Polar Vector

• Parity
•So far our coordinate transformations have been restricted to pure
passive rotations. We now consider the effect of reflections or inver-
sions.

•If we have transformation coefficients aij = −δij , then

xi = −x0i (1)
which is an inversion or parity transformation.

• Important Point:
The above transformation changes our initial right-handed coordi-
nate system into a left-handed coordinate system. This is an in-
version of the coordinate system or coordinate axes, objects in the
physical world remaining fixed.
•Our prototype vector r with components (x1, x2, x3) transforms to:

r0 = (x01, x02, x03) = (−x1, −x2, −x3) (2)

⇒ This new vector r0 has negative components, relative to the new
transformed set of axes.
1
•If we reverse the directions of the coordinate axes (Fig. (1)) and
also change the signs of the components then
r0 = r
.

Figure 1: Inversion of Cartesian coordinates

⇒ The vector (an arrow in space) stays exactly as it was before the
transformation was carried out.
⇒ “The position vector r and all other vectors whose components
behave this way (reversing sign with a reversal of the coordinate
axes) are called polar vectors and have odd parity”.

• Axial Vector
•A fundamental difference appears when we encounter a vector de-

2
fined as the cross product of two polar vectors.

Let C = A × B, where both A and B are polar vectors. Then the

components of C are given by

C 1 = A2B 3 − A3B 2 (3)

and so on.

•Now, when the coordinate axes are inverted, Ai → −A0i, Bj →

−Bj0 , but from its definition C k → +C 0k .
⇒ Our cross-product vector, vector C, does not behave like a polar
vector under inversion.

•To distinguish, we label it a pseudovector or axial vector (see

Fig. (2)) that has even parity.

•The term axial vector is frequently used because these cross prod-
ucts often arise from a description of rotation.

3
 Figure 2: Inversion of Cartesian coordinates - axial vector

•Examples are
angular velocity, v =ω×r
orbital angular momentum, L=r×p
torque, force = F, N=r×F
∂B
magnetic field field B, = −∇ × E
∂t
•v = ω × r:
The axial vector is the angular velocity ω
r and v = dr/dt are polar vectors.

•Clearly, axial vectors occur frequently in physics, although this fact

4
is usually not pointed out. In a right-handed coordinate system an
axial vector C has a sense of rotation associated with it given by a
right-hand rule.

•In the inverted left-handed system the sense of rotation is a left-

handed rotation. This is indicated by the curved arrows in the above
Fig. (2).

5
• Distinction between polar and axial vectors:

•The distinction between polar and axial vectors may also be illus-
trated by a reflection. A polar vector reflects in a mirror like a real
physical arrow, Fig. (3)a.

•Earlier (in Figs. (1) and (2)) the coordinates are inverted; the phys-
ical world remains fixed.
But here the coordinate axes remain fixed; the world is reflected
- as in a mirror in the xz-plane.

•Specifically, in this representation, we keep the axes fixed and asso-

ciate a change of sign with the component of the vector.

•For a mirror in the xz-plane, Py → −Py . We have

P = (Px, Py , Pz )
P0 = (Px, −Py , Pz ) polar vector

6
 Figure 3: (a) Mirror in xz-plane; (b) mirror in xz-plane.

•An axial vector such as a magnetic field H or a magnetic moment

µ (= current × area of current loop) behaves quite differently under
reflection.
7
 Implications of Axial Vector

•Consider the magnetic field H and magnetic moment µ to be pro-

duced by an electric charge moving in a circular path. Reflection
reverses the sense of rotation of the charge.

•The two current loops and the resulting magnetic moments are
shown in Fig. (3)b. We have
µ = (µx, µy , µz )
µ0 = (−µx, µy , −µz ) reflected axial vector

•If we agree that the universe does not care whether we use a right-
or left-handed coordinate system, then it does not make sense to add
an axial vector to a polar vector.

•In the vector equation A=B, both A and B are either polar vectors
or axial vectors.

•Note:
“The big exception to this is in beta decay, weak inter-
actions, where the universe distinguishes between right- and left-
handed systems. So we add polar and axial vector interactions.”

8
•Similar restrictions apply to scalars and pseudoscalars and, in gen-
eral, to the tensors and pseudotensors considered subsequently.

