Advances in Applied Clifford Algebras Is

You might also like

Download as pdf or txt
Download as pdf or txt
You are on page 1of 28

Adv. appl. Clifford alg. 19 (2009), No.

3–4, 793–817
c 2009 Birkhäuser Verlag Basel/Switzerland Advances in
DOI 10.1007/s00006-003-0000 Applied Clifford Algebras

Isometry from Reflections


Versus Isometry from Bivector
Zbigniew Oziewicz

Abstract. Unipotent isometry is said to be a reflection. In 1937 Élie Cartan


proved that every isometry can be expressed as a composition of reflections.
The Lie subalgebra of bivectors inside a Clifford algebra generate isometries
without the Grassmann exponential. The main result is the coordinate-free
and basis-free construction of two isometries from a simple bivector. Hestenes
introduced in 1966 a rotor in a Spin group as a square root of the Clifford
product of two vectors. We compare a rotor from reflections, with a rotor
from a simple bivector. This result generalizes the Lorentz transformations
considered by Pertti Lounesto in 1997.
Mathematics Subject Classification (2000). 15A66 Clifford algebras, spinors,
15A75 Exterior algebra, Grassmann algebra, 20L05 Groupoids viewed as a
category in which all morphisms are isomorphisms, 53A35 Non-Euclidean
geometry.
Keywords. Isometry, groupoid viewed as a category, the Hestenes rotor = the
Hestenes spinor.

Contents
1. Introduction 794
2. Duality and Derivation 795
3. Metric Tensor as non-natural Isomorphism 797
4. Grassmann -Hodge stars 799
4.1. Simple bivector 801
5. Isometry is basis-free 802
5.1. Isometry from single simple endomorphism must be a reflection 803
6. Homogeneous manifold as a groupoid category 805
7. Isometry from Bivector 809
8. Proof of the main theorem 30 812
The last Section 11 was removed by Editor.
794 Zbigniew Oziewicz AACA

9. Challenge: isometry from trivector 813


10. Comparison of rotors 814
11. Public Referee. Section removed by Editor. 815
Acknowledgements 817
References 818

1. Introduction
A unipotent isometry is said to be a reflection. In 1937 Élie Joseph Cartan
proved that every isometry can be expressed as a compositions of reflections. This
statement is known as the Cartan-Dieudonné theorem, because Dieudonné pro-
vided a shorter proof [Cartan 1937, Dieudonné 1948, Artin 1957]. Within Clifford
algebra a reflection can be expressed in terms of the Clifford multiplication of vec-
tors, Z 7−→ −XZX −1 , X being non-isotropic. In particular, the composition of
two reflections, Z 7−→ (XY )Z(XY )−1 , give rise to the concept of a rotor intro-
duced by David Hestenes in√ 1966, as the square root of the Clifford product of two
non-isotropic vectors, R = XY , being an element of the even part of the Spin
group inside of the Clifford subalgebra of the even multivectors, [Hestenes 1966].
Bivectors inside the Clifford algebra form a Lie subalgebra, and generate
isometries without using reflections [Cartan 1937, Section 19]. Our main result
is the coordinate-free and basis-free construction of two isometries from a sim-
ple bivector. This allows considerations of another rotor, a rotor from a simple
bivector. This construction generalizes the Lorentz transformations considered by
Lounesto in four dimension with a specific Lorentz signature [Lounesto 1997, pages
125–127]. We are comparing a rotor from a bivector versus a rotor from the com-
position of two reflections.
In Sections 2-3-4 we recall some technical details about Grassmann algebra,
that are necessary for understanding the main part of this paper in Section 7.
In Section 2 we discuss the convention relating graded duality of multivectors &
multiforms, to graded derivations in Grassmann algebra.
In Section 4 we interpret a Grassmann multivector as a morphism of K-
modules, for K being a ground ring, as a morphism from multiforms to multivec-
tors, i.e. the push-forward covariant functor sends a multivector to this morphism.
Analogously a multiform is interpreted as a morphism from a K-module of multi-
vectors to a K-module of multiforms. Such a morphism need not be invertible, and
in the case of a simple (decomposable) bivector it is, for example, a non-invertible
morphism from one-forms to one-vectors (a one-vector within Grassmann algebra
is a synonym of a vector). We show that the non-null simple bivectors possess
an ‘inner’ Riemannian inverse and therefore such bivectors are regular, accord-
ing to terminology introduced in [Nashed 1976]. This regularity is reflexive, and
resembles the Galois relation (named by Oystein Ore).
Rotor of a Simple Bivector 795

This paper was written in May 2007, rejected in February 2008, and miracu-
lously accepted in July 2008 without appeal. In the mean time this journal AACA
published at least one paper on a related subject, by Aragón et al., Reflections,
rotations, and Pythagorean numbers, submitted in April 2008 and accepted in June
2008, [Argón González et al. 2009]. Therefore there is a lot of overlap between these
papers, that made publishing the present paper almost superfluous. Aragón at al.
are restricted to the Euclidean space Rn (possessing the canonical basis), whereas
we stress basis-free approach, including non-Euclidean geometry of arbitrary signa-
ture. Moreover, Aragón et al. studied Pythagorean triples and Pythagorean boxes
that are outside of the scope of our paper. However the Pythagorean subject was
published in 2006 in AACA by Jerzy Kocik who observed that the Euclid parame-
terization of the Pythagorean triples (rediscovered independently by Aragón et al.
in 2009) is the earliest appearance of the concept of a spinor in the Clifford algebra
of three-dimensional Minkowski space, Cℓ(2, 1) (Lorentzian, pseudo-Euclidean or
non-Euclidean) [Kocik 2006-2007].

2. Duality and Derivation


Throughout this paper, K is a unital associative and commutative ring, and E
is a finite-dimensional K-module of vectors. A dual K-module of forms is denoted
by E ∗ . The Grassmann Hopf K-algebras are denoted by E ∧ and E ∗∧ . The under-
lying K-module of multivectors, an image of the forgetful functor from K-algebras
to K-modules, is denoted by |E ∧ |.
The aim of this Section is to recall the rather well known fact that an iso-
morphism E ∗ ≃ E does not determine the isomorphism between the following
N-graded K-modules,
|E ∧ |∗ ≃ |E ∗∧ |, and |E ∗∧ |∗ ≃ |E ∧ |. (1)
On the other hand, a choice of a non-degenerate N-graded pairing (duality),
|E ∗∧ | ⊗K |E ∧ | −→ K, (2)
gives an injection of a K-module |E ∗∧ | into a dual K-module |E ∧ |∗ , and vice-versa,
|E ∗∧ | ֒→ |E ∧ |∗ ≃ |E ∗∧ |. (3)

