RESEARCH Revista Mexicana de Fı́sica 61 (2015) 211–223 MAY-JUNE 2015

1
Geometry of spin 2 particles
G. Sobczyk∗
Universidad de las Américas-Puebla, Departamento de Fı́sico-Matemáticas,
72820 Puebla, Pue., México.
Received 18 August 2014; accepted 18 March 2015

The geometric algebras of space and spacetime are derived by sucessively extending the real number system to include new mutually
anticommuting square roots of ±1. The quantum mechanics of spin 1/2 particles are then expressed in these geometric algebras. Classical 2
and 4 component spinors are represented by geometric numbers which have parity, providing new insight into the familiar bra-ket formalism
of Dirac. The classical Dirac Equation is shown to be equivalent to the Dirac-Hestenes equation, so long as the issue of parity is not taken
into consideration, the latter quantity being constructed in such a way that it is parity invarient.

Keywords: Bra-ket formalism; geometric algebra; spacetime algebra; Schrödinger-Pauli equation; Dirac equation; Dirac-Hestenes equation;
spinor; spinor operator.

PACS: 02.10.Xm; 03.65.Ta; 03.65.Ud

1. Geometric Concept of Number extended number system, and give the new quantity i := e12
the geometric interpretation of a directed plane segment or
In his book “Number: The Language of Science” Tobias bivector. Note that
Dantzig, in describing the invention of matrices has this to
say: i 2 = e1 e2 e1 e2 = −e21 e22 = −1
(It is) a theory in which a whole array of elements is so that i has the same algebraic property as the imaginary unit
regarded as a number-individual. These “filing cabi- of the complex numbers C. Indeed, every unit bivector of the
nets” are added and multiplied, and a whole calculus of n-dimensional Euclidean space Rn shares this property and
matrices has been established which may be regarded is the generator of rotations in the vector plane of that bivec-
as a continuation of the algebra of complex numbers. tor.
These abstract beings have lately found a remarkable The Euler identity
interpretation in the quantum theory of the atom, and
in man’s other scientific fields, [2]. eiθ = cos θ + i sin θ

We now show how these “filing cabinets” take on the for θ ∈ R and the unit bivector i = e12 , and the hyperbolic
interpretation of geometric numbers in a geometric number Euler identity
system called geometric algebra. It was William Kingdon
ee1 φ = cosh θ + e1 sinh θ
Clifford (1845-1879), himself, who first referred to his fa-
mous algebras as geometric algebras, but some authors still for φ ∈ R and the unit vector e1 , are easily established as
call these algebras Clifford algebras [3]. David Hestenes and special cases of the general algebraic definition of the expo-
other theoretical physicists over the last half-century have nential function
shown this comprehensive geometric interpretation imbues X∞
Xn
the equations of physics with new meaning [4]. We shall re- eX ≡ = cosh X + sinh X.
strict our attention to developing geometric algebras of the n=0
n!
plane, 3-dimensional space, and Minkowski spacetime and
In [5, Chp. 2] and [6], the author shows that the hy-
their application to quantum mechanics, but the ideas apply
perbolic number plane, regarded as the extension of √ the real
to higher dimensional spaces as well [5].
number system to include a new sqrare root e1 := +1 of
+1, has much in common with the more famous complex
1.1. Geometric algebra of the plane
number plane. Since it took centuries for the complex num-
We extend the real number system R to include two new an- bers to be recognised as bonified numbers, I suppose that it
ticommuting square roots e1 , e2 of +1, which we identify as is not surprising that the sister hyperbolic numbers are dis-
unit vectors along the x and y axis of the coordinate plane criminated against even to this day. After all, it was Leopold
R2 . Thus, Kronecker (1823-1891) who said, “God made the integers,
all the rest is the work of man”.
e21 = e22 = 1, and e12 := e1 e2 = −e2 e1 . The standard basis over the real numbers of the geomet-
ric algebra G2 := G(R2 ) is
We assume that the associative and distributive laws of mul-
tiplication of real numbers remain valid in our geometrically G2 = span{1, e1 , e2 , e12 }. (1)
212 G. SOBCZYK

With respect to this basis a geometric number g ∈ G2 can be with respect to the spectral basis (3). From (4) it follows that
written µ ¶
1
g = (1 e1 )u+ [g] . (6)
g = α + v = (a + bi) + (ce1 + de2 ) (2) e1
The interesting thing about equation (6) is that it can be
where α = a + bi = reiθ behaves formally like a complex
turned around and solved directly for for the matrix of [g] of
number and v = ce1 + de2 ∈ R2 is a vector in the Euclidean
g. To accomplish this we define the e1 -conjugate of g ∈ G2
plane of the bivector i = e12 . It can also be easily checked
by
that vα = αv, where α = a − βi is the (complex) conjugate
g e1 := e1 ge1 .
of α. It is this property that gives α the geometric significance
as the generators of rotations in the plane of e12 , [5, p.55]. Multiplying Eq. (6) on the left and right by
µ ¶
Another basis of G2 , the spectral basis, gives the alge- 1
bra of real 2 × 2 matrices the interpretation of representing u+
e1
geometric numbers in G2 . We write
µ ¶ and
1 (1 e1 )u+ ,
G2 = span{ u+ (1 e1 )}
e1 respectively, gives
µ ¶ µ ¶
u+ e1 u− 1
= span{ }, (3) u+ g(1 e1 )u+
e1 u+ u− e1
µ ¶ µ ¶
or G2 = span{u+ , e1 u+ , e1 u− , u− }, where the idempo- 1 e1 1 e1
= u+ u+ [g] u+ = u+ [g].
tents u± := (1/2)(1 ± e2 ). The relationship between the e1 1 e1 1
standard basis (1) and the spectral basis (3) is directly ex-
By taking the e1 -conjugate of the last equation, we get
pressed by µ ¶
     1
1 1 0 0 1 u+ u− g e1 (1 e1 )u− = u− [g].
e1
 e1   0 1 1 0   
    e1 u+ 
 e2  =  1 0 0 −1  e1 u−  Adding the last two equations gives the desired result
µ ¶ µ e ¶
e12 0 1 −1 0 u− g g e1 g 1 e1 g
[g] = u+ u+ + u− u− . (7)
e1 g g e 1 g e1 g
Noting e1 u+ = u− e1 , it follows that
µ ¶ Let µ ¶
1 1
(1 e1 )u+ = u+ + e1 u+ e1 = u+ + u− = 1. gk = (1 e1 )u+ [gk ]
e1 e1
Using this last relationship, and the fact that where [gk ] is the matrix of the geometric number gk ∈ G2
for k = 1, 2 as given in (2), (4) and (6). The algebra of real
u2+ = u+ = e2 u+ , and u+ e1 u+ = 0 = u+ e12 u+ , 2 × 2 matrices M atR (2) is algebraically isomorphic to the
geometric algebra G2 because [g1 + g2 ] = [g1 ] + [g2 ] and
with similar relationships for u− gu+ , u+ gu− and u− gu− , [g1 g2 ] = [g1 ][g2 ] for all 2 × 2 matrices [g1 ], [g2 ] ∈ M atR (2)
we calculate for the geometric number g given in (2), and their corresponding geometric numbers g1 , g2 ∈ G2 . For
µ ¶ µ ¶
1 1 example, using (6) and u+ e1 u+ = 0,
g = (1 e1 )u+ g(1 e1 )u+ µ ¶ µ ¶
e1 e1 1 1
µ ¶ µ ¶ g1 g2 = (1 e1 )u+ [g1 ] (1 e1 )u+ [g2 ]
α+v (α + v)e1 1 e1 e1
= (1 e1 )u+ u+ µ ¶ µ ¶
e1 (α + v) e1 (α + v)e1 e1 u+ u+ e1 u+ 1
µ ¶µ ¶ = (1 e1 )[g1 ] [g2 ]
a+d c−b 1 u+ e1 u+ u+ e1
= (1 e1 )u+ (4) µ ¶ µ ¶
c+b a−d e1 1 1
= (1 e1 )u+ [g1 ][g2 ] = (1 e1 )u+ [g1 g2 ]
= (a + d)u+ e1 e1
which shows that [g1 g2 ] = [g1 ][g2 ].
+ (c − b)e1 u− + (c + b)e1 u+ + (a − d)u− . (5)
Extending the real number system R to include the new
We say that anticommuting square roots e1 , e2 of +1 is well defined since
µ ¶ the resulting geometric number system G2 is algebraically
a+d c−b isomorphic to the 2 × 2 matrix algebra M atR (2). Indeed,
[g] =
c+b a−d each geometric algebra Gp,q is either algebraically isomor-
is the matrix of the geometric number phic to a real or complex square matrix algebra, or matrix
subalgebra, of the appropriate dimension, [5, p.205]. Next,
g = (a + be12 ) + (ce1 + de2 ) ∈ G2 we consider the case of the geometric algebra G3 of space.

