps as

we get α ' 3 × 10–5  1. While a typical neutron star has
above. Hence skipping intermediate working gives,
a ' 104 m, ω ' 2π rad/s, " ! # !
α2 ieiρ α3 –ieiρ –eiρ

1 iλ
Hϕ ' < R1 (a) 9i – 3i
cos 2θ – + i sin χe
which gives α ' 2 × 10–4  1. We can also express this as a 2 ρ i ρ ρ2
α4 α3
 6  3 3
scaled equation,
  
1 α 1 α α
=< R1 (a) cos 2θ + + 2 i sin χei(λ+ρ) .
2 9 – 3α ρ ρ iρ
 
–4 a ω 
α = 2 × 10
104 m 2π rad/s The largest term in the line above is the α3 /ρ term. Removing
all smaller magnitude terms yields,
The fastest recorded neutron star has a ' 1.6 × 104 m and
ω ' 4.5 × 103 rad/s (Hessels et al., 2006). We see from these α3
 
1 i(λ+ρ)
values that α ' 0.2 < 1. Though this limit is not as strong as Hϕ ' < R1 (a) i sin χe
2 ρ
slower neutron stars, it shows an appreciable upper limit that
1 ω2 3 1 h r  i
is still less than one. Suggesting that taking the approximation = a R 1 (a ) sin χ sin ω – t + ϕ .
α  1 in the following derivations is valid for slower stars but 2 c2 r c
can still apply to faster stars, but with some error.
Now consider the E field. Using similar intermediate steps
and the same principle as before where the largest order term
5.1 Radiation Form Derivation is the only term considered and the limit forms of h1 (ρ) and
To derive the radiation form we take the limit in which r  h1 (α) are used. The E field’s radial component will be,
c/ω . In taking this limit, we only consider terms that will
be of the largest magnitude for each electromagnetic field
! !
1 ieiρ α
component, neglecting terms containing high powers of r Er ' < ωµ0 aR1 (a)3 sin χ sin 2θeiλ
in their denominator. Recall < denotes the real part of an
2 ρ2 α α94i – α33i
expression from (22) and λ = ϕ – ω t from (6). Working firstly α4 1
 
3 i(λ+ρ)
on the H field’s radial component, after neglecting smaller =< ωµ0 aR1 (a) sin χ sin 2θ e
2 6 ρ2
magnitude terms and substituting the limit forms of h1 (ρ) and
h1 (α) in, Hr becomes however, due to the α4 /ρ2 term, the radial component is equiv-
alently equal to zero when compared to the other E field com-
e i ρ α3
 
Hr ' < R1 (a) 2 sin χ sin θeiλ ponents, this implies that Er ≈ 0. The polar component is,
ρ i
ω a3 1
     iρ
  
4 –eiρ + ieρ 2
= < R1 (a) 2 sin χ sin θei(λ+ρ) 1
Eθ ' <  ωµ0 aR1 (a) 
α  cos 2θ – α i ρ i
e  sin χe  λ
c r i

2 6i ρ iρ
ω a3 1
 
 4
–1 α2
   
= < R1 (a) 2 sin χ sin θ [cos(λ + ρ) + i sin(λ + ρ)] 1 α i
c r i =< ωµ0 aR1 (a) + 2 cos 2θ – sin χei(λ+ρ)
2 6i ρ ρ iρ
ω 1 h r 
1 ω 2 µ0 3
i h r 
1
= a3 R1 (a) 2 sin χ sin θ sin ω
i
–t +ϕ , '– a R1 (a) sin χ sin ω –t +ϕ .
c r c 2 c r c
Publications of the Astronomical Society of Australia 9

By using some of the previous intermediate steps from above, Now we evaluate the integrals over the functions that
the azimuthal component is, depend on θ and ϕ. The double integral can firstly be expressed
 4 as,
–1 α2
   