We discuss here the freedom of choice of the isomorphisms above (1). The
conventions vary from author to author, and sometimes it is not realized explicitly
that there is a freedom of choice. Different conventions lead to different expressions
that could be embarrassing.
The Grassmann K-algebra E ∧ is noncommutative, and therefore there are
left and right ‘exterior’ Grassmann products, known as the regular/adjoint left L,
and right R representations of the Grassmann K-algebra E ∧ in the endomorphism
K-algebra End(|E ∧ |). In fact L and R are inner Grassmann actions within the
796 Zbigniew Oziewicz AACA

same Grassmann algebra. Let X and Y ∈ E ∧ be multi-vectors, then


L, R
E ∧ −−−−→ End |E ∧ |, LX Y ≡ X ∧ Y = RY X. (4)
In (4), L is an algebra map, and R is the anti-algebra map respectively (or algebra
map for the opposite Grassmann algebra).
The pull-backs, R∗ & L∗ , (destructors) are the exterior actions of a Grass-
mann algebra on the underlying K-module of its dual, see Figure 1.
L,R
E∧ / End (|E ∧ | ≃ |E ∗∧ |∗ )
5
g∧ R∗ R∗
 )
E ∗∧ L,R
/ End (|E ∗∧ | ≃ |E ∧ |∗ )

Figure 1. Creators R, and destructors R∗ .

Note that if R is an anti-algebra map, then its pull-back R∗ is an algebra


map, and vice-versa, (7).
If α ∈ |E ∧ |∗ , then α(LX Y ) = (L∗X α)Y ∈ K, (5)
∗ ∗
L and R
E ∧ −−−−−−−−−−→ End(|E ∧ |∗ ), (6)
L∗X∧Y ≡ (LX∧Y )∗ = (LX ◦ LY )∗ = L∗Y ◦ L∗X ,
∗ (7)
RX∧Y ≡ (RX∧Y )∗ = (RY ◦ RX )∗ = RX∗
◦ RY∗ .
Here the algebra map, (6)-(7), needs a priori a choice of K-module-isomorphisms
(1)-(3). One can express the choice of K-module isomorphisms, (1)-(3), by means
of the graded derivations of the Grassmann algebra as follows.
1. Definition (Graded derivation). Every vector X ∈ E, extends to left and right
graded derivations of the Grassmann algebra of multiforms, iL R
X , iX ∈ der(E
∗∧
).
The notation ‘i’, for ‘internal’ or ‘inner’, is misleading because in fact it is an
exterior action of vectors on multi-forms. For all homogeneous multi-forms, α and
β ∈ E ∗∧ , we have by definition

iL L α L
X (α ∧ β) ≡ (iX α) ∧ β + (−1) α ∧ iX β,

iR R β R
X (α ∧ β) ≡ α ∧ iX β + (−1) (iX α) ∧ β,

iL
X ◦ L α = L iL

+ (−1)α Lα ◦ iL
X,
(8)
iL L
X ◦ R β = R β ◦ i X + R iL

◦ (−1)grade ,
iR
X ◦ L α = L iR

◦ (−1)grade + Lα ◦ iR
X,

iR β R
X ◦ Rβ = (−1) Rβ ◦ iX + RiR

.
The Z-graded commutator of graded derivations is a graded derivation, there-
fore the K-module of graded derivations is a super Z2 -Lie K-module.
Rotor of a Simple Bivector 797

2. Convention. An isomorphism (1), and an injection (3) can be fixed by the


following choice of assignment,
i
E ∗∧ −−−−→ End(|E ∧ |)
i|K is a K-module structure of |E ∧ |,
iα∧β ≡ iα ◦ iβ , and i|E ∗ is a left derivation of E ∧ . (9)
Then, for each α ∈ E ∗∧ , iα = Rα

, and a graded pairing (duality) is given by,
(E ∗∧ ) ⊗K (E ∧ ) ∋ α ⊗ X −→ αX ≡ iX α = iα X ∈ K. (10)
The choice of ‘inner’ homomorphism ‘i’ from the Grassmann algebra, to en-
domorphism algebra of the dual Grassmann algebra, (9), fixes the duality pairing
between multiforms and multivectors. Another option to extend the graded deriva-
tion, iL or iR , from vectors to multivectors, is by the anti-homomorphism from
Grassmann algebra to endomorphism algebra. The anti-homomorphism choice or
convention is suggested by determinants and permanents, as e.g. in [Oziewicz
1986], however this choice is not the only one. A given pairing, implies, i = R∗ or
i = L∗ , [Cartan 1922; Greub 1958, 1967, 1978].
For example, for one-forms, α and β ∈ E ∗ , and for one-vectors, X, Y ∈ E,
we can chose the following convention,

(α ∧ β)(X ∧ Y ) = iX∧Y (α ∧ β)
= (iX ◦ iY )(α ∧ β) = iX {(αY )β − (βY )α}
= (βX)(αY ) − (αX)(βY ). (11)
In what follows, i ≡ R∗ . Moreover we assume, as in (9), that, i|E and i|E ∗ ,
are graded left derivations of Grassmann algebras. Then, for every Grassmann
multiform, β ∈ E ∗∧ , Z ∈ E ∧ , and for each vector X ∈ E, and each form α ∈ E ∗ ,
the following commutation formulas hold
iβ ◦ RX = RX ◦ iβ + iiX β ◦ (−1)grade ,
(12)
iZ ◦ Rα = Rα ◦ iZ + iiα Z ◦ (−1)grade .

3. Metric Tensor as non-natural Isomorphism


The usual practice is to denote the square of the magnitude of a vector X ∈ E,
or the square of the magnitude of a form α ∈ E ∗ , by X 2 , and by α2 ∈ K. This
could made incorrect impression that the magnitude is a property of a vector or
covector only. In fact, the square of the magnitude of a vector,
X 2 ≡ X · X ≡ g(X ⊗ X) ≡ igX X ∈ K, (13)
is a function of two independent tensorial variables. It depends on the vector X
and on the choice of a metric tensor g, a free choice for the non-natural, or non-
canonical, scalar product.
798 Zbigniew Oziewicz AACA

A non-degenerate (invertible) metric-tensor is non-natural isomorphism, that


can be chosen to be symmetric, and is denoted as follows, [Kocik 1989, Chapter
0.3, pages 12-13],
g = g ∗ ∈ isoK (E, E ∗ ) ≃ E ∗ ⊗K E ∗ , g −1 ∈ isoK (E ∗ , E) ≃ E ⊗K E. (14)
A metric-tensor usually is interpreted as the scalar-valued ‘product’ on ‘pairs’
g
of vectors, E ⊗2 −→ K, and by abuse of notation, we use the same symbol for
g ∈ isoK (E, E ∗ ). One can pass from one to the other interpretation by means of
evaluation and coevaluation in a bi-closed category.
g
+
E j E ∗ ≃ (E 7→ K)
g −1
Figure 2. Invertible scalar product as an isomorphism.