Rev. Mex. Fis. 61 (2015) 211–223

 1
GEOMETRY OF SPIN 2
PARTICLES 213

1.2. Geometric algebra of space Since the unit trivector i commutes with all the elements
of G3 , we can simply complexify the spectral basis (3) of G2
The geometric algebra G2 of the xy-plane in R2 can be ex- to obtain the spectral basis of G3 ,
tended to the geometric algebra G3 of the space R3 simply by µ ¶
extending G2 to include e3 , a new square root of +1, which 1
G3 = span{ u+ (1 e1 )}
anticommutes with both e1 and e2 . Naturally, e3 has the ge- e1
ometric interpretation of a unit vector along the z-axis. The µ ¶
u+ e1 u −
standard basis of G3 , with respect to the coordinate frame = span{ }, (13)
e1 u+ u−
{e1 , e2 , e3 } of the Euclidean space R3 , is
where in this case u± := (1/2)(1 ± e3 ). We can then directly
G3 : =G(R3 )=span{1, e1 , e2 , e3 , e12 , e13 , e23 , e123 }. (8) apply the analogous formulas (4), (5), (6), and (7) to elements
g ∈ G3 merely by allowing the values of a, b, c, d to be of the
The elements e12 , e13 , e23 are unit bivectors in the xy-, xz-,
form s + it where s, t ∈ R and i = e123 . The famous Pauli
and yz-planes and generate rotations in those planes, and
matrices of the coordinate frame {e1 , e2 , e3 } are simply ob-
i := e123 := e1 e2 e3 is the unit trivector or oriented vol-
tained using the spectral basis (13), with u± = (1/2)(1±e3 ),
ume element of R3 . The unit trivector i also has the property
getting
that i 2 = −1, as follows from
µ ¶ µ ¶
0 1 0 −i
i 2 = e1 (e2 e3 )e1 (e2 e3 ) = e21 (e2 e3 )2 = −1. [e1 ] = , [e2 ] =
1 0 i 0
µ ¶
Any geometric number g ∈ G3 can be written in the form 1 0
[e3 ] = −i[e1 ][e2 ] = . (14)
0 −1
3
X
g= αk ek (9) In this representation, the imaginary unit i acquires the
k=0 geometric interpretation of the oriented volume element in
G3 , as follows from
where e0 := 1 and αk = ak + ibk for ak , bk ∈ R and where
µ ¶
0 ≤ k ≤ 3. The conjugation known as the reverse g † of the 1 0
geometric number g, is defined by reversing the orders of all [e123 ] = [e1 ][e2 ][e3 ] = i . (15)
0 1
the products of the vectors that make up g, giving
In terms of the geometric product, all other products in
3
X the geometric algebra are defined. For example, given vec-
g† = α k ek . (10) tors a, b ∈ G13 ≡ R3 ,
k=0
ab = a · b + a ∧ b ∈ G0+2
3 , (16)
In particular, writing g = s + v + B + T, the sum of a real
number s ∈ G03 , a vector v ∈ G13 , a bivector B ∈ G23 and
where a · b := (1/2)(ab + ba) ∈ R is the symmetric inner
a trivector T ∈ G33 , g † = s + v − B − T as can be easily
product and a ∧ b := (1/2)(ab − ba) is the antisymmetric
checked.
outer product of the vectors a and b, respectively. The outer
Another conjugation widely used in geometric algebra is
product satisfies a∧b = i(a×b), expressing the duality rela-
the grade inversion, obtained by replacing each vector in a
tionship between the standard Gibbs-Heaviside cross product
product by its negative. It corresponds to an inversion in the
a × b and the outer product a ∧ b ∈ G23 . In what follows the
origin, otherwise known as a parity inversion. For the geo-
outer product a ∧ b ∧ c of three vectors is also used. Similar
metric number g given in (9), the grade inversion is
to (16), we write
3
X
g − := α0 − α k ek . (11) a(b ∧ c) = a · (b ∧ c) + a ∧ (b ∧ c), (17)
k=1
where
When g is written in the form g = s + v + B + T, the grade 1¡ ¢
inversion takes the form g − = s − v + B − T. a · (b ∧ c) : = a(b ∧ c) − (b ∧ c)a
2
Alternatively, we can obtain the geometric algebra G3 by
extending or complexifying G2 to include a commuting imag- = −a × (b × c) ∈ G13 ,
inary i, which we subsequently identify as the unit trivector.
We write and
1¡ ¢
G3 =G2 (i):=span{1, e1 , e2 , e21 i, e12 , −e2 i, e1 i, i}, (12) a ∧ (b ∧ c) : = a(b ∧ c) + (b ∧ c)a
2
and then define e3 := e21 i, e13 = −e2 i and e23 = e1 i. = a · (b × c)i ∈ G33 .