1 α i
i sin χ cos θei(λ+ρ)
Eϕ ' < ωµ0 aR1 (a) + 2 –
Z 2π Z π h r  i
2 6i ρ ρ iρ sin(θ) sin2 ω –t +ϕ
1 ω 2 µ0 3 1 h r  i 0 0 c
'– a R1 (a) sin χ cos θ cos ω –t +ϕ . h r  i
2 c r c + sin(θ) cos2 (θ) cos2 ω – t + ϕ d θ d ϕ,
c
The results of the radiation field equations are gathered in (35)
Appendix 2 as equations (39) and (40). The results obtained and using the integration identities,
here match the results obtained by D55. Z 2π
These results are interesting as we can see the retarded time sin(θ) d θ = 2,
affect, ( rc – t), on the magnetic field in the radiation zone. Z 2π
0
That is we would not expect an instantaneous change to the 2
sin(θ) cos2 (θ) d θ = ,
electromagnetic fields when it is far from the surface from 3
Z 0π
the star as electromagnetic “news” travels at the speed of light h r  i
(G17). Instead we have a delayed affect which depends on the sin2 ω – t + ϕ dϕ
c
distance from the star. Z π0 h r  i
= cos2 ω – t + ϕ d ϕ = π,
0 c
5.2 Power Equation - D55 Equation (15)
which leads to the integral (35) yielding 8π /3. On combining
The power P passing through a surface is the integral of the
this result with (34) we get
Poynting vector (S = E × H) over said surface. From G17,
Eqn. (11.1), dW 2πω 4 µ0 6 2
I – = a R1 (a) sin2 (χ). (36)
P(r , t) = S · da. (32) dt 3c3
This is the same result obtained by D55 in his equation (15).
This is the power passing through the surface, so a negative The RHS is always positive. Hence, dW/dt must always be
must be introduced to give the rate at which energy is radiated negative. That is, energy is always being radiated away from
away from the system, the star. The energy that is radiated comes from the rotational
kinetic energy causing the star’s spin down.
dW
I
– = S · da. (33)
dt 5.2.1 Spin Down Rate - D55 Equation (16)
To derive the spin down rate of a star in the VRD solution,
For the E and H fields G17 advocates taking the fields when
we require two standard relationships (see for example Moebs
r is very large, when they are in radiation form. Therefore, by
et al. (2016), Eqns. (10.31) and (11.8)),
taking the field equations (39) and (40) (which apply when
r  c/ω ) leads to the Poynting vector, dL dW dW dL
τ = , = P = τ · ω, =⇒ = · ω,
      dt dt dt dt
0 Hr Eθ Hϕ – Eϕ Hθ
where L is the angular momentum and τ is the torque. As the
S =  Eθ  ×  H θ  =  Eϕ Hr 
angular velocity and the angular momentum are in the same
Eϕ Hϕ Eθ Hr
direction, we can rearrange the equation above to become,
and by also taking the surface to be a sphere of radius r, this dL ω dW
implies that Sr k a, Sθ ⊥ a and Sϕ ⊥ a, as a is the normal of = ,
dt kωk2 dt
the surface (see Fig. 1a). So (33) becomes,
Z 2π Z π which is the result obtained by D55.
dW However, another typo is present in D55. It is not obvi-
– = (Eθ Hϕ – Eϕ Hθ )r 2 sin(θ) d θ d ϕ.
dt 0 0 ously shown that his ω is actually a vector and that the ω in
the bottom of the fraction is the norm of ω . Explaining why
Upon applying expressions for the H and E fields from (39) he has ω /ω 2 rather than just 1/ω . Lastly, D55 displays it with
and (40) we get, a negative so it can be compared easier with (36),
 
dL ω dW
Z 2π Z π – = – . (37)
dW 1 ω 4 µ0 6 2 n h r  i
dt kωk2 dt
– = 3
a R1 (a) sin2 (χ) sin2 ω –t +ϕ
dt 0 0 4 c c
2 2
h r  io It can be seen from this equation that the angular momentum
+ cos (θ) cos ω – t + ϕ sin(θ) d θ d ϕ. also decreases with time. The star will be brought to rest by
c
(34) its own radiation as D55 states.
10 J. C. Satherley et al.