A non-degenerate scalar product considered as an isomorphism from vectors


to covectors, g ∈ isoK (E, E ∗ ), is a genuine invertible non-natural isomorphism,
Figure 2.
Metric tensors, g and g −1 , lift to the isomorphisms, g ⊗ , g ∧ and g Cℓ , of tensor
algebras, Grassmann algebras, and Clifford algebras, Figure 3,

⊗ 1 g ⊗ ∈ homK (E ⊗ , E ∗⊗ )


homK (E, E ∗ ) ∋ g / g ∧ ∈ homK (E ∧ , E ∗∧ )

Cℓ
, 
g Cℓ ∈ homK Cℓ(E, g), Cℓ(E ∗ , g −1 )
Figure 3. Three functors.

3. Example.
g ⊗ g −1 = g ◦ g −1 ◦ g = g ∈ homK (E, E ∗ ),
(15)
(g −1 )⊗ g = g −1 ◦ g ◦ g −1 = g −1 ∈ homK (E ∗ , E).
Similarly, if A ∈ EndK E, then, A⊗ and A∧ , are homomorphisms of the tensor
algebra and Grassmann algebra, correspondingly. For X, Y ∈ E ∧ , and α ∈ E ∗∧ ,
we have
iX ◦ g ∧ = g ∧ ◦ ig ∧ X , g ∧ ◦ iα = ig−1∧ α ◦ g ∧ , (16)
2 2 2
(X ∧ Y ) ≡ ig∧ (X∧Y ) (X ∧ Y ) = (Y · X)(X · Y ) − X Y . (17)
Secret Referee considers that (16) is erroneous.
Rotor of a Simple Bivector 799

4. Grassmann -Hodge stars


Similarly to the metric tensor, g ∈ iso(E, E ∗ ), also every bivector, b ∈ E ∧2 ,
can be interpreted as a morphism from forms to vectors, and this is denoted by
‘push-forward’, b∗ ∈ hom(E ∗ , E) (like a pre-Poisson structure), or, more generally
as a K-module morphism, b∗ ∈ hom(|E ∗∧ |, |E ∧ |).
In what follows it is convenient to use the following push-forward notation.
For every Grassmann multiform, α ∈ E ∗∧ , and for every multivector, X ∈ E ∧ ,
one can define the following morphisms of K-modules, see Figure 4,
α∗ X ≡ iX α, X∗ α ≡ iα X. (18)

‘symplectic’
α∗ *
|E ∧ | j |E ∗∧ |
X∗
‘Poisson’

Figure 4. Push-forward of a multiform, and of a multi-vector,


as a morphism of K-modules.

The above push-forward of Grassmann multi-vectors and the push-forward


of multi-forms, generalize the Weyl isomorphism (also called the Poincaré isomor-
phism), that holds for the volume forms and volume vectors of the highest grade.
Note that for a bivector, b ∈ E ∧2 (a kind of ‘pre-Poisson’ structure), and for a
1-form α ∈ E ∗ (a ‘hamiltonian’ form), a vector b∗ α ∈ E, looks like a ‘hamiltonian’
vector in the case that a bivector b gives a bijection b∗ from E ∗ to E, i.e. if
b∗ ∈ iso(E ∗ , E).
4. Lemma. Let, X ∈ E ∧ , be a homogeneous multivector. The pull-back of a mor-
phism is as follows,
X∗ ∈ homK (|E ∗∧ |, |E ∧ |) =⇒ (X∗ )∗ = X∗ · (−1)(1+X)·grade . (19)
5. Corollary. Let grade X = odd, for a homogeneous multivector X ∈ E ∧ . A
‘Poisson’ morphism X∗ ∈ homK (|E ∗∧ |, |E ∧ |) is symmetric (X∗ )∗ = X∗ .
6. Grassmann - Hodge stars (Kocik 1979). The compositions, g ∧ ◦X∗ , and X∗ ◦g ∧ ,
are generalized ‘Hodge stars’ that are not necessarily invertible,
g∧ X α g −1∧
E∧ −−−−→ E ∗∧ −−−−

→ E∧, |E ∧ | −−−∗−→ E ∗∧ −−−−→ E ∧ ,
(20)
X∗ g∧ g −1∧ α
|E ∗∧ | −−−− → E ∧ −−−−→ E ∗∧ , E ∗∧ −−−−→ E ∧ −−−∗−→ E ∗∧ .
The concept of a ‘Hodge-star’ was introduced by Hermann Grassmann in his
second monograph in 1862, under the name ‘Ergänzung’ (supplement) [Grass-
mann 1862, §3.4]. The present-day star-notation, ∗, was introduced by Hermann
Weyl in 1945. Prior to 1980 Jerzy Kocik observed that the symmetry of the
metric tensor is not an obligatory condition for the Hodge star, i.e. g ∗ 6= g,
800 Zbigniew Oziewicz AACA

leads to a more general concept. We refer to the WEB page by Jerzy Kocik,
http://www.math.siu.edu/jkocik.htm.
7. Corollary. Let b ∈ E ∧2 ≡ E ∧ E, be a bivector. Then, b∗∗ |E ∗ = −b∗ . An
endomorphism, eg b ≡ b∗ ◦ g ∈ EndK E, is g-skew-symmetric, and trace-less,
g
E −−−−→ E ∗ −−−∗−→ E,
b (b∗ ◦ g)∗ |E = g ◦ b∗∗ = −g ◦ b∗ ,
g ◦ (b∗ ◦ g) = −(b∗ ◦ g)∗ ◦ g, =⇒ tr(b∗ ◦ g) = gb = 0 ∈ K.
For a homogeneous multivector, X ∈ E ∧ , a scalar X 2 ∈ K, is defined by
2
X ≡ igX X = X∗ (gX) = iX gX, (16), and therefore depends on the choice of the
inner product ‘i’ (the choice of the duality in Convention 2).
8. Lemma. For a homogeneous multivector, X ∈ E ∧ , that does not need to be
decomposable, we have
(g ∧ ◦ X∗ )2 |E ∗ = iX ◦ Rg∧ X − (−1)X X 2 · id .
(21)
(X∗ ◦ g ∧ )2 |E = ig∧ X ◦ RX − (−1)X X 2 · id .
Proof. One needs to use the commutation (16), and then the commutations (12).