Rev. Mex. Fis. 61 (2015) 211–223

214 G. SOBCZYK

A much more detailed treatment of G3 is given in [5,Chp.3], where h i0 denotes real scalar part of the enclosed expres-
and in Ref. 7 I explore the close relationship that exists be- sion, and using a·b = habi0 that we found in (16). A closely
tween geometric algebras and their matrix counterparts. Ge- related calculation gives almost the same answer
ometric algebra is developed in Ref. 5 as the geometric com-
pletion of the real number system, providing a new founda- (â+ b̂+ )† (â+ b̂+ ) = (b̂+ â+ )(â+ b̂+ ) = b̂+ â+ b̂+
tion for much of mathematics and physics. b̂+ b̂+
= (1 + â)(1 + b̂) = (1 + â + b̂ + âb̂)
4 4
1.3. Properties of idempotents and nilpotents
b̂+ 1
An idempotent p ∈ G3 has the defining property p2 = p. = (1 + â + b̂ − b̂â + 2â · b̂) = (1 + â · b̂)b̂+ ,
4 2
√ course +1, 0 are idempotents, and so is p = (1/2)(1 +
Of
but multiplied by the idempotent b̂+ . In summary,
2e1 + i e2 ). For a general geometric number g = ω + m +
in ∈ G3 to be an idempotent, we must have (â+ b̂+ )† (â+ b̂+ ) = b̂+ â+ b̂+ = P rob(â+ |b̂+ )b̂+ , (22)
g 2 = ω 2 + m2 − n2 + 2i m · n where P rob(â+ |b̂+ ) := (1/2)(1+ â· b̂) and (g1 g2 )† = g2† g1†
+ 2ω(m + i n) = ω + m + i n, (18) for all g1 , g2 ∈ G3 is the operation of reverse defined in (10).
Another important property that easily follows is
or equivalently, 1
b̂+ mb̂+ = b̂+ m(1 + b̂)
2
ω 2 + m2 − n2 + 2i m · n = ω
1
= b̂+ (m − b̂m + 2b̂ · â) = (b̂ · m)b̂+ . (23)
and 2
(2ω − 1)(m + i n) = 0. A geometric number g is a nilpotent if g 2 = 0. A similar
analysis to (18) for a general idempotent shows that a nilpo-
If m + i n 6= 0, it follows that ω = (1/2), m · n = 0, and tent N has the canonical form
m2 − n2 = (1/4). Equivalently, a geometric number A, not
equal to 0 or 1, is an idempotent iff 1 1
N = (r + i s) = r (1 + i r−1 s) = rn̂+ = n̂− r, (24)
2 2
1
A= (1 + m + i n), where where n̂ := i r−1 s ∈ G13 for the orthogonal vectors r · s = 0
2 and r2 = s2 . Thus, every nilpotent N has associated with it
m2 − n2 = 1, and m · n = 0, (19) a corresponding idempotent n̂+ .

for m, n ∈ G13 . It follows immediately that any idempotent

A has the important cannonical form 2. Quantum mechanics of spin
1 1 The observable corresponding to the spin of an electron in
A= (1 + m + i n) = m (1 + m−1 + i m−1 n) a coordinate frame {e1 , e2 , e3 } is a unit vector m̂ = m1 e1
2 2
+m2 e2 + m3 e3 in the Euclidean space R3 of that frame. The
1
= m (1 + â) = mâ+ , (20) Hermitian matrix of the observable m̂ is
2 µ ¶ µ ¶
m3 m1 − im2 m3 m−
where â = m−1 + i m−1 n ∈ G13 , â+ = (1/2)(1 + â) and [m̂] = = ,
m1 + im2 −m3 m+ −m3
â · m = 1. We use the notation â± = (1/2)(1 ± â) for the
mutually annihiliating idempotents â+ and â− . as given in (13) and (14), with u± = (1/2)(1 ± e3 ) and
For the idempotents â± , note that m± = m1 ± im2 , where
µ ¶
1
ââ+ = â+ and ââ− = −â− . m̂ = (1 e1 )u+ [m̂]
e1
Because of these properties, we would like to identify â± as = m3 u+ + m− e1 u− + m+ e1 u+ − m3 u− . (25)
the two pure eigenprojector spin states of an electron. Unfor-
tunately, as is discussed in the next section, this is not quite By a geometric spinor |αi with respect to
correct. u+ = (1/2)(1 + e3 ), we mean
Given a second idempotent b̂+ , we calculate √
|αi := 2(α0 + α1 e1 )u+ ⇐⇒ hα|
1 √
2hâ+ b̂+ i0 = h(1 + â)(1 + b̂)i0 := |αi† = 2u+ (α0 + α1 e1 ) (26)
2
1 where α0 , α1 ∈ G0+3 . The spinor |αi is said to be non-
= (1 + â · b̂), (21) 3
2 degenerate if in addition α0 α0 + α1 α1 6= 0. Spinors have

Rev. Mex. Fis. 61 (2015) 211–223

 1
GEOMETRY OF SPIN 2
PARTICLES 215

the important superposition property that if |αi and |βi are and the unit vector â = m̂e3 m̂ = a1 e1 + a2 e2 + a3 e3 is
spinors with respect to u+ then specified by
|ωi := λ1 |αi + λ2 |βi for λ1 , λ2 ∈ G0+3
3 (27) α0 α1 + α0 α1 i(α0 α1 − α1 α0 )
a1 = , a2 = ,
is also a spinor with respect to u+ . This is precisely the prop- α0 α0 + α1 α1 α0 α0 + α1 α1
erty that naked idempotents m+ , n+ do not have. α0 α0 − α1 α1
Given two spinors |αi and |βi, the geometric product a3 = . (38)
α0 α0 + α1 α1
hα||βi = 2u+ (α0 + α1 e1 )(β0 + β1 e1 )u+ For a normalized spinor (31),
= 2(α0 β0 + α1 β1 )u+ (28) 2 α0 α0 + α1 α1 1
m2 = = = . (39)
is also a spinor. The inner product hα|βi is defined to be the 1 + â · e3 α0 α0 α0 α 0
scalar and pseudoscalar parts of the geometric product of the
Using the above information (34) and (35) for the idem-
spinors,
D E potent A, the spinor |αi takes the canonical form
hα|βi := hα||βi = α 0 β0 + α 1 β1 . (29) √ √ √
0+3 |αi = 2α0 A = 2α0 mu+ = 2α0 m2 â+ u+ . (40)
An important observation which will be used later is that
D¡ ¢† ¡ ¢E The matrix [â+ ] of â+ is the density matrix of the pure state
hα||βi hα||βi = 2hα|βihα|βi, (30) |αi. By taking the reverse of the spinor equation (40), we get
0+3

as can be easily verified. The spinor |αi is said to be a nor- √ √

hα| = 2α0 u+ m = 2α0 m2 u+ â+ , (41)
malized spinor if
hα|αi = α0 α0 + α1 α1 = 1. (31) and find that
Traditionally, a 2-component column spinor is 1
µ ¶ |αihα| = α0 α0 m2 m̂u+ m̂ = â+ . (42)
α0 2
[α]2 := where α0 , α1 ∈ C. (32)
α1 It is interesting to note that a Cartan spinor is specifed by
√
In general, for a spinor |αi := 2(α0 + α1 e1 )u+ with
|αie1 |αi∗ = |αie1 hα|−
α0 6= 0, we can write
√ ³ α1 ´ √ = (α02 − α12 )e1 + i(α02 + α12 )e2 − 2α0 α1 e3
|αi = 2α0 1 + e1 u+ = 2α0 A (33)
α0 α0 2
³ ´ =− m̂(e1 + ie2 )m̂ = − 2 â+ m̂e1 m̂
for A = 1 + (α1 /α0 )e1 u+ . It is easily checked that A is α0 α0
an idempotent, so that A can be expressed in the cannonical
and represents a null complex vector, [8,p.41].
form (20),
There are several other operations on spinors which are
1³ α1 α1 ´
important. By multiplying the spinor equation (40) on the
A= 1+ e1 − i e2 + e3
2 α0 α0 left and right by e3 , we find that
1 1 √
= (1 + m + i n) = m (1 + â) = mu+ , (34) |αie3 : = e3 |αie3 = 2α0 e3 me3 u+
2 2
√
since â = (m + i mn/m2 ) = e3 . Another closely related = 2α0 m2 e3 â+ e3 u+ . (43)
canonical form for the idempotent A, with the help of (23), is
Multiplying the spinor equation (40) on the left and right
A = mu+ = mm̂m̂u+ mu+ = m2 â+ u+ , (35)
by e1 , we get
where we have redefined the unit vector â := m̂e3 m̂. √
All of the above quantities can be expressed in terms of |αie1 : = e1 |αie1 = 2α0 e1 me1 u−
α0 and α1 . We find that √
= 2α0 m2 e1 â+ e1 u− . (44)
1 ³
m= (α1 α0 + α0 α1 )e1
2α0 α0 The spinor |αie3 is called the e3 -conjugate of |αi, and
´
the spinor |αie1 is called the e1 -conjugate of |αi. Using (11)
+ i(α0 α1 − α1 α0 )e2 + e3 , (36)
and (40) the grade inversion |αi− of the spinor |αi ∈ G3 is
1 ³ √ ³ ´− √
n= i(α0 α1 − α1 α0 )e1
2α0 α0 |αi− : = 2 (α0 + α1 e1 )u+ = 2(α0 − α1 e1 )u−
´
√ √
− (α0 α1 + α1 α0 )e2 , (37) = − 2α0 mu− = 2α0 m2 â− u− . (45)

