a

r
X
i
v
:
1
1
0
8
.
5
4
4
7
v
1


[
m
a
t
h
-
p
h
]


2
7

A
u
g

2
0
1
1
Concepts of trace, determinant and inverse of
Cliord algebra elements
D. S. Shirokov
Steklov Mathematical Institute
Gubkin St.8, 119991 Moscow, Russia
email: shirokov@mi.ras.ru
August 30, 2011
Abstract
In our paper we consider the notion of determinant of Cliord
algebra elements. We present some new formulas for determinant of
Cliord algebra elements for the cases of dimension 4 and 5. Also we
consider the notion of trace of Cliord algebra elements. We use the
generalization of the Paulis theorem for 2 sets of elements that satisfy
the main anticommutation conditions of Cliord algebra.
Keywords: Cliord algebra, determinant, trace, inverse
MSC classes: 15A66
1
1 Introduction
The notion of determinant of Cliord algebra elements was considered in
[3]. In our work we present some new formulas for determinant of Cliord
algebra elements for the cases of dimension n = 4 and 5. Also we consider the
notion of trace of Cliord algebra elements. We use the generalization of the
Paulis theorem for 2 sets of elements that satisfy the main anticommutation
conditions of Cliord algebra.
After writing this paper author found the article [4] on the subject that
is close to the subject of this paper. In particular, the article [4] contains the
formulas that are similar to the formulas for the determinant in this paper.
However, note that for the rst time most of these formulas (n = 1, 2, 3) were
introduced in [3].
2 Complex Cliord algebras
Let p and q be nonnegative integers such that p + q = n 1. We consider
complex Cliord algebra C(p, q). The construction of Cliord algebra C(p, q)
is discussed in details in [2].
Generators e
1
, e
2
, . . . , e
n
satisfy the following conditions
e
a
e
b
+ e
b
e
a
= 2
ab
e,
where = ||
ab
|| is the diagonal matrix whose diagonal contains p elements
equal to +1 and q elements equal to 1.
The elements
e
a
1
. . . e
a
k
= e
a
1
...a
k
, 1 a
1
< . . . a
k
n, k = 1, 2, . . . n
together with the identity element e form a basis of Cliord algebra C(p, q).
The number of basis elements equals to 2
n
.
Any Cliord algebra element U C(p, q) can be written in the following
form
U = ue + u
a
e
a
+

a
1
<a
2
u
a
1
a
2
e
a
1
a
2
+ . . . + u
1...n
e
1...n
, (1)
where u, u
a
, u
a
1
a
2
, . . . u
1...n
are complex constants.
2
We denote the vector subspaces spanned by the elements e
a
1
...a
k
enumer-
ated by the ordered multi-indices of length k by C
k
(p, q). The elements of
the subspace C
k
(p, q) are denoted by
k
U
and called elements of rank k. We
have
C(p, q) =
n
k=0
C
k
(p, q). (2)
Cliord algebra C(p, q) is a superalgebra, so we have even and odd sub-
spaces:
C(p, q) = C
Even
(p, q) C
Odd
(p, q), (3)
where
C
Even
(p, q) = C
0
(p, q) C
2
(p, q) C
4
(p, q) . . . ,
C
Odd
(p, q) = C
1
(p, q) C
3
(p, q) C
5
(p, q) . . .
3 Operations of conjugation
Let denote complex conjugation of matrix by

A, transpose matrix by A
T
,
Hermitian conjugate matrix (composition of these 2 operations) by A

.
Now let dene some operations on Cliord algebra elements.
Complex conjugation. Operation of complex conjugation U

U acts in
the following way

U =

u e +

u
a
e
a
+

a
1
<a
2

u
a
1
a
2
e
a
1
a
2
+

a
1
<a
2
<a
3

u
a
1
a
2
a
3
e
a
1
a
2
a
3
+ . . . (4)
We have
e
a
= e
a
, a = 1, . . . , n, U = U, (UV ) =