9. Lemma. Let X ∈ E ∧ be a multivector. Then, g ∧ X ∈ E ∗∧ , is a multiform. The
following identity holds
(g ∧ X)∗ = g ∧ ◦ X∗ ◦ g ∧ : E ∧ −→ E ∗∧ . (22)
10. Notation (Clifford algebra). The presentation of Clifford algebra in terms of
generators and relations, Cℓ(E, g), seems to be the most popular definition of the
Clifford algebra [Porteous 1969, 1995; Riesz 1993]. This presentation is not unique,
and is not intrinsic.
In my lecture presented at the 7th Conference on Clifford Algebras in Toulouse
in 2005 I proposed to view presentation-free Clifford algebra, as a particular in-
stance of Frobenius algebra. Such an axiomatic definition avoids the choice of
generating space – there is no need to see Clifford algebra as a factor algebra or
as the Chevalley deformation of an N-graded Grassmann algebra. The grading
of multivectors is of lesser or no importance and one can see very explicitly the
intrinsic global properties of the Clifford binary operation and cooperation (Clif-
ford coalgebra) within the axioms of Frobenius algebra. Such a point of view is
necessary when we are not interested in selecting a generating privileged subspace
and whenever only the entire Clifford algebra is important as a generator-free and
presentation-free algebra structure. Secret Referee was not happy with this idea
and gently suggested to me that ‘the author must assume the well known definition
of Clifford algebra, and must not invent new a one.’
Forgetting for awhile Frobenius algebra, Clifford algebra can be seen as a K-
module |E ∧ | with a binary associative multiplication denoted by γ, γ|E ≡ L+L∗ ◦g,
given in terms of the Chevalley g-deformation, where a metric tensor g is considered
to be the deforming tensor, [Chevalley 1954, Oziewicz 1986, 1997].
Rotor of a Simple Bivector 801

Sometimes, for simplicity, the Clifford multiplication is denoted by juxta-


position. Let b denote a bivector, and let X and Y be one-vectors. Then, XY,
bXb, bb, etc, denote examples of Clifford products, and these a priori need not be
homogeneous Grassmann multivectors. For example, following Chevalley,

XY = X · Y + X ∧ Y.

11. Lemma. Let X ∈ E, and b ∈ E ∧2 . Moreover let b2 ≡ ig∧ b b ∈ K. The following


identity holds,

2ig∧ b (X ∧ b) = −b2 X − bXb. (23)

4.1. Simple bivector


From now on 0 6= b ∈ E ∧2 stands for a non-zero simple (decomposable)
bivector, b ∧ b = 0 ∈ E ∧4 . Then a kernel of the push-forward b∗ |E ∗ consists of all
forms associated to the plane of b (or all vectors orthogonal to this plane.) In the
sequel b∗ ∈ homK (E ∗ , E).

12. Lemma. Let b ∈ E ∧2 ⊂ Cℓ(E, g), be a simple bivector. The Clifford product of b
with b, denoted by juxtaposition, bb, is equal to a scalar, b2 ≡ ig∧ b b = b∗ (g ∧ b) ∈ K.

Although b∗ , is a non-invertible morphism, it does nonetheless possess a ‘gen-


eralized inverse’ as follows.

13. Theorem. Let b ∈ E ∧2 ⊂ Cℓ(E, g), be a simple bivector, and let, b∗ ≡ b∗ |E ∗ .


The following reflexive identities hold

b∗ ◦ (g ∧ b)∗ ◦ b∗ = b2 b∗ ,
(24)
(g ∧ b)∗ ◦ b∗ ◦ (g ∧ b)∗ = b2 (g ∧ b)∗ .

This means that a morphism (g ∧ b)∗ /b2 is a generalized (inner) inverse of b∗ ,


and b∗ /b2 is an inner inverse of (g ∧ b)∗ . The terminology of regular morphism, and
inner inverse, were introduced in [Nashed 1976].
The above theorem can be rephrased in terms of the minimal polynomial,
and this holds for any dimension, and for an arbitrary signature of the invertible
metric tensor g ∈ homK (E, E ∗ ). If b is a simple bivector, then there are the
following polynomials, that is analogous to the matrix problem in [Jackson 1962,
1999, Problem 11.8],

(b∗ ◦ g)3 = b2 (b∗ ◦ g) ∈ EndK E, (g ◦ b∗ )3 = b2 (g ◦ b∗ ) ∈ EndK E ∗ . (25)

Note that the reflexive relations (24) reminds one of the relation of the meson
algebra, aba = f (a ⊗ b)a, for a symmetric form, f (a ⊗ b) = f (b ⊗ a). For meson
algebra we refer to [Micali and Rachidi 2008].
802 Zbigniew Oziewicz AACA

5. Isometry is basis-free
Textbooks frequently define an isometry as a basis-dependent matrix. For ex-
ample, the particular isometry given by a particular Lorentz boost is frequently
defined as the following matrix,
 
γ −(v/c)γ 0 0
−(v/c)γ 0 0
Lorentz boost ≡ 
γ , γ ≡ q 1 . (26)
 0 0 1 0 2
1 − vc2
0 0 0 1
trace(Lorentz boost) = 2 + 2γ. (27)
In this section we wish to recall and clarify that the concept of an isometry
is basis-free. Bases, frames and coordinates obscure mathematical and physical
ideas, introducing irrelevant arbitrariness that has nothing to do with the sub-
ject. In particular, a vector is basis-free, so it would be strange that an isometry
can not be understood as the basis-free concept. A Lorentz transformation is an
isometry therefore understanding isometry as a basis-free concept will help to re-
understand special relativity based on the postulate that every permutation of
reference systems must be an isometry.
To understand isometry, it is good to be familiar with the endomorphism
algebra, EndK E ≃ E ⊗K E ∗ . Let X ∈ E ≃ hom(K, E), and α ∈ E ∗ ≃ hom(E, K).
Then, X ⊗ α ≃ X ◦ α ∈ End E, is said to be a simple endomorphism. Pull-backs
give, X ∗ ∈ hom(E ∗ , K ∗ ≃ K), and α∗ ∈ hom(K ∗ , E ∗ ). Therefore, (X ⊗ α)∗ =
α∗ ⊗ X ∗ ∈ End E ∗ .

X⊗α α∗ ⊗X ∗
E 7/ E EO ∗ o EO ∗
α X g α∗ g∗
  v X∗
K / E∗ K∗ ≃ K o E
(gX)∗ =g ◦X gX=X ∗ ◦ g ∗

g ◦ (X ⊗ α) = (gX)∗ ⊗ α ∈ hom(E, E ∗ ),
(28)
(α∗ ⊗ X ∗ ) ◦ g ∗ = α∗ ⊗ (gX).
Figure 5. Commuting diagrams.