Rev. Mex. Fis. 61 (2015) 211–223

216 G. SOBCZYK

Given normalized spinors |αi, |βi, we calculate with the Using (48) and (49), we also calculate
help of (40) and (23),
h0|b̂|0i = b3 , and h1|b̂|1i = −b3 , (50)
hα|βi = 2αβ0 u+ mnu+
³ ´ and
= 2α0 β0 m · n + i(m × n) · e3 u+ = 0
h1|b̂|0i = b+ = b1 + ib2 , and
iff m · n = 0 and (m × n) · e3 = 0. Expressed in terms of the
associated spinor directions m̂, n̂, the inner product hα|βi is h0|b̂|1i = b− = b1 − ib2 . (51)
quite different than the component expression (29). A great
√
simplification occurs when we calculate hα|βihα|βi, finding Often an electron in the spinor state |0i = 2u √+ is said to be
with the help of (30) and (39), up, and an electron in the spinor state |1i = 2e1 u+ is said
1 D¡ ¢† ¡ ¢E to be down. Using (50), the average value of the observable
hα|βihα|βi = hα||βi hα||βi b̂ in the state |0i is b3 , and the average value of b̂ in the state
2 0+3
|1i is −b3 .
1D ¡ ¢ E
= hβ| |αihα| |βi We can also carry out the above calculation using spinor
2 0+3
operators. Following [10,p.63], given the spinor |αi, the
D E 1
= hβ|â+ |βi = (1 + â · b̂). spinor operator ψ of |αi is defined by
0+3 2
1 ³ ´
Given an observable â, suppose that an electron has been ψ : = √ |αi + |αi−
prepared in the pure spin state |αi. Then the probability of 2
finding the electron in the pure spin state |βi is = m(α0 u+ − α0 u− ) ∈ SU(2), (52)

P rob(|αi, |βi) : = hα|βihα|βi where the inversion |αi− has been defined in (45). The spinor
1 operator is an even multivector in G+ 3 and hence invarient un-
= (1 + â · b̂) = 2hâ+ b̂+ i0 , (46) der a parity swap. It retains the normalization property that
2
ψψ † = det[ψ] = 1, and allows us to calculate the associated
just as we have already seen in (21) and (22). Immediately
direction â of |αi as a rotation of the unit vector e3 . With the
after the measurment is carried out, the electron will be in the
help of (22), (38) and (52), we find that
pure state |βi. Identity (46) is independent of the basis spinor
u+ that was used in the definition (26) of the spinors |αi and
ψe3 ψ † = m(α0 u+ − α0 u− )e3 (α0 u+ − α0 u− )m
|βi. Clearly, the spinor states |αi and |βi are orthogonal iff
â = −b̂. = m(α0 u+ + α0 u− )(α0 u+ − α0 u− )m
Given a spinor state |αi, the average value of the observ-
able b̂ in that state is defined by hα|b̂|αi. Using (22), (23), = α0 α0 m2 m̂(u+ − u− )m̂ = â, (53)
(39), (40), we calculate
√ √ and also
b̂|αi = 2α0 b̂mu+ = 2α0 (b̂ · m + b̂ ∧ m)u+ , (47) 1
ψu+ ψ † = ψ(1 + e3 )ψ † = â+ . (54)
2
and The relationship (54) should be compared with (42).
hα|b̂|αi : = 2hu+ (m̂b̂m̂)u+ i0+3 = (m̂b̂m̂) · e3

= b̂ · (m̂e3 m̂) = b̂ · â. (48) 3. Geometric algebra of spacetime

It follows that the average value lies in between the values The geometric algebra G3 of the coordinate frame
±1. It is important to note that the final values in both (46) {e1 , e2 , e3 } in R3 can be extended to the geometric alge-
and (48) in no way depend upon the idempotent u+ , so u+ bra G1,3 of the pseudoeuclidean space R1,3 of Minkowski
could be replaced by any other idempotent ĉ+ defined by any spacetime, by extending G3 by an additional square root γ0
unit vector ĉ, but we would then also have to redefine |αi. of +1 which anticommutes with each of the space vectors
In the special cases where e1 , e2 , e3 . So γ0 satisfies γ02 = 1 and γ0 ek = −ek γ0 for
√ √ k = 1, 2, 3. Equivalently, the spacelike vectors
|αi = |0i := 2u+ or |αi = |1i := 2e1 u+ ,
γk := ek γ0 ⇐⇒ ek = γk γ0 = γk0 ,
we have
√
b̂|0i = 2(b3 + b+ e1 )u+ and together with the timelike vector γ0 , define the spacetime al-
√ gebra
b̂|1i = 2(b− − b3 e1 )u+ . (49) G1,3 = gen{γ0 , γ1 , γ2 , γ3 }