9
 Pseudoscalars, pseudovectors, and pseudotensors

•Usually, pseudoscalars, pseudovectors, and pseudotensors will trans-

form as

S 0 = JS,
Ci0 = Jaij Cj ,
A0ij = Jaik ajl Akl (4)
where J is the determinant of the array of coefficients,
amn, i.e., the Jacobian of the parity transformation.

•In our inversion the Jacobian is

−1 0 0
J = 0 −1 0 = −1
0 0 −1

•For a reflection of one axis, say x-axis:

−1 0 0
J = 0 1 0 = −1
0 0 1
and again the Jacobian J = −1.

10
•Pure rotations J is always +1.

• Pseudoscalar:
Under Rotations:
The triple scalar product S = A × B · C was shown to be a scalar.

•Under Parity Transformation:

Now by considering the parity transformation, we see that S → −S.
•Conclusion: The triple scalar product is actually a pseudoscalar.

•This behavior was foreshadowed by the geometrical analogy of a vol-

ume. If all three parameters of the volume: length, depth, and height,
change from positive distances to negative distances, the product of
the three will be negative.

11
 Levi-Civita Symbol in 3-dimension

For future use it is convenient to introduce the three-dimensional

Levi-Civita symbol εijk , defined by
ε123 = ε231 = ε312 = 1
ε132 = ε213 = ε321 = −1 (5)
all other εijk = 0 (6)
⇒ εijk is antisymmetric with respect to all pairs of indices.

•Suppose now that we have a third-rank pseudotensor δijk , which in

one particular coordinate system, is equal to εijk .

•Then by definition of pseudotensor:

0
δijk = |a|aipajq akr εpqr (7)

•Now, by direct expansion of the determinant:

a1pa2q a3r εpqr = |a| (8)
0
⇒ Showing that δ123 = |a|2 = 1 = ε123.

12
•Considering the other possibilities one by one, we find
0
For rotations and reflections : δijk = εijk (9)

⇒ Hence εijk is a pseudotensor.

•The usefulness of εpqr extends far beyond this section. For instance,
the (matrix) representation for the generators of the rotation group
are derived from
(Mr )pq = iεpqr
0 0 0 0 0 i 0 −i 0
Mx = 0 0 −i My = 0 0 0 Mz = i 0 0
0 i 0 −i 0 0 0 0 0

•A particle of spin 1/2 in nonrelativistic quantum mechanics is rep-

resented by the set of three 2 × 2 Pauli matrices σ:
     
0 1 0 −i 1 0
σ1 = , σ2 = , σ3 = (10)
1 0 i 0 0 −1
• Algebra of Pauli Matrices:

σiσj + σj σi = 2δij I2, anticommutation

σiσj = iσk , i, j, k a cyclic permutation of 1, 2, 3
(σi)2 = I2,
13
where I2 is the 2 × 2 unit matrix.

• Commutation Relation:
The vector σ/2 satisfies the same commutation relations, as the or-
bital angular momentum, L (L × L = iL:

[σi, σj ] ≡ σiσj = 2iεijk σk (11)

•Conclusion:
The three Pauli matrices σ and the unit matrix form a complete set,
so that any Hermitian 2 × 2 matrix M may be expanded as:

M = m0 I2 + m1σ1 + m2σ2 + m3σ3 = m0 + m · σ , (12)

where the mi form a constant vector m.

• Using

(σi)2 = I2
Trace (σi) = 0 (13)
• Pauli Algebra:
Adding and multiplying such 2 × 2 matrices we generate the Pauli
algebra.

14
• Dirac Matrices
• Dirac extended this formalism to fast-moving particles
of spin 1/2 , such as electrons (and neutrinos).
⇒ Dirac Equation or Relativistic Wave Equation:
⇒ 4 × 4 Gamma (γ) matrices.

•The (matrix) representation of spin 1 particle:

0 1 0 0 −i 0 1 0 0
1 1 1
Mx = √ 1 0 1 My = √ i 0 −i Mz = √ 0 0 0
20 1 0 20 i 0 2 0 0 −1

15
 Levi-Civita Symbol continued

•Much of elementary vector analysis can be written in

a very compact form by using εijk and the identity:
In R3 : εijk εpqk = δipδjq − δiq δjp

•The numerical value of εpqr is given by the triple scalar product of

coordinate unit vectors:
x̂p · x̂q × x̂r

⇒ From this point of view each element of εpqr is a pseudoscalar,

but the εpqr collectively form a third-rank pseudotensor.