U

V , (U + V ) =

U +

V ,
(U) =

U, U, V C(p, q), C.
Reverse. Let dene operation reverse for U C(p, q) in the following way
U

=
n

k=0
(1)
k(k1)
2
k
U
.
3
We have
(e
a
)

= e
a
, a = 1, . . . , n, U

= U, (UV )

= V

U

,
(U + V )

= U

+ V

, (U)

= U

.
Pseudo-Hermitian conjugation. Let dene Pseudo-Hermitian conjuga-
tion as composition of reverse and complex conjugation:
U

=

U

.
We have
(e
a
)

= e
a
, a = 1, . . . , n, U

= U, (UV )

= V

U

,
(U + V )

= U

+ V

, (U)

.
Grade involution. Let dene operation of grade involution U U

in the
following way
U

=
n

k=0
(1)
k
k
U
.
We have
(e
a
)

= e
a
, a = 1, . . . , n, U

= U, (UV )

= U

,
(U + V )

= U

+ V

, (U)

= U

.
Cliord conjugation. Let dene Cliord conjugation as composition of
grade involution and reverse U U

:
U

=
n

k=0
(1)
k(k+1)
2
k
U
.
We have
(e
a
)

= e
a
, a = 1, . . . , n, U

= U, (UV )

= V

,
(U + V )

= U

+ V

, (U)

= U

.
4
Hermitian conjugation. In [2] we consider operation of Hermitian conju-
gation. We have the following formulas for these operation:
U

= (e
1...p
)
1
U

e
1...p
, if p - odd,
U

= (e
1...p
)
1
U

e
1...p
, if p - even, (5)
U

= (e
p+1...n
)
1
U

e
p+1...n
, if q - even,
U

= (e
p+1...n
)
1
U

e
p+1...n
, if q - odd,
We have
(e
a
)

= (e
a
)
1
, a = 1, . . . , n, U

= U, (UV )

= V

,
(U + V )

= U

+ V

, (U)

.
4 Matrix representations of Cliord algebra el-
ements, recurrent method.
Complex Cliord algebras C(p, q) of dimension n and dierent signatures
(p, q), p + q = n are isomorphic. Cliord algebras C(p, q) are isomorphic to
the matrix algebras of complex matrices. In the case of even n these matrices
are of order 2
n
2
. In the case of odd n these matrices are block diagonal of
order 2
n+1
2
with 2 blocks of order 2
n1
2
.
Consider the following matrix representations of Cliord algebra ele-
ments.
Identity element e of Cliord algebra C(p, q) maps to identity matrix of
corresponding order: e 1.
For C(1, 0) element e
1
maps to the following matrix
e
1

_
1 0
0 1
_
.
For C(2, 0) we have
e
1

_
1 0
0 1
_
, e
2

_
0 1
1 0
_
.
Further, suppose we have a matrix representation for C(2k, 0), n = 2k:
e
1
, . . . , e
n

1
, . . . ,
n
.
5
Then, for Cliord algebra C(2k + 1, 0) we have
e
a

_

a
0
0
a
_
, a = 1, . . . , n, e
n+1

_
i
k

1
. . .
n
0
0 i
k

1
. . .
n
_
.
For Cliord algebra C(2k + 2, 0) we have the same matrices for e
a
, a =
1, . . . , n + 1 as in the previous case n = 2k + 1 and for e
n+2
we have
e
n+2

_
0 1
1 0
_
.
So, we have matrix representation for all Cliord algebras C(n, 0). In the
cases of other signatures elements e
a
, a > p maps to the same matrices as in
signature (n, 0) but with multiplication by imaginary unit i.
For example, we have the following matrix representations for Cliord
algebras C(3, 0), C(4, 0) and C(1, 3).
C(3, 0):
e
1