Rev. Mex. Fis. 61 (2015) 211–223

 1
GEOMETRY OF SPIN 2
PARTICLES 217

of the orthonormal spacetime frame {γµ | 0 ≤ µ ≤ 3} in and similarly γ31 = e13 , it follows that
R1,3 . In summary, the spacetime vectors obey the rules
e13 u++ = u+− e13 , e3 u++
γ02 = 1, γk2 = −1, γµ γν = −γν γµ = u−+ e3 , e1 u++ = u−− e1 , (56)
for µ 6= ν, µ, ν = 0, 1, 2, 3, and k = 1, 2, 3. Note also that where
pseudoscalar 1
u+− : = (1 + γ0 )(1 − iγ12 ),
4
γ0123 = γ10 γ20 γ30 = e123
1
u−+ : = (1 − γ0 )(1 + iγ12 ),
of G1,3 is the same as the pseudoscalar of the rest frame 4
{e1 , e2 , e3 } of G3 , and it anticommutes with each of the 1
u−− : = (1 − γ0 )(1 − iγ12 ).
spacetime vectors γµ for µ = 0, 1, 2, 3. 4
The forgoing shows that the geometric algebra G3 is nat- The idempotents u++ , u+− , u−+ , u−− are mutually anni-
urally isomorphic to the even subalgebra G+ 1,3 of the space- hiliating in the sense that the product of any two of them is
time algebra G1,3 generated by the bivectors γµ γν of G1,3 , zero, and partition unity
[9]. Any element g ∈ G1,3 can be expressed in the form
g = G1 + G2 γ0 where G1 , G2 ∈ G3 . In particular, a unit u++ + u+− + u−+ + u−− = 1. (57)
vector m̂ = m1 e1 + m2 e2 + m3 e3 in the geometric algebra By the spectral basis of the Dirac algebra G1,3 we mean
G3 of the rest frame {e1 , e2 , e3 } becomes the unit spacetime the elements of the matrix
bivector  
1
e13 
m̂ = (m1 γ1 + m2 γ2 + m3 γ3 )γ0 ∈ G21,3 ,   u++ (1 − e13 e3 e1 )
 e3 
with the spacelike vector m1 γ1 + m2 γ2 + m3 γ3 ∈ G11,3 as a e1
factor.  
u++ −e13 u+− e3 u−+ e1 u−−
In the above, we have carefully distinguished the rest e13 u++ u+− e1 u−+ −e3 u−− 
frame {e1 , e2 , e3 } in the spacetime algebra G1,3 . Any =  e3 u++
 . (58)
e1 u+− u−+ −e13 u−− 
other rest frame {e01 , e02 , e03 } can be obtained by an ordi-
e1 u++ −e3 u+− e13 u−+ u−−
nary space rotation of the rest frame {e1 , e2 , e3 } followed
by a Lorentz boost. In the spacetime algebra G1,3 , this Any geometric number g ∈ G1,3 can be written in the
is equivalent to defining a new frame of spacetime vectors form  
1
{γµ0 | 0 ≤ µ ≤ 3} ⊂ G1,3 , and the corresponding rest frame −e13 
g = (1 e13 e3 e1 )u++ [g]  
0
{e0k = γk0 γ00 | k = 1, 2, 3} of a Euclidean space R3 moving  e3  (59)
with respect to the Euclidean space R3 defined by the rest
e1
frame {e1 , e2 , e3 }. The way we introduced the geometric
algebras G3 and G1,3 may appear novel, but they perfectly where [g] is the Dirac matrix corresponding to the geometric
reflect all the common relativistic concepts [5,Chp.11]. number g. In particular,
 
The well-known Dirac matrices can be obtained as a sub- 1 0 0 0
algebra of the 4×4 matrix algebra M atC (4) over the complex  0 1 0 0 
[γ0 ] = 
 0 0 −1 0  ,

numbers. We first define the idempotent
0 0 0 −1
1
u++ : = (1 + γ0 )(1 + iγ12 )  
4 0 0 0 −1
1  0 0 −1 0 
= (1 + iγ12 )(1 + γ0 ), [γ1 ] = 
 0 1 0
, (60)
4
(55) 0 
√ 1 0 0 0
where the unit imaginary i = −1 is assumed to com-
and
mute with all elements of G1,3 . Whereas it would be nice  
to identify this unit imaginary i with the pseudoscalar ele- 0 0 0 i
ment γ0123 = e123 as we did in G3 , this is no longer possible  0 0 −i 0 
[γ2 ] = 
 0
,
since γ0123 anticommutes with the spacetime vectors γµ as −i 0 0 
previously mentioned. i 0 0 0
Noting that  
0 0 −1 0
 0 0 0 1 
γ12 = γ1 γ0 γ0 γ2 [γ3 ] = 
 1
.
0 0 0 
= e2 e1 = e21 0 −1 0 0

Rev. Mex. Fis. 61 (2015) 211–223

218 G. SOBCZYK
 
It is interesting to see what the representation is of the 1
basis vectors of G3 . We find that for k = 1, 2, 3, −e13 
ψ = (1 e13 e3 e1 )u++ [ψ]α 
 e3 

µ ¶
[0]2 [ek ]2 e1
[ek ]4 = [γk ][γ0 ] = and
[ek ]2 [0]2
µ ¶ = α1 + α2 e13 + α3 e3 + α4 e1 ∈ G+
1,3 . (64)
[0]2 [1]2
[e123 ]4 = i , To establish the last equality above, with the help of (57),
[1]2 [0]2
note that
where the outer subscripts denote the order of the matrices
ψ = ψu++ + ψu+− + ψu−+ + ψu−− ,
and, in particular, [0]2 , [1]2 are the 2 × 2 zero and unit matri-
ces, respectively. The last relationship shows that the i occur- and then succesively show that
ing in the Pauli matrix representation (15), which represents ³ ´
√ ψu±± = α1 + α2 e13 + α3 e3 + α4 e1 u±± ,
the oriented unit of volume, is different than the i = −1
which occurs in the complex matrix representation (60) of for each of different sign combinations ++, +−, −+, −−.
the of Dirac algebra. As an even geometric number in G+ 1,3 , ψ generates
A Dirac spinor is a 4-component column matrix [ϕ]4 , Lorentz boosts in addition to ordinary rotations in the
  Minkowski space R1,3 . The amazing property of (64) is
ϕ1
 ϕ2  that where as we started
√ by formally introducing the com-
[ϕ]4 :=  
 ϕ3  for ϕk ∈ C. (61) plex number i = −1, in order to represent the Dirac
gamma matrices, we have ended up with √ a spinor operator
ϕ4
ψ ∈ G+ 1,3 in which the role of i = −1 is taken over by
Following [10,p.143], we use the Dirac spinor to construct the γ21 = e12 ∈ G3 . This is the key idea in the Hestenes
the (matrix) spinor operator representation of the Dirac equation [18].
  Although we√ have formally eliminated the need for the
ϕ1 −ϕ2 ϕ3 ϕ4 artificial i = −1 in the Dirac spinor by replacing it with
 ϕ2 ϕ1 ϕ4 −ϕ3 
[ψ] =   the bivector γ21 = e12 in the spinor operator, such a quan-
 ϕ3 ϕ4 ϕ1 −ϕ2  . (62)
tity occuring in a real
ϕ4 −ϕ3 ϕ2 ϕ1 √ geometric algebra is unacceptable.
The imaginary i = −1 has the effect of complexifying
the geometric algebra G1,3 to the complex geometric alge-
Unlike the Dirac spinor [ϕ]4 , the spinor operator [ψ] is invert-
bra G4 (C)=M e at4 (C), [10,p.217]. We might, instead, search
ible iff det[ψ] 6= 0. We find that
for a larger real geometric algebra containing the geometric
det[ψ] = r2 + 4a2 ≥ 0, (63) algebra G1,3 , just as we enlarged the geometric algebra G3
of space to the geometric algebra G1,3 of spacetime. The
where real geometric algebra G2,3 , obtained from the geometric
algebra G1,3 by assuming the existence of a new timelike
r = |ϕ1 |2 + |ϕ2 |2 − |ϕ3 |2 − |ϕ4 |2 and vector γ00 which anticommutes with all the spacetime vec-
¡ ¢ tors γµ ∈ G1,3 , has the required properties. The unit pseu-
a = = ϕ1 ϕ3 + ϕ2 ϕ4 .
doscalar i 0 := γ00 γ0 γ1 γ2 γ3 ∈ G2,3 has
√ all the desired alge-
Where-as the spinor operator [ψ] obviously contains the braic properties of the imaginary i = −1, commuting with
same information as the Dirac spinor [ϕ]4 , we will shortly all the elements of G1,3 and having square −1. All of the pre-
see that it acquires the geometric interpretation of an even vious arguments in the complex algebra G4 (C) remain valid
multivector in G+1,3 .
in G2,3 , but now i 0 is the oriented element of volume of the
Noting that 5-dimentional pseudoeclidean space R2,3 .
However, the real geometric algebras of the pseudoeu-
1 clidean spaces R4,1 and R0,5 also fit the bill, as is seen in
u++ γ21 = (1 + γ0 )(γ21 + iγ12 γ21 )
4 Table I in [10, p.217]. In [11, p.326], Hestenes argues that
1 the geometric algebra G4,1 of R4,1 might be an even bet-
= (1 + γ0 )(i − γ12 ) = iu++ = γ21 u++ ,
4 ter choice because the even subalgebra G+ 4,1 is isomorphic to
G1,3 in exactly the same way that G+ is isomorphic to G3 .
it follows
¡ that u++ ϕk =¢ u++ <(ϕk ) + u++ iγ12 =(ϕk )i = 1,3
In the paper, “Vector Analysis of Spinors” [12], the author
u++ <(ϕk ) + γ21 =(ϕk ) and hence
explores the interpretation of a geometric Pauli spinor on the
u++ [ψ] = u++ [ψ]α , Riemann sphere, ideas which he then generalizes in another
paper to geometric Dirac spinors in the complex geometric
where each of the elements αk in [ψ]α is defined by αk = algebra G3 (C) [13]. These real and complex geometric alge-
ϕk |i→γ21 . Using (59), the geometric number ψ ∈ G+
1,3 cor- bras open up many new possibilities for the interpretation of
responding to the matrix spinor operator [ψ] is relativistic quantum mechanics.