6. Conclusion Michel, F. C. 1991, Theory of neutron star magnetospheres, Theoretical

By considering an idealised star and assuming a dipolar form astrophysics (Chicago: University of Chicago Press)
Moebs, W., Ling, S. J., & Sanny, J. 2016, University Physics Volume 1 (Hous-
for the internal magnetic field, the general field equations
ton, Texas: OpenStax)
of the electromagnetic fields external of the star were found
Pétri, J. 2013, MNRAS, 433, 986
analytically. These general field equations were then able to be Pétri, J. 2015, Monthly Notices of the Royal Astronomical Society, 450, 714
reduced to their radiation limit (r  c/ω ). From the radiation Rezzolla, L., Ahmedov, B. J., & Miller, J. C. 2001, Monthly Notices of the
form, the relationship for the power radiated by the rotating Royal Astronomical Society, 322, 723
magnetic field was found. Roberts, P. H. 2007, in Encyclopedia of Geomagnetism and Paleomagnetism,
It was found that the power equation matches the equation ed. D. Gubbins & E. Herrero-Bervera (Dordrecht: Springer Netherlands),
7–11
found in D55 and the derivation for which is shown in this
Stratton, J. A. 1941, Electromagnetic Theory, International series in physics
report for scrutiny. However, a typo appears in D55’s general (McGraw-Hill book Company, Incorporated)
field equations. In particular, the Eθ (31) component appears Travelle, P. A., ed. 2011, Pulsars: discoveries, functions, and formation (Haup-
to evaluate the wrong fraction at α. This error in D55 causes pauge, N.Y: Nova Science Publishers)
trouble when reducing the general field equations to their Van Brummelen, G. 2012, Heavenly Mathematics (Princeton University Press)
radiation form. When using the Eθ component found in this
report the reduction to radiation form matches D55.
A plot of the magnetic field using the general field equations
was also produced, showing the summation of the magnetic
dipole and the wave solution. This generalises a similar plot
given in D55 which was zoomed in around the r  c/ω region.
A 1969 paper by Goldreich and Julian argues that a pulsar
cannot be surrounded by a vacuum but rather a magnetosphere
containing plasma (Goldreich & Julian, 1969). A related ap-
proach can be taken when estimating the power equation.
The Force-Free (FF) magnetosphere, which is the solution to
Maxwell’s Equations for a rotating star with a dipole field. It
ignores the plasma pressure and requires that the electric field
parallel to the magnetic field is zero everywhere (Contopoulos
et al., 1999). This solution requires approximations just like
the VRD solution. The most realistic way is to use numerical
simulations. The results from Kalapotharakos et al. (2018) find
their power equations ranging from the near FF solution to
near the VRD solution given in equation (36). The VRD so-
lution proves to be an interesting limiting case of numerical
simulations.
The VRD solution discussed in this review has been ex-
tended in a number of ways. For example, general relativistic
corrections to the Deustch solution can be added (Rezzolla
et al., 2001; Pétri, 2013). Also, considering an electromagnetic
field that consists of more multipole terms than just a dipole
adds extra correction terms that are potentially required to
match reality (Pétri, 2015; Bonazzola et al., 2020).

References
Bonazzola, S., Mottez, F., & Heyvaerts, J. 2020, arXiv e-prints,
arXiv:2007.02539
Contopoulos, I., Kazanas, D., & Fendt, C. 1999, The Astrophysical Journal,
511, 351, arXiv: astro-ph/9903049
Deutsch, A. J. 1955, Annales d’Astrophysique, 18
Goldreich, P., & Julian, W. H. 1969, ApJ, 157, 869
Griffiths, D. J. 2017, Introduction to Electrodynamics, 4th edn. (Cambridge
University Press), doi:10.1017/9781108333511
Hessels, J. W. T., Ransom, S. M., Stairs, I. H., et al. 2006, Science, 311, 1901
Kalapotharakos, C., Brambilla, G., Timokhin, A., Harding, A. K., & Kazanas,
D. 2018, The Astrophysical Journal, 857, 44
Melatos, A. 1997, Monthly Notices of the Royal Astronomical Society, 288,
1049
Michel, F., & Li, H. 1999, Physics Reports, 318, 227
Publications of the Astronomical Society of Australia 11

Appendix 1. General Field Equations

The results of the general field equations found in Sec. 4 are gathered here for reference. The fields external a star for the VRD
solution is the real part of the following,
 3 
a α h1 (ρ) iλ
Hr = R1 (a) cos χ cos θ + sin χ sin θ e ,
r3 h1 (α) ρ
α ρh01 (ρ) + h1 (ρ)
 3
α2
  
1 a iλ
Hθ = R1 (a) 3 cos χ sin θ + + 0 h2 (ρ) sin χ cos θe ,
2 r h1 (α) ρ αh2 (α) + h2 (α)
α ρh01 (ρ) + h1 (ρ) α2
 
1
Hϕ = R1 (a) + 0 h2 (ρ) cos 2θ i sin χeiλ ,
2 h1 (α) ρ αh2 (α) + h2 (α)
 4  (38)
1 1a α h2 (ρ)
Er = ωµ0 aR1 (a) – 4 cos χ(3 cos 2θ + 1) + 3 0 sin χ sin 2θeiλ ,
2 2r αh2 (α) + h2 (α) ρ
ρh02 (ρ) + h2 (ρ)
 4   
1 a α h1 (ρ) iλ
Eθ = ωµ0 aR1 (a) – 4 cos χ sin 2θ + cos 2θ – sin χe ,
2 r αh02 (α) + h2 (α) ρ h1 (α)
ρh02 (ρ) + h2 (ρ) h1 (ρ)
 
1 α
Eϕ = ωµ0 aR1 (a) – i sin χ cos θeiλ .
2 αh02 (α) + h2 (α) ρ h1 (α)

Appendix 2. Radiation Field Equations We will omit the superscript denoting the functions as the
The results of the radiation field equations found in Sec. 5 are third kind for convenience. Differentiating these functions
gathered here for reference. The magnetic field when r  c/ω with respect to ρ yields,
is,
 2
iρ – 2ρ – 2i

h r  iρ
ω 1 h01
= –e
i
Hr = a3 R1 (a) 2 sin χ sin θ sin ω –t +ϕ , ,
c r c ρ3
 3 2
(42)
iρ ρ + 4iρ – 9ρ – 9i