1
=
_
_
_
_
1 0 0 0
0 1 0 0
0 0 1 0
0 0 0 1
_
_
_
_
, e
2

2
=
_
_
_
_
0 1 0 0
1 0 0 0
0 0 0 1
0 0 1 0
_
_
_
_
,
e
3

3
=
_
_
_
_
0 i 0 0
i 0 0 0
0 0 0 i
0 0 i 0
_
_
_
_
.
C(4, 0):
e
1

1
, e
2

2
, e
3

3
, e
4

4
=
_
_
_
_
0 0 1 0
0 0 0 1
1 0 0 0
0 1 0 0
_
_
_
_
.
C(1, 3):
e
1

1
, e
2
i
2
, e
3
i
3
, e
4
i
4
.
6
5 Operation of trace of Cliord algebra ele-
ments
Consider complex Cliord algebra C(p, q) and introduce the operation of
trace of Cliord algebra element U C(p, q) as the following operation of
projection onto subspace C
0
(p, q):
Tr(U) = U
0
|
e1
. (6)
For arbitrary element U C(p, q) in the form (1) we have
Tr(ue + u
a
e
a
+ . . .) = u.
Theorem 1 . Operation trace (6) of Cliord algebra element U C(p, q)
has the following properties:
linearity:
Tr(U + V ) = Tr(U) + Tr(V ), Tr(U) = Tr(U)
U, V C(p, q), C,
cyclic recurrence:
Tr(UV ) = Tr(V U), Tr(UV W) = Tr(V WU) = Tr(WUV )
U, V, W C(p, q),
but, in general:
Tr(UV W) = Tr(UWV ).
invariance under similarity transformation:
Tr(U
1
V U) = Tr(V ) V C(p, q), U C

(p, q),
where C

(p, q) is the set of all invertible Cliord algebra elements.

invariance under conjugations:
Tr(U) = Tr(U

) = Tr(U

) =

Tr(U) =

Tr(U

) =

Tr(U

).
7
Proof. Linearity follows from the denition (6).
We have
Tr(UV ) = Tr(V U)
because Tr([
k
U
,
l
V
]) = 0 for k, l = 0, . . . n (see [2]). Then we obtain cyclic
recurrence for 3 elements. We obtain invariance under similarity transfor-
mation as a simple consequence of cyclic recurrence. Last properties follow
from properties of conjugations.
There is a relation between operation trace Tr of Cliord algebra element
U C(p, q) and operation trace tr of quadratic matrix. To obtain this
relation, at rst, we will prove the following statement.
Lemma 1 . Consider recurrent matrix representation of Cliord algebra
C(p, q) (see above). For this representation U U we have
tr(U) = 2
[
n+1
2
]
Tr(U), tr(U

) = tr(U).
Proof. Coecient 2
[
n+1
2
]
equals to the order of corresponding matrices.
It is not dicult to see that trace of almost all matrices that correspond to
basis elements equals to zero
tr(e
A
) = 0, where A - any multi-index except empty.
The only exception is identity element e, which corresponds to the identity
matrix. In this case we have tr(e) = 2
[
n+1
2
]
. Further we use linearity of trace
and obtain
tr(U) = 2
[
n+1
2
]
u = 2
[
n+1
2
]
Tr(U).
The second property is a simple consequence of the rst property, because
tr(U

) = 2
[
n+1
2
]
Tr(U

) = 2
[
n+1
2
]
Tr(U) = tr(U).

Theorem 2 . Consider complex Cliord algebra C(p, q) and operation trace

Tr. Then
Tr(U) =
1
2
[
n+1
2
]
tr((U)), (7)
where (U) - any matrix representation of Cliord algebra C(p, q) of min-
imal dimension. Moreover, this denition of trace (7) is equivalent to the
denition (6). New denition is well-dened because it doesnt depend on the
choice of matrix representation.
8
Proof. This property proved in the previous statement for the recurrent
matrix representation. Let we have besides recurrent matrix representation
U = U|
e
a

a
another matrix representation
U = U|
e
a

a.
Then, by Paulis theorem in Cliord algebra of even dimension n there exists
matrix T such that

a
= T
1

a
T, a = 1, . . . , n.
Then, we have
U = T
1
UT
and
tr(U) = tr(T
1
UT) = tr(U).
In the case of odd n we can have also another case (by Paulis theorem),
when two sets of matrices relate in the following way

a
= T
1

a
T, a = 1, . . . , n.
In this case we have
U = T
1
U

T.
From tr(U

) = tr(U) (see Lemma 1) we obtain

tr(U) = tr(T
1
U

T) = tr(U

) = tr(U).