•Furthermore, it is seen to be an isotropic pseudotensor with the

same components in all rotated Cartesian coordinate systems.

16
 Dual Tensors

•With any antisymmetric second-rank tensor C (in three-dimensional

space), we may associate a dual pseudovector Ci defined by

1
Ci = εijk C jk (14)
2

•Here the antisymmetric C may be written

C 12 −C 31
 
0
C = −C 12 0 C 23  (15)
C 31 −C 23 0

•We know that Ci must transform as a vector under rotations from

the double contraction of the fifth-rank (pseudo) tensor εijk Cmn but
that it is really a pseudovector from the pseudo nature of εijk .

•Specifically, the components of C are given by

(C1, C2, C3) = (C 23, C 31, C 12) (16)

17
•Notice the cyclic order of the indices that comes from the cyclic
order of the components of εijk .

C = (C1, C2, C3) = (C 23, C 31, C 12)

means that our three-dimensional vector product may literally be
taken to be either a pseudovector or an antisymmetric second-rank
tensor, depending on how we choose to write it out.

•If we take three (polar) vectors A, B, and C, we may define the

direct product:

V ijk = AiB j C k (17)

•By an extension of the earlier analysis, V ijk is a tensor of third rank.

•Define a dual quantity:

1
V = εijk V ijk (18)
3!
⇒ It is clearly a pseudoscalar.

•By expansion it is seen that

18
 A1 B 1 C 1
V = A2 B 2 C 2 (19)
A3 B 3 C 3
is our familiar triple product.

•For use in writing Maxwell’s equations in covariant form, we want

to extend this dual vector analysis to four-dimensional space and,
in particular, to indicate that the four-dimensional volume element
dx0dx1dx2dx3 is a pseudoscalar.

1 µνρσ
F̃ µν = ε Fρσ
2
∂µF̃ µν = 0 (20)

•Due to the antisymmetry of εµνρ,σ ,

∂ λF µν + ∂ µF νλ + ∂ ν F λµ = 0
∂ λF̃ µν + ∂ µF̃ νλ + ∂ ν F̃ λµ = 0 (21)

19
 Levi-Civita symbol in 4-dimension

•We introduce the Levi-Civita symbol εijkl , the four-dimensional

analog of εijk . This quantity εijkl is defined as totally antisymmetric
in all four indices:

• εijkl is defined as +1, if (ijkl) is an even permutation of (0, 1, 2, 3).

A permutation is odd if it involves an odd number of interchanges
of adjacent indices, such as (0 1 2 3) −→ (0 2 1 3).
Even permutations arise from an even number of transpositions
of adjacent indices. ε0123 = +1.

• εijkl is defined as -1, if it is an odd permutation.

• εijkl is 0 if any two indices are equal.

•The Levi-Civita εijkl may be proved a pseudotensor of rank 4 by

analysis similar to that used for establishing the tensor nature of εijk .

•Introducing the direct product of four vectors as fourth-rank tensor

with components

H ijkl = AiB j C k Dl (22)

20
built from the polar vectors A, B, C, and D, we may define the dual
quantity:

1
H= εijkl H ijkl (23)
4!
⇒ A pseudoscalar due to the quadruple contraction with the pseu-
dotensor εijkl .

•Now we let A, B, C, and D be infinitesimal displacements along

the four coordinate axes (Minkowski space),

A = (dx0, 0, 0, 0)
B = (0, dx1, 0, 0), and so on (24)
and
H = dx0dx1dx2dx3. (25)

•The four-dimensional volume element is now identified as a pseu-

doscalar. This result could have been expected from the results of
the special theory of relativity.

•The Lorentz-Fitzgerald contraction of dx1dx2dx3 just balances the

time dilation of dx0.

21
•We slipped into this four-dimensional space as a simple mathemat-
ical extension of the three-dimensional space and, indeed, we could
just as easily have discussed 5-, 6-, or N-dimensional space.

•This is typical of the power of the component analysis. Physically,

this four-dimensional space may be taken as Minkowski space:

(x0, x1, x2, x3) = (ct, x, y, z) (26)

where t is time.

•This is the merger of space and time achieved in special relativity.

The transformations that describe the rotations in four-dimensional
space are the Lorentz transformations of special relativity.

22