By an abuse of notation we identify both interpretations, X ∗ ≃ X and α∗ ≃


α, see Figure 5.
14. Definition (Isometry). An endomorphism L of K-module E, L ∈ EndK E ≃
E ⊗K E ∗ , is said to be a g-isometry if the scalar product, g ∈ hom(E ⊗2 , K), is
invariant
∀ X, Y ∈ E, (LX) · (LY ) = X · Y = g((LX) ⊗ (LY )) ∈ K. (29)
Rotor of a Simple Bivector 803

Invariance of a morphism, g ∈ hom(E, E ∗ ), means that


∀ X, Y ∈ E, (gLX)(LY ) = (gX)Y ⇐⇒ L∗ ◦ g ◦ L = g. (30)
Every isometry can be parameterized in terms of the sum of simple endomor-
phisms, where αi ∈ E ∗ , and, Xi ∈ E,
X X
L = idE − Xi ⊗ α i , tr L = (dim E) − α i Xi , (31)
X1 ∧ X2 ∧ . . . 6= 0, α1 ∧ α2 ∧ . . . 6= 0, g(X1 ∧ X2 ∧ . . .) 6= 0. (32)
Note that in (31)-(32), the summation need not necessarily extend to dim E, and
therefore the set of vectors {Xi } is not presupposed to be a basis.
15. Proposition. The isometry (30) together with ansatz (31)-(32), implies the
following condition,
X X X
(Xi · Xj ) αj ⊗ αi = (gXi ) ⊗ αi + αi ⊗ (g ∗ Xi ), (33)
∀ i, (gXi ) ∧ α1 ∧ α2 ∧ . . . = 0, and (g ∗ Xi ) ∧ α1 ∧ α2 ∧ . . . = 0. (34)

5.1. Isometry from single simple endomorphism must be a reflection


Consider just one simple endomorphism,
L = idE −X ⊗ α =⇒ L2 = (2 − αX)L − (1 − αX) id . (35)
The isometry condition (33)-(34),
X 2 α ⊗ α = (gX) ⊗ α + α ⊗ (g ∗ X), (36)

gives the following unique solution for g = g,
(gX) ∧ α = 0 =⇒ 2gX = X 2 α 6= 0, αX = 2, (37)
X ⊗ gX
LX = id −2 ∈ Aut(g), L2X = id, tr LX = (dim E) − 2. (38)
X2
(Alternative solution, αX = 0, needs (g ∗ + g)X = 0, and implies (L − id)2 = 0.)
A unipotent isometry is said to be a reflection. In other words: a g-automorphism
(isometry), L ∈ Aut(E, g) ⊂ EndK E, is said to be a reflection, iff it is unipotent,
L2 = idE .
16. Corollary. Every non light-like (non-isotropic) vector X ∈ E, is invertible in
Clifford algebra, X −1 = X/X 2 , and determines a reflection with a fixed hyperplane
ker gX,
LX ≡ id −2 X ⊗ gX −1 ∈ Aut(E, g), LX X = −X. (39)
There is a 1:1 correspondence between idempotents and unipotents,
   
idempotent, f 2 = f unipotent, u2 = id
⇐⇒ (40)
f = 12 (id −u) u = id −2f
The commutator of the Clifford product gives Clifford Lie algebra. In 1996
Bernd Schmeikal studied non-homogeneous unipotents, idempotents, and nilpo-
tents in Clifford algebra and discovered the Lie algebra of the standard model,
804 Zbigniew Oziewicz AACA

su(3), within the 16-dimensional real Dirac-Clifford algebra, concluding that the
standart model of elementary particles is a property of the Clifford algebra of
Minkowski spacetime, [Schmeikal 2005, 2006, 2007, 2009].
Within Clifford algebra, reflection (38) satisfies the following identity, e.g.
[Hestenes and Sobczyk 1984, 1986, §3.8, page 103; Aragón et al. 2009, Lemma 8],
∀ Z ∈ E, LX Z = −XZX −1 . (41)
Every isometry can be presented as a composition of reflections [Cartan 1937;
Dieudonné 1948; Artin 1957, pages 129-130; Urbantke 2003; Kocik 2007]. The
identity (41), leads to the following definition of the Clifford group: the Clifford
group is the subgroup of all invertible elements of the Clifford algebra, a ∈ Cℓ(E, g),
such that, aEa−1 = E.
Within a Clifford algebra the composition of two reflections gives rise to the
concept of a rotor ∈ (Spin group), [Hestenes 1966, Section 17; Hestenes & Sobczyk
1984, 1986; Hestenes 1986, Chapter 5.3],
LX LY Z = (XY )Z(XY )−1 . (42)
17. Definition (Hestenes rotor). Let X and Y ∈ E. A rotor of the Clifford product
XY, is a Lipschitzian element of the Clifford algebra, R(XY ) ∈ Cℓ(E, g), defined
by the following condition, [Hestenes 1966],
{R(XY )}2 = XY. (43)
18. Lemma. Let X and Y ∈ E be vectors, such that
X 2Y 2 = 1 and 0 < 1 + X · Y. (44)
Then, up to ±, the rotor is given by the following expression,
1 + XY
R(XY ) = p , R(XY )R(Y X) = 1. (45)
2(1 + X · Y )
Proof. The expression above follows from the identity
(1 + XY )2 |(X 2 Y 2 =1) = 2(1 + X · Y )XY.  (46)
Since 1975, the Hestenes rotor acquired a synonymous name of ‘the Hestenes
spinor’. The Hestenes spinor lay in the even subalgebra of the Clifford algebra
(subalgebra of even multivectors). On the other hand, the spinor invented by Elié
Cartan in 1913, and published again in 1937, reinvented by Marcel Riesz in 1946,
and by Claude Chevalley in 1954, etc., is a one-sided ideal of Clifford algebra.
Even subalgebra is not an ideal, and these different spinors can not be related in
spite of the coincidence of dimensions = 2n−1 .
Rotor of a Simple Bivector 805

6. Homogeneous manifold as a groupoid category

19. Definition (Groupoid category). A directed graph with loops = {nodes ⇔


arrows}, is given by source and target maps: arrows ⇒ nodes. Each arrow must
have a specific source (domain) node and target (codomain) node. A graph with
the injection of each node into the identity arrow (an identity arrow is a loop),
id
nodes ֒→ arrows, where ‘id’ is a section for the source and target maps, is said to
be reflexive. A reflexive directed graph together with the associative composition
of composable pairs of arrows is said to be a category. An arrow that possess two-
sided unique inverse is said to be iso-arrow. A category in which every arrow is
iso-arrow is said to be the Brandt groupoid 1 .
A Brandt groupoid category with exactly one node is said to be a transfor-
mation group, i.e. the set of all arrows is a group of transformations of this single
node,
Transformation group = {single node, group of iso-arrows}.
A generalization of a group to a category of many-nodes is said to be a Brandt
groupoid category,
Groupoid category = {many nodes, groupoid of iso-arrows}.
A manifold on which a Lie group G acts transitively is said to be G-homoge-
neous. If H ⊂ G is the isotropy subgroup, then a G-homogeneous manifold is
homeomorphic to the coset manifold G/H, e.g. a sphere or a hyperboloid. The
Lobachevsky space is an example of a (Aut g)-homogeneous manifold. Every G-
homogeneous manifold can be seen as a one-node category, i.e. as a group G of
iso-arrows of transformations of a single node G/H,