6 Determinant of Cliord algebra elements

Determinant of Cliord algebra element U C(p, q) is a complex number
DetU = det(U), (8)
which is a determinant of any matrix representation U of minimal dimension.
Now we want to show that this denition is well-dened. Let prove the
following Lemma.
9
Lemma 2 . Consider the recurrent matrix representation (see above) of
Cliord algebra C(p, q). For this representation U U we have
det(U

) = det(U).
Proof. In the case of Cliord algebra of even dimension n we have
U

= (e
1...n
)
1
Ue
1...n
.
So, we obtain
det(U

) = det((e
1...n
)
1
Ue
1...n
) = det((e
1...n
)
1
)det(U)det(e
1...n
) = det(U).
In the case of Cliord algebra of odd dimension generators maps to the
block diagonal matrices and blocks are identical up to the sign:
e
a

_

a
0
0
a
_
.
Then for elements of the rank 2 we obtain
e
ab

_

a

b
0
0
a

b
_
.
It is not dicult to see that even part U
Even
of arbitrary element U = U
Even
+
U
Odd
maps to the matrix with identical blocks, and odd part U
Odd
of the
element U maps to the matrix with the blocks diering in sign:
U
Even

_
A 0
0 A
_
, U
Odd

_
B 0
0 B
_
.
Then we have
U
_
A + B 0
0 AB
_
, U

_
AB 0
0 A + B
_
and
det(U) = (AB)(A+ B) = det(U

).

Theorem 3 . Denition (8) is well-dened, i.e. it doesnt depend on the

matrix representation.
10
Proof. Consider the recurrent matrix representation
U = U|
e
a

a.
The statement for this representation proved in the previous lemma. Let we
have another matrix representation
U = U|
e
a

a.
Then, by Paulis theorem in Cliord algebra of even dimension n there exists
a matrix T such that

a
= T
1

a
T, a = 1, . . . , n.
Then we have
U = T
1
UT
and obtain
det(U) = det(T
1
UT) = det(T
1
)det(U)det(T) = det(U).
In the case of odd n, by Paulis theorem we also have another case, where
2 sets of matrices relate in the following way

a
= T
1

a
T, a = 1, . . . , n.
In this case we have
U = T
1
U

T.
From det(U

) = det(U) (see Lemma 2) we obtain

det(U) = det(T
1
U

T) = det(U

) = det(U).

Let formulate some properties of operation determinant of Cliord algebra

element.
Theorem 4 . Operation determinant (8) of Cliord algebra element U
C(p, q) has the following properties

Det(UV ) = Det(U)Det(V ), Det(U) =

2
[
n+1
2
]
Det(U) (9)
U, V C(p, q), C.
11
Arbitrary element U C(p, q) is invertible if and only if DetU = 0.
For any invertible element U C(p, q)
Det(U
1
) = (DetU)
1
. (10)
Invariance under similarity transformation:
Det(U
1
V U) = Det(V ) V C(p, q), U C

(p, q),
where C

(p, q) is set of all invertible Cliord algebra elements.

Invariance under conjugations:
Det(U) = Det(U

) = Det(U

) =

Det(U) =

Det(U

) =

Det(U

).
Proof. The rst 4 properties are simple and follow from the denition of
determinant (8).
In lemma 2 we have the property Det(U) = Det(U