Rev. Mex. Fis. 61 (2015) 211–223

 1
GEOMETRY OF SPIN 2
PARTICLES 219

4. Tensor products of geometric numbers where we are using the qubit notation

Tensor products of geometric numbers, and their correspond-

ing geometric algebras, are necessary for describing multi- |00i := 2u+ ⊗ u+ , |01i := 2u+ ⊗ e1 u+
particle systems used, for example, in the construction of a
quantum computer, or even in the much simpler “double slit” and, similarly, for the definitions of |10i and |11i. Taking the
experiments. The resulting quantum entanglement, which reverse of the product state |αi ⊗ |βi gives
implies the “spooky action-at-a distance” that even Einstein
couldn’t stomach, is discussed in Sec. 5. ³ ´†
By the tensor product G3 ⊗ G3 of the geometric alge- |αi ⊗ |βi = hα| ⊗ hβ|
bra G3 with itself, we essentially mean two distinguishable
copies of the same geometric algebra G3 . We can express and
this as ³ ´³ ´
G3 ⊗ G3 = {g1 ⊗ g2 | g1 , g2 ∈ G3 }. (65) hα| ⊗ hβ| |αi ⊗ |βi := hα|αihβ|βi.
In addition, the subspace consisting of everything which
commutes with the entire algebra, called the center of G3 ,
The product state |αi ⊗ |βi represents the state of two
may be identified with the complex numbers
unentangled electrons, where each electron is prepared in-
C ≡ span{1, i = e123 }. dependently, and can be measured independently. A direct
consequence of (48) is that for any values s, t ∈ [−1, 1] cho-
The tensor products of g1 , g2 , g3 , g4 ∈ G3 and α, β ∈ C sat- sen, directions m̂ and n̂ can always be found resulting in the
isfies the rules average values
α(g1 ⊗ g2 ) = (αg1 ) ⊗ g2 = g1 ⊗ (αg2 ), (66)
hα|m̂|αi = s, and hβ|n̂|βi = t.
(αg1 + βg2 ) ⊗ g3 = α(g1 ⊗ g3 ) + β(g2 ⊗ g3 ), (67)
and The states |S± i = |01i ± |10i are entangled states. In
an entangled state if one of the electrons is measured to be up
g1 ⊗ (αg2 + βg3 ) = α(g1 ⊗ g2 ) + β(g1 ⊗ g3 ). (68)
along a given direction, then the other will definitely be down
Either the standard basis (8) or the spectral basis (13) can without the need for measurement. For any direction m̂, the
be tensored with itself to get the respective standard basis or averages of the states |S± i measured in the directions m̂ ⊗ 1
spectral basis of the tensor algebra G3 ⊗ G3 over C. By the and 1 ⊗ m̂ is 0, meaning that measurements of the first and
product of tensors g1 ⊗ g2 and g3 ⊗ g4 , we mean second electrons along this direction are equally likely to be
³ ´³ ´ ±1. With the help of (49), we easily calculate
g1 ⊗ g2 g3 ⊗ g4 = (g1 g3 ) ⊗ (g2 g4 ). (69)

The tensor product (65) satisfies the same algebraic rules (m̂ ⊗ 1)|S± i = m̂|ui ⊗ |1i ± m̂|1i ⊗ |0i
(67) and (68) as the geometric product itself. This suggests ³ ´
= m3 |0i + m+ |1i ⊗ |1i
extending the tensor product to commuting copies of the geo-
metric algebra G1,3 of spacetime. In [14,15] the authors iden- ³ ´
tify the tensor product of N copies of G3 with commuting ± − m3 |1i + m− |0i ⊗ |0i
copies of G+ 1,3 in the geometric algebra GN,3N to represent
³ ´
the spinor operators of an N -qubit quantum computer. As ex- = m3 |01i + m+ |11i
plained in Ref. 14, in order to get the usual tensor products of ³ ´
complex matrices that is used for multi-distinguishable spin ± − m3 |10i + m− |00i ,
(1/2) particles, one must introduce a commuting idempotent
to project down to only a single set of complex numbers. from which we find that
³ ´
5. Non-entangled and entangled states hS± |m̂ ⊗ 1|S± i = h01| ± h10|
√ √
If |αi = 2(α0 + α1 )u+ and |βi = 2(β0 + β1 )u+ are nor- ³ ´
malized states, then the product state is given by the tensor m3 |01i + m+ |11i ∓ m3 |10i ± m− |00i
product
³√ ´ ³√ ´ = m3 − m3 = 0,
|αi ⊗ |βi = 2(α0 + α1 e1 )u+ ⊗ 2(β0 + β1 e1 )u+
and a similar calculation shows that hS± |1 ⊗ m̂|S± i = 0.
= α0 β0 |00i + α0 β1 |01i + α1 β0 |10i + α1 β1 |11i, (70) For a further insightful discussion of these results, see [1].