G-homogeneous space
= {single node = G/H, group of arrows = G}. (47)
as a group category
20. Clarification (Group-link problem). The link problem for a given pair, X, Y ∈
G/H, is to determine all links, L(X, Y ) ∈ G, from X to Y, L(X, Y )X = Y, under
the condition that, L(X, X) ≡ idG/H ∈ G. Such a group-link is also called a
boost from X to Y. The important fact is that the link problem for the group
action for dim(G/H) > 1, does not possesses a unique solution, a non-unique
path in G/H, and, in particular, the isometry-link L(X, Y ) ∈ Aut g, from X to
Y, on Lobachevsky space is not unique – contrary to the frequent false claims of
uniqueness in the literature. Non-uniqueness of a group-link-boost can be seen for
example on two-dimensional sphere S 2 ⊂ E ≃ R3 , considered as the single-node of
the rotation-group category O3 . For, p, q ∈ S 2 , let s ∈ S 2 ⊂ E be the vector that
bisects the Euclid angle between p and q, s ∧ p ∧ q = 0. Then, for r orthogonal
to (p ∧ q)-plane, a rotation that leaves invariant a vector v in the (r ∧ s)-plane,
v ∧ r ∧ s = 0, does a v-boost from p to q for every such vector v.
1 Oftenthe term ‘groupoid’ is used in another meaning, as a binary global operation (all pairs
are composable) known as a ‘magma’ by Bourbaki.
806 Zbigniew Oziewicz AACA

21. Conjecture (Oziewicz 2006). Let G/H be G-homogeneous manifold such that
2 ≤ dim(G/H). Thus almost every triple of elements of such G-homogeneous
manifold G/H determines a unique ternary boost (group-arrow) between each pair
of this triple, i.e. every group-boost on such homogeneous manifold is ternary,
G/H × G/H × G/H −→ G − links. (48)
For some examples of the group-link problem on homogeneous manifolds we
refer for instance to [van Wyk 1986, Urbantke 2003, Oziewicz 2006, 2007], among
many others.
We may be interested in a category possessing unique link-arrow for each pair
of his nodes, G/H × G/H −→ Arrows. For this purpose we should consider every
element of G/H as a node, i.e. as 0-morphism. Each object X ∈ G/H possesses
its own identity arrow idX , and X 6= Y implies idX 6= idY ,
  
G-homogeneous space set of all nodes = G/H groupoid
= , . (49)
as a groupoid category node ∈ (G/H) of arrows
22. Definition (Unique link). Consider the following (Aut g)-homogeneous mani-
fold Nk of non-isotropic vectors,
Nk ≡ {X ∈ E|X 2 = k ∈ K, k 6= 0} = (Aut g)/Hk . (50)
We wish to consider Nk as a set of nodes (objects = 0-morphisms) of a category
possessing a unique arrow between every pair of nodes.
An idempotent, f 2 = f ∈ EndK E, that is g-self-adjoint, g ◦ f = f ∗ ◦ g, for

g = g, is said to be g-projection.
Each unique arrow will be parameterized in terms of the g-projection as
follows
 
w X ⊗ gX
Nk × Nk ∋ X, Y −−→ w(X, Y ) ≡ idNk − Y ∈ ker(gX), (51)
k
(X · Y )2
{w(X, Y )}2 = k − . (52)
k
We wish to consider a vector w ∈ ker(gX) ⊂ E, (51)-(52), as a parameter of
the unique morphism from vector-object X to vector-object Y, see Figure 6, for
Nk = S 1 .
23. Example (Euclid’s geometry). Let a sphere S 2 be a set of 0-morphisms,
i.e. nodes (objects). Then three-dimensional ball B 3 , ∂B 3 = S 2 , with blow-up zero,
(59), parameterizes the set of all arrows (morphisms). Consider a triple of mutu-
ally orthogonal vectors, X, Y ∈ S 2 , and 0 6= w ∈ B 3 . Then, w(X, X ′ ) = w(Y, Y ′ )
and, 0 = X · Y 7→ X ′ · Y ′ = w2 , i.e. this w-arrow is not an isometry.
On Figure 6, S 1 is a set of nodes, and the ball B 2 with blow-up zero (59)
parameterizes
√ a set of arrows. A vector w ∈ B 2 possesses the Lorentz scalar factor
2
γw = 1 − w , but w-arrow (denoted by Bw below in Definition 25) is not an
isometry, −1 = (−X) · X 7→ Z · Y = 2w2 − 1.
Rotor of a Simple Bivector 807

γw |X| w ⊕ ker(gw)

w Y
Z
X ker(gw)
−X
w−1

Figure 6. A sphere S 1 of radius |X| is a manifold of 0-


morphisms, nodes, objects of a groupoid category. A set of arrows
is parameterized by vectors in the ball B 2 with blow-up center
as explained in (59). ‘w-Rotations’, X 7→ Yp = w + γw X, and,
−X 7→ Z, possess the Lorentz factor γw = 1 − w2 /k, but this
is not an isometry.

The unique directed arrow between ordered two nodes of a category Nk (50)-
(52), will be defined in terms of the Lorentz scalar factor γ ∈ K by means of the
Pythagorean theorem. The scalar factor γ in (26) was introduced by Heaviside in
1888, and used by FitzGerald, before the Lorentz isometry group was discovered
and studied by Hendrik Antoon Lorentz and Henri Poincaré in 1904. A priori, the
Lorentz scalar factor generalizes the definition of the angle in Euclidean geome-
try, where γ is the cosine of the angle between two vectors of the same positive
magnitude = 1, −1 ≤ γ ≤ 1.
24. Definition (Heaviside - Lorentz factor). A Lorentz scalar factor γ ∈ K is
defined for each pair of non-isotropic vectors X, Y ∈ Nk ⊂ E, see definition (50),
1
γk ≡ X ·Y ∈ K. (53)
k
Definition 24, (53), generalizes the Euclidean angle for the case that also
includes the negative values2 of 0 6= X 2 = k ∈ K.
What does Definition 24 have to do with isometry(26)? Every isometry
L
L ∈ Aut g ≡ Og ⊂ End E, X −−−−→ Y = LX ∈ Nk , (54)