) for the recurrent

matrix representation. But it is also valid for other matrix representations
because of independence on the choice of representations (see Theorem 3).
It is known that operation relates to the operation of matrix transpose
(as similarity transformation) and also we have det(U) = det(U
T
). Analo-
gously we can consider another operations of conjugation.
Denition (8) of determinant of Cliord algebra element U C(p, q) is
connected with its matrix representation. We have shown that this denition
doesnt depend on matrix representation. So, determinant is a function of
complex coecients u
a
1
...a
k
located before basis elements e
a
1
...a
k
in (1). In
the cases of small dimensions n 5 we give expressions for determinant of
Cliord algebra elements that doesnt relate to the matrix representation.
Now we need also 2 another operations of conjugations , :
(
0
U
+
1
U
+
2
U
+
3
U
+
4
U
)

=
0
U
+
1
U
+
2
U
+
3
U

4
U
, n = 4,
(
0
U
+
1
U
+
2
U
+
3
U
+
4
U
+
5
U
)

=
0
U
+
1
U
+
2
U
+
3
U

4
U

5
U
, n = 5,
(
0
U
+
1
U
+
2
U
+
3
U
+
4
U
+
5
U
)

=
0
U
+
1
U
+
2
U
+
3
U
+
4
U

5
U
, n = 5.
12
Theorem 5 . We have the following formulas for the determinant of Cliord
algebra element U C(p, q):
Det U =
_

_
U, n = 0;
UU

, n = 1;
UU

, n = 2;
UU

= UU

, n = 3;
UU

(U

= UU

(U

, n = 4;
UU

(U

(UU

(U

, n = 5.
(11)
Note, that these expressions are Cliord algebra elements of the rank 0.
In this case we identify them with the constants: ue u.
Proof. The proof is by direct calculation. .
Note, that properties (9) and (10) for small dimensions also can be proved
with the formulas from Theorem 5. For example, in the case n = 3 we have
Det(UV ) = UV (UV )

(UV )

(UV )

= UV V

U

=
= UU

V V

V

V

= Det(U)Det(V )
We used the fact that V V

and V

V

= (V V

)

are in Cliord algebra

center C
0
(p, q) C
3
(p, q) and commute with all elements.
Theorem 5 give us explicit formulas for inverse in C(p, q). We have the
following theorem.
Theorem 6 . Let U be invertible element of Cliord algebra C(p, q). Then
we have the following expressions for U
1
:
(U)
1
=
_

_
e
U
, n = 0;
U

UU

, n = 1;
U

UU

, n = 2;
U

UU

=
U

UU

, n = 3;
U

(U

UU

(U

=
U

(U

UU

(U

, n = 4;
U

(U

(UU

(U

UU

(U

(UU

(U

, n = 5.
(12)
13
Note, that in denominators we have Cliord algebra elements of the rank
0. We identify them with the constants: ue u.
Proof. Statement follows from Theorem 5.
Note, that formulas for determinant in Theorem 5 are not unique. For
example, in the case of n = 4 we can use the following formulas, but only for
even and odd Cliord algebra elements U C
Even
(p, q) C
Odd
(p, q).
Consider operation +, that acts on the even elements such that it changes
the sign before the basis elements that anticommutes with e
1
. For exam-
ple, elements e, e
23
, e
24
, e
34
maps under + into themselves, and elements
e
12
, e
13
, e
14
, e
1234
change the sign.
Theorem 7 . Let U C
Even
(p, q), n = p + q = 4. Then
DetU = UU

U
+
U
+
. (13)
Let U C
Odd
(p, q), n = p + q = 4. Then
DetU = UU

U. (14)
Proof. The proof is by direct calculation.
References
[1] Lounesto P., Cliord Algebras and Spinors, Cambridge Univ. Press
(1997, 2001).
[2] Marchuk N.G., Shirokov D.S., Unitary spaces on Cliord algebras, Ad-
vances in Applied Cliord Algebras, Volume 18, Number 2 (2008).
[3] Marchuk N.G., Martynova S.E., Notions of determinant, spec-
trum and Hermitian conjugation of Cliord algebra elements,
arXiv:math-ph/0307043 (2008).
[4] P. Dadbeh, Inverse and determinant in 0 to 5 dimensional Cliord alge-
bra, arXiv:1104.0067v1 [math-ph] (2011).
14