Rev. Mex. Fis. 61 (2015) 211–223

220 G. SOBCZYK

6. The Schrödinger, Pauli, and Dirac Equa- for the electric field potential V and the magnetic field B =
tions −i(∇ ∧ A) = ∇ × A of a particle of mass m with electric
charge e, and where i = e123 is the unit pseudoscalar of G3 .
Spin and quantum entanglement have been studied by many Expanding Eq. (76), we find that
authors. The Susskind Lectures [1] give a fascinating ac- 1 ³ e2
count, presenting clearly the issues that separate quantum i~∂t |αi = − ~2 ∇2 + 2 A2
2mc c
and classical mechanics. Another elementary presentation is ¢´
e¡
given in [16]. Of course, no account of quantum mechan- + i~ ∇A + A∇ |αi + eV |αi
ics can be complete without at least a cursory presentation of c
the fundamental Schrödinger and Dirac equations, the foun- 1 ³ e2 e¡ ¢
= − ~2 ∇2 + 2 A2 + i~ ∇ · A + 2A · ∇
dation upon which all of quantum mechanics is built. 2mc c c
The Schrödinger equation is obtained from the classical e ´
expression for the total energy H of a particle at the position − ~ B |αi + eV |αi
c
x = xe1 + ye2 + ze3 , This is equivalent to the classical 2 × 2 matrix form of this
1 1 2 equation,
H= mv2 + V = p + V, (71) ∂[α]2 1 ³ e ´2
2 2m i~ = i~[∇] + [A] [α]2 + eV [α]2 , (77)
∂t 2mc c
where p = px e1 + py e2 + pz e3 is the momentum vector of
the particle and V is the potential, by way of the substitution obtained by taking the matrix of both sides of (76) and ex-
tracting the first column, where [ek ] are the Pauli matrices
p → −i~∇, (72) given in (14) and [ϕ]2 is the 2-component spinor (32). The
matrix Eq. (77) can be expanded in the same way that we
where ∇ := ∇x is the standard gradient with respect to the expanded Eq. (76). See reference [17,eqn(A.4)]. The reader
position x ∈ R3 , ~ = h/2π and h is Planck’s constant. The is referred to the Appendix for more technical details.
√ ¡
Schrödinger equation thus becomes Recall
¢ that the spinor operator (52) is ψ = (1/ 2) |αi +
³ ´ |αi− . If |αi satisfies the Schrödinger-Pauli equation (76),
∂ϕ ~ 2 then |αi− will satisfy the parity inversion of this equation,
i~ = Ĥϕ = − ∇ + V ϕ, (73)
∂t 2m
1 ³ e ´2
−i~∂t |αi− = i~∇ − A |αi− + eV |αi− . (78)
for the Hamiltonian Ĥ = −(~/2m)∇2 + V , where 2mc c
ϕ := ϕ(x, t) ∈ C. √ spinor ψ will satisfy the equa-
It then follows that the operator
When the potential V is independent of time, there will tion obtained by taking (1/ 2) times the sum of equations
be a complete set of stationary states (76) and (78), giving the parity invarient Schrödinger-Pauli-
Hestenes equation
ϕn (x, t) = e−iEn (x)t/~ 1 ³ e ´2
i~∂t ψe3 = i~∇ + A ψu+
where the spatial wave function ϕn (x) satisfies the time in- 2mc c
dependent Schrödinger equation 1 ³ e ´2
+ i~∇ − A ψu− + eV ψ
³ ´ 2mc c
~ 2
− ∇ + V ϕn (x) = En ϕn (x), (74) 1 ³ e2 ´
2m = − ~2 ∇2 + 2 A2 ψ
2mc c
in terms of which the time dependent Schrödinger equation i~e ¡ ¢
has the general solution + 2
∇ · A + 2A · ∇ ψe3
2mc
X
ϕ(x, t) = cn ϕn (x)e−iEn t/~ , (75) ~e
− Bψ + eV ψ,
n 2mc2
[17, eqn.(1.8)].
[16,p.122]. According to the Born interpretation of the wave
The Schrödinger equation (73) applies to all quantum
function, the probability of finding the particle in the in-
phenomena except magnetism and relativity. The magnetism
finitesimal volume
and spin of an electron are taken care of in the Schrödinger-
|d3 x| = dxdydz is |ϕ(x, t)|2 |d3 x|. Pauli equation (76), or equivalently (77). In order to extend
quantum mechanics to special relativity, we start with the ex-
The Schrödinger equation (73) is generalized to include pression for relativistic energy (E 2 /c2 ) − p2 = m2 c2 . Fol-
spin in the Schrödinger-Pauli equation [10,p.51]. Expressed lowing [19, (5.5)] and [10, p.136], we express the Dirac equa-
in terms of our spinor (26) with respect to u+ , tion in terms of the Dirac spinor operator (64), getting what
is known as the Dirac-Hestenes equation
1 ³ e ´2
i~∂t |αi = i~∇ + A |αi + eV |αi (76) ~∂ψγ21 − eAψ = mψγ0 , (79)
2mc c

Rev. Mex. Fis. 61 (2015) 211–223

 1
GEOMETRY OF SPIN 2
PARTICLES 221
P3 1
where ∂ = µ=0 γ µ (∂/∂xµ ), and an interaction with a elec- √ [|βie1 ] = [(β0 + β1 e1 )u− ]
tromagnetic 2
P3field is included by way of the spacetime poten- µ ¶µ ¶ µ ¶
tial A = µ=0 γµ Aµ . 0 β1 0 0 0 β1
Just like for the Shrödinger-Pauli equation (76), the clas- = =
0 β0 0 1 0 β0
sical Dirac equation can be retrieved from its spinor oper- µ ¶
ator form (79). First, multiply both sides of the Eq. (76) β1
so the right column spinor is represented by the matrix
on the right by the primitive idempotent u++ , using that µ ¶ β0
u++ γ0 = u++ and u++ γ21 = iu++ , and using the spec- 0 β1
. Unlike with ordinary 2-component spinors (32),
tral basis relations (58), (59), and (60). Finally, extracting the 0 β0
first column on each side of the resulting matrix equation, we the full matrix [g] of a geometric number g ∈ G3 can be
obtain recovered from the corresponding matrices of left and right
i~[∂][ϕ]4 − e[A][ϕ]4 = m[ϕ]4 (80) column spinors,
√ 1
where [ϕ]4 is the Dirac spinor (61) and i = −1. To show g = gu+ + gu− = √ (|αi + |βie1 )
the full equivalence of (80) and (79), one must also show 2
µ ¶µ ¶
that (80) implies (79), which is taken up in the Appendix. α0 β1 1
= (1 e1 )u+ .
In [17-19], and in many other papers, Hestenes explores new α1 β0 e1
geometric features of the Dirac theory of the electron made Taking the matrix representation of (28), we find that
possible by the spinor operator form (79) of the Dirac equa-
tion. [hα||βi] = 2[u+ ][(α0 + α1 e1 )][(β0 + β1 e1 )][u+ ]
= 2(α0 β0 + α1 β1 )[u+ ]
Appendix or
µ ¶µ ¶
In this Appendix, we explore in greater detail how our pre- 1 0 α0 α1
sentation of quantum mechanics is related to the more usual 0 0 α1 α0
approaches. µ ¶µ ¶ µ ¶
β0 β1 1 0 α0 β0 + α1 β1 0
Multiplying the spectral basis (13) on the right by = ,
β1 β0 0 0 0 0
u+ = (1/2)(1 + e3 ) gives the left spinor basis
µ ¶ µ ¶ from which it follows that
1 u+ e1 u− D E
u+ (1 e1 )u+ = u+ hα|βi = [hα||βi]
e1 e1 u + u− 0+3
µ ¶ µ ¶
u+ 0 α0 β0 + α1 β1 0
= (81) = Tr = α0 β0 + α1 β1 .
e1 u + 0 0 0
The matrix of the spinor operator (52) is given by
Similarly, multiplying the spectral basis (13) on the right
by u− = (1/2)(1 − e3 ) gives the right spinor basis 1 ¡ ¢
[ψ] = √ [|αi] + [|αi− ]
µ ¶ µ ¶ 2
1 u+ e1 u− µ ¶ µ ¶
u+ (1 e1 )u− = u− α0 0 0 −α1
e1 e1 u + u− = +
µ ¶ α1 0 0 α0
0 e1 u − µ ¶
= . (82) α0 −α1
0 u− =
α1 α0
Taking the matrix of each side of Eq. (26), we get √
Recalling (40), |αi = 2α0 mu+ , the calculation (42), is
1 equivalent to the matrix calculation
√ [|αi] = [(α0 + α1 e1 )u+ ]
2 1
µ ¶µ ¶ µ ¶ [|αihα|] = α0 α0 [m][u+ ][m] = [â+ ]
2
α0 α1 1 0 α0 0
= = where the matrices [m], [u+ ] and [â+ ] are specified by
α1 α0 0 0 α1 0
µ α1 ¶
µ ¶ 1 α0
α0 [m] = α1 ,
so the left column spinor is represented by the matrix α0 −1
µ ¶ α1 µ ¶
α0 0 √ 1 0
. The factor of (1/ 2) is inserted as a matter of [u+ ] = ,
α1 0 0 0
convenience in the way that we defined the Hermitian inner µ ¶
product (29). Similarly, taking the matrix of each side of the 1 1 + a3 a−
[â+ ] = .
e1 -conjugate (44), we get 2 a+ 1 − a3

Rev. Mex. Fis. 61 (2015) 211–223

222 G. SOBCZYK

Analogously to (82) for the spectral basis (13) of the Pauli corresponds to the Dirac spinor (61).
algebra, we can extract any column of the spectral basis (58) The classical Dirac equation (80) is equivalent to the
of the Dirac algebra. For example, to extract the third column Dirac-Hestenes equation (79) multiplied on the right by u++ ,
we multiply (58) on the right by u−+ to get ³ ´
  i~∂ψ − eAψ − mψ u++ = 0. (83)
u++ −e13 u+− e3 u−+ e1 u−−
e13 u++ u+− e1 u−+ −e3 u−−  Taking the spacetime inversion of this equation, by replac-
 u
 e3 u++ e1 u+− u−+ −e13 u−−  −+ ing all spacetime vectors by their negatives, gives the parity
e1 u++ −e3 u+− e13 u−+ u−− related equation
  ³ ´
0 0 e3 u−+ 0
 0 0 e1 u−+ 0  i~∂ψ − eAψ + mψ u−+ = 0.
= 0 0
.
u−+ 0 
Two more equations are obtained, which are parity related to
0 0 e13 u−+ 0
(83), by taking the complex conjugate of both of these last
Multiplying the spinor operator (64) on the right by u++ two equations, giving
we obtain ³ ´
i~∂ψ + eAψ + mψ u+− = 0,
ψu++ = ϕ1 u++ + ϕ2 e13 u++ + ϕ3 e3 u++ + ϕ4 e1 u++ ,
and ³ ´
whose spinor matrix i~∂ψ + eAψ − mψ u−− = 0.
  But the sum of these four parity related equations is exactly
ϕ1 0 0 0
 ϕ2 0 0 0  the parity invarient Dirac-Hestenes equation (79).
 
[ψu++ ] = 
 ϕ3 0 0 0 

 ϕ4 0 0 0  Acknowledgement
  I thank Professor Melina Gomez Bock of the Universidad de
α1 0 0 0 Las Americas-Puebla for many fruitful hours discussing the
 α2 0 0 0 

=  M od(u++ ) intricacies of quantum mechanics. I am also grateful to Dr.
α3 0 0 0  Timothy Havel for invaluable remarks leading to extensive
α4 0 0 0 revisions of an earlier version of this work.

∗. http://www.garretstar.com 10. P. Lounesto, Clifford Algebras and Spinors, 2nd Edition. Cam-
1. L. Susskind, 9 YouTube Lectures: Quantum Entanglements, bridge University Press, Cambridge, 2001.
Part 1, (Stanford University 2008). 11. D. Hestenes, Clifford Algebra and the Interpretation of Quan-
2. T. Dantzig, NUMBER: The Language of Science, Fourth Edi- tum Mechanics, in Clifford Algebras and Their Applications
tion, (Free Press, 1967). in Mathematical Physics, edited by J.S.R. Chisholm and A.K.
Common, NATO ASI Series C: Mathematical and Physical Sci-
3. W.K. Clifford, On the Classification of Geometric Algebras, in
ences Vol. 183, D. Reidel (Publishing Company 1985).
Mathematical Papers by William Kingdon Clifford, edited by
Robert Tucker, London, Macmillan and Co., 1882. 12. G. Sobczyk, Vector Analysis of Spinors (to appear), http://
www. garretstar.com/nyuvas3-10-15.pdf
4. D. Hestenes, New Foundations for Classical Mechanics, 2nd
Ed. (Kluwer 1999). 13. G. Sobczyk, Spacetime Algebra of Dirac Spinors, (2015) (to
5. G. Sobczyk, New Foundations in Mathematics: The Geometric appear), http://www.garretstar.com/diracspin03-24-15.pdf
Concept of Number, Birkhäuser, (New York 2013). 14. T.F. Havel, J.L. Doran, Geometric Algebra in Quantum Infor-
http://www.garretstar.com/ mation Processing, Contemporary Mathematics, ISBN-10: 0-
6. G. Sobczyk, Hyperbolic Number Plane, The College Mathe- 8218-2140-7, Vol. 305 (2002).
matics Journal 26, No. 4 (1995) 268-280. 15. C. Doran, A. Lasenby, Geometric Algebra for Physicists, (Cam-
7. G. Sobczyk, Geometric Matrix Algebra, Linear Algebra and its bridge 2007).
Applications, 429 (2008) 1163-1173. 16. D.J. Griffiths, Introduction to Quantum Mechanics, (Prentice
8. E. Cartan, The Theory of Spinors, Dover Publications, (New Hall, Inc. 1995).
York 1981). 17. D. Hestenes and R. Gurtler, Local Observables in Quantum
9. D. Hestenes, Spacetime Algebra, (Gordon and Breach 1966). Theory, Am. J. Phys. 39 (1971) 1028-1038.

Rev. Mex. Fis. 61 (2015) 211–223

 1
GEOMETRY OF SPIN 2
PARTICLES 223

18. D. Hestenes, Zitterbewegung in Quantum Mechanics, Found ORY”, in: A Symposium on the Mathematics of Phys-
Physics 40 (2010) 1-54. http://geocalc.clas.asu.edu/pdf/ZBWin ical Space-Time, Facultad de Quimica, Universidad Na-
QM15**.pdf http://geocalc.clas.asu.edu/pdf-preAdobe8/Loc cional Autonoma de Mexico, Mexico City, 67-96, (1981).
ObsinQT.pdf http://geocalc.clas.asu.edu/pdf/Geom Dirac.pdf
19. D. Hestenes, “GEOMETRY OF THE DIRAC THE-

Rev. Mex. Fis. 61 (2015) 211–223