2 Ifa vector X is isotropic-light, X 2 = 0, then for Y 2 6= 0, one can still define a scalar νX
interpreted as the ‘frecuency’ or the ‘energy’ of the light X relative to Y, as follows, νX Y 2 ≡ X ·Y.
808 Zbigniew Oziewicz AACA

gives rise to its own isometry-dependent Lorentz scalar factor γL ,


X · (LX)
Aut g ∋ L 7−→ γL ∈ K, γL,X ≡∈ K. (55)
X ·X
25. Definition (Arrow from Pythagorean theorem). Let w · X = 0. One can solve
B
(53) in terms of the γ-dependent map, X −−→
w
Y, as follows,
B
Nk ∋ X −−−−w−−→ Bw X ≡ γ w X + w ∈ Nk , (56)
2
2 w
(Bw X)2 = X 2 = k ⇐⇒
(γw ) − 1 = − , (57)
r k
B w2
Nk ∋ X −−−−w−−→ ± 1 − X + w. (58)
k
We postulate the blow-up of zero vector as follows
Bw(X,±X) ≡ ± idX , and X 6= Y imply idX 6= idY . (59)
2
If K = R, and if, 0 < w
k , then the following inequalities must hold,
2
w
If 0< , then, 0 < w2 < k,k < w2 < 0.
or, (60)
k
 
2 2
Alternatively to (56), setting, w ≡ γv, we have, (γv ) 1 + vk ≡ 1.

26. Theorem (Groupoid associative composition). A vector Bw X ∈ Nk exists iff,


w · X = 0, and a composition (Bu ◦ Bw )X exists iff u ◦ (Bw X) = 0. (61)
From the uniqueness of the link, X 7→ Bw X, it follows that the composition,
Bu◦w ≡ Bu ◦ Bw is associative, and the following hold,
u·w u·w
u ◦ w = u + γu w + X, γu◦w = γu γw − . (62)
kγw kγw
(i) (Bw )−1 (γw X + w) = Bw−1 (X) (γw X + w).
2
(ii) w−1 (X) ≡ −γw w − (γw − 1)X.
−1 2 2 2 2
(iii) (w (X)) = w ⇐⇒ γw −1 = γw .

(iv) γw−1 = γw =⇒ Bw−1 (X) ◦ Bw X = X,

and, Bw ◦ Bw−1 (X) (γw X + w) = γw X + w.
(v) Inverse is involutive: {w−1 (X)}−1 (γw X + w) = w.
Theorem 26 was deduced and proved in [Oziewicz 2005], and applied to the
associative addition of not necessarily constant relative velocities between not nec-
essarily inertial reference systems.
27. Clarification (Non-isometry does not mean the violation). Whenever we deal
with the metric tensor, g ∈ E ∗ ⊗ E ∗ , we have the isometry Lie group, Aut g ⊂
End E, g ⇐⇒ Aut g, and g ∧ ⇐⇒ Aut(g ∧ ) ⊂ End(E ∧ ). Therefore it is hard psy-
chologically to accept non-isometries and non-group structures coexisting with the
metric tensor. An obstacle may be the erroneous belief that the only alternative to
Rotor of a Simple Bivector 809

isometry is violation, i.e. non-isometries inside End E, that change the metric ten-
sor. The domain of the groupoidal iso-arrow, Bw , Definition 25, is Nk ∩ (ker gw),
and this can not be extended to the entire K-module E. Therefore Bw 6∈ End E
can not violate the metric tensor. The tranformation Bw coexists perfectly with
the metric tensor, although Bw is not an isometry because it does not act on ten-
sors E ∗ ⊗ E ∗ . It is wrong to conclude that Bw ‘is not an isometry because it does
not preserve the metric tensor’.
A sceptical reader still might insist that there exist an isometry L ∈ Aut g,
such that when restricted to the domain of Bw , it will coincide with the γ-possessor
?
Bw , L|{Nk ∩(ker gw)} = Bw , or, that Bw can be ‘extended’ to act on isotropic vec-
tors, for k = 0, as is the case for every isometry. This is not the case. Again, couple
more reasons showing the distinction of the groupoid, {Bw |w as in (51)-(52)-(58)},
from the group Aut g of symmetries of the metric tensor:
• The isomorphism Bw , (56), preserve ‘length’ but not ‘angle’, and therefore
it is not an isometry, Bw 6∈ Aut g. See for example in Figure 6,

(Bw (−X)) · (Bw X) = Z · Y = (−γw X + w) · (γw X + w) = −k + 2w2


6= X · (−X) = −k. (63)
• The groupoid, {Bw , (56) − (58)}, is not a group. The composition of vectors
(from the ball B 2 ) u ◦ w, as well as of composition of boosts Bu ◦ Bw =
Bu◦w is partial. A vector (Bu ◦ Bw )X ∈ Nk , needs, among other things, two
conditions, w · X = 0 and u · (Bw X) = 0. The composition if it exists is
associative.
• The square of the transformation (56)-(58), Bw Bw X, is allowed if and only
if w2 = 0.
We can conclude this section. There are two alternatives
• Group action implies non-unique link.
• Unique link implies groupoid action.
The O’Raifeartaigh theorem of 1965 state that the relativity group contradicts to
mass splitting of elementary particles. One can ask if this theorem also holds true
for relativity groupoid?

7. Isometry from Bivector


In 1937 Elié Cartan observed that the Lie algebra of the isometry group
Og = Aut(E, g), is given by bivectors E ∧2 ⊂ Cℓ(E, g). Clifford algebra with a
commutator gives rise to Clifford Lie algebra, studied for example in [Ablamowicz
et al. 1982, Mignaco and Linhares 1993]. Most exhaustive studies of Clifford Lie
algebra and Lie subalgebras was carried out by Bernd Schmeikal, who throughly
studied basis-dependent unipotents, idempotents, and nilpotents in Clifford alge-
bra, and discovered the Lie algebra of the standard model, su(3), from the real
Dirac-Clifford algebra, concluding that the standart model of elementary particles
810 Zbigniew Oziewicz AACA

is a property of the Clifford algebra of Minkowski spacetime, [Schmeikal 1996, 2005,


2006, 2007]. We also recommend strongly reading the forthcoming two monographs
by Bernd Schmeikal Lie group guide to the universe, (102 pages), and Primordial
Space (Nova Publishers 2009).
The Clifford product of two bi-vectors is a sum of Grassmann’s 4-vector, 2-
vector and 0-vector. Let u, v, w, x, be a set of 1-vectors. The Clifford product of
simple bivectors is
Clifford
Cℓ⊗2 ∋ (u ∧ v) ⊗ (w ∧ x) −−−−−−−−→ u ∧ v ∧ w ∧ x
+ (u · x)v ∧ w − (u · w)v ∧ x + (v · w)u ∧ x − (v · x)u ∧ w
− (u ∧ v) · (w ∧ x) ∈ Cℓ. (64)
This implies that the submodule of bi-vectors, E ∧2 , is a Lie algebra of the
isometry orthogonal group, e.g. [Ablamowicz et al. 1982],
1
2 [u ∧ v, w ∧ x]
= (u · x)v ∧ w − (u · w)v ∧ x + (v · w)u ∧ x − (v · x)u ∧ w. (65)
Let us recall that a metric tensor g, and a push-forward of a bivector b, gives
rise to two composed endomorphisms, cf. with Section 4,
g

E _ E∗
b∗

Figure 7. Composed endomorphisms.

We denote the above composed endomorphism as follows


g
e
E ∧2 ∋ b −−−−→ g b ≡ b∗ ◦ g ∈ End E ≃ E ⊗ E ∗ .
e (66)
In fact, the above map (66), eg, is a Lie algebra isomorphism from the Lie algebra of
bi-vectors (65), to the Lie algebra of g-skew-symmetric endomorphisms/operators
(hence traceless),
 
  
 Lie algebra 

 
Lie algebra g
e of trace-less
−−−−−−−−−−−−−−−→ (67)
of bivectors Lie algebra isomorphism 
 g-skew-symmetric 
 
endomorphisms
28. Definition (Representation of Lie algebra). Let b ∈ E ∧2 ⊂ Cℓ(E, g), be a
bivector. We use the convention 2 (in Section 2). Then the following morphism
extends to a particular representation of the Lie algebra of the orthogonal group,
E ∋ X −−−−→ (b∗ ◦ g)X ≡ igX b ∈ E. (68)
∗ ∗ 1
g(u ∧ v) ≡ v ⊗ g u − u ⊗ g v
e [e
g(u ∧ v), e
g(w ∧ x)] = 2 g([u ∧ v, w ∧ x]). (69)
e
Rotor of a Simple Bivector 811

29. Observation (Lorentz scalar factor). If Z ∈ E, is a non-isotropic vector, then,


a map Z 7→ LZ, such that (LZ)2 = Z 2 (a map L does not need to be neither a
g-isometry nor even an E-endomorphism) defines the scalar factor γL ∈ K as,

Z · (LZ) ≡ γL Z 2 , see Definition 24. (70)

This scalar γ generalize the Heaviside - FitzGerald - Lorentz -scalar factor (26),
known from special relativity, based on the postulate that each transformation of
reference systems is obliged also to be an isometry. This postulate implies that ev-
ery relativity transformation must be an E-endomorphism also necessarily acting
on the light cone and also on a zero vector that is hard to consider as a refer-
ence system. A Brandt groupoid category (not groupoid as magma), generalizes
the concept of a group with unique neutral element, to ‘group with many neu-
trals/units’, it also leads to a similar Lorentz scalar factor, see Definition 25, and
[Oziewicz 2007, 2008].

30. Main Theorem (Isometry from bivector). Let g ∗ = g be a symmetric metric.


g b)2 = 2b2 ∈ K. There are exactly two
Let b ∈ E ∧2 be a simple bivector, hence, tr(e
(g ∗ = g)-isometries that can be constructed from the simple bivector as follows.
(This theorem was named ‘strange’ by Secret Referee.)

(i) A reflection from a simple bivector is given by the following expression for
b2 6= 0,

g b)2
(e
Lb = id −2 =⇒ L2b = idE , tr Lb = (dim E) − 4. (71)
b2
The trace of the reflection from a vector is different, compare with (38).

(ii) An isometry from a bivector with a third-order minimal polynomial. Let the
scalar γ ∈ K be defined as follows,

γ 2 = 1 + b2 , and γ + 1 6= 0. (72)

Then, the following endomorphism Lb is a g-isometry,

g b)2
(e
Lb ≡ idE +(e
g b) + ∈ Og = Aut(E, g), and (Lb )−1 = L−b , (73)
γ+1
trace Lb = (dim E) − 2 + 2γ. (74)

The above trace agrees with (26) for dim E = 4. Isometry (73) possesses a third-
order minimal polynomial,
n  p  on  p  o
(L − id) L − γ + γ 2 − 1 id L − γ − γ 2 − 1 id = 0. (75)

The following implication holds, P ∧ b = 0, implies, P · (Lb P ) ≡ γ P 2 .


812 Zbigniew Oziewicz AACA

In fact, for endomorphism (73), one can check directly the following identities,

g b)2
(e
L∗b ◦ g ◦ Lb = g − (γ 2 − b2 − 1) ◦ g,
(γ + 1)2
(76)
g b)2
(e
Lb ◦ L−b = id −(γ 2 − b2 − 1) .
(γ + 1)2

8. Proof of the main theorem 30


Let P and Q be vectors in E, and let α, β ∈ E ∗ be unknown covectors to be
determined. Consider the following endomorphism

L = idE +P ⊗ α − Q ⊗ β, α ∧ β 6= 0, b ≡ P ∧ Q 6= 0. (77)

Van Wyk in 1958 studied isometry derived from a pair of vectors, i.e. from a pair of
simple endomorphisms. A (g ∗ = g)-isometry from a pair of simple endomorphisms,
(77), is in fact an isometry from a simple bivector b = P ∧ Q, as we are going to
show explicitly.

31. Lemma. Let, b2 , c ∈ K, be such that, c2 6= b2 . The endomorphism (77) gives a


one-parameter c-family of isometries of the following form
2  4b2
Lc = idE + c (e g b)2 ,
g b) + (e tr Lc = (dim E) + . (78)
c 2 − b2 c 2 − b2
32. Corollary. L2c = idE ⇐⇒ c = 0. This proves (71).

Proof of Lemma 31. We need to determine unknown covectors, α and β, as the


functions of the given vectors P and Q, see (81) below. The ansatz-endomorphism
(77) inserted into the isometry condition, Proposition 15, allows an arbitrary scalar
c ∈ K,
    
gP −P 2 Q·P +c α
2 = (79)
gQ −P · Q + c Q2 β
−4 g ∧ b = −4 g ∧ (P ∧ Q) = (c2 − b2 ) α ∧ β 6= 0, (80)
    
α 2 −Q2 Q·P +c gP
= 2 , (81)
β c − b2 −P · Q + c P2 gQ
2 2
αP = 2 2
(c P · Q + b2 ), βP = 2 c P 2,
c −b c − b2
(82)
2 2
αQ = 2 c Q2 , βQ = 2 (c P · Q − b2 ),
c − b2 c − b2
4b2
(α ∧ β)