1ω 32 1 r 
0
h i
Hθ = a R1 (a) sin χ cos θ cos ω – t + ϕ , (39) h2 = –e .
2 c2 r c ρ4
1 ω2 3 1 h r  i
Hϕ = a R 1 (a ) sin χ sin ω – t + ϕ .
2 c2 r c In the limit ρ  1 eqns. (41) and (42) become,

The electric field when r  c/ω is,

–i –3i
h1 ≈ 2 , h2 ≈ 3 ,
Er = 0, ρ ρ
(43)
2i 9i
1 ω 2 µ0 3 1 h r  i
h01 ≈ 3 , h02 ≈ 4 .
Eθ = – a R1 (a) sin χ sin ω –t +ϕ , ρ ρ
2 c r c
2
1 ω µ0 3 1 h r  i
Eϕ = – a R1 (a) sin χ cos θ cos ω –t +ϕ . In the limit ρ  1 eqns. (41) and (42) become,
2 c r c
(40)
–eiρ ieiρ
h1 ≈ , h2 ≈ ,
Appendix 3. Analysis of Spherical Bessel Functions ρ ρ
(44)
The general solution for the fields includes spherical Bessel –ieiρ –eiρ
functions of the third kind (first kind Hankel functions). This h01 ≈ , h02 ≈ .
ρ ρ
appendix highlights what these functions look like and what
they are approximately equal to when considered at the limits
ρ  1 and ρ  1. From Ref. Stratton (1941) page 405, The form these Bessel functions take at these limits are
relatively simple and are required in the derivation of a eqns.
  (13), (14), (18) and (19) from D55 (equations (39) and (40) in
(1) iρ ρ + i
h1 (ρ) = –e , this report).
ρ2
 2 (41)
i ρ ρ + 3i ρ – 3

(1)
h2 (ρ) = ie .
ρ3
12 J. C. Satherley et al.

Appendix 4. Associated Legendre Functions

In Appendix IV of Stratton (1941), the associated Legendre
functions are given as,

P0 (cos θ) = 1,
P1 (cos θ) = cos θ,
P11 (cos θ) = sin θ,
(45)
1 1
P2 (cos θ) = (3 cos2 θ – 1) = (3 cos 2θ + 1),
2 4
1 3
P2 (cos θ) = 3 cos θ sin θ = sin 2θ,
2
where θ ∈ [0, π], the subscript denotes the degree of the poly-
nomial, and the superscript denotes the order of the derivative.

C
Appendix 5. Vector Calculus in Spherical Coordinates a b
Let f (r , θ, ϕ) be a scalar function and A(r , θ, ϕ) = (Ar , Aθ , Aϕ )
be a vector-valued function in spherical coordinates (r , θ, ϕ). A
B
From appendix A of G17, the gradient of a scalar function is,
c
∂f 1 ∂f 1 ∂f
∇f = r̂ + θ̂ + ϕ̂. (46) (a)
∂r r ∂θ r sin θ ∂ϕ z, ω

The divergence of a vector field is,

∇ · A = r12 ∂∂r r 2 Ar + r sin

1 ∂ 1 ∂ Aϕ λ


θ ∂θ (Aθ sin θ ) + r sin θ ∂ϕ .
χ θ
e ξ
(47) –ν ψ̂
ψ
The curl of a vector field is, ξ
  θ̂
1 ∂ ∂ Aθ
∇×A= (Aϕ sin θ) – r̂
r sin θ ∂θ ∂ϕ
 
1 1 ∂ Ar ∂
+ – (rAϕ ) θ̂ (48)
r sin θ ∂ϕ ∂ r
 
1 ∂ ∂ Ar
+ (rAθ ) – ϕ̂. (b)
r ∂r ∂θ
Figure 4. Spherical triangles with general notation and the labels accord-
ing to parameters defined in the report. (a) Spherical Triangle with side
Appendix 6. Solutions of Spherical Triangles lengths and angles labelled according to the notation in this appendix. Note
From Van Brummelen (2012) on page 98 the spherical Law of that sides and angles labelled with the same letter are opposite each other.
Cosines is given as, (b) Description according to coordinate parameters and inclination of e axis.
Note that ν is an angle which is 2π periodic so –ν ≡ 2π – ν (note angles
increase though anticlockwise).
cos c = cos a cos b + sin a sin b cos C, (49)

and on page 63 the spherical Law of Sines is given as,

sin a sin b
= . (50)
sin A sin B
Upon rearranging the spherical Law of Cosines for cos C we
get,
cos c – cos a cos b
cos C = . (51)
sin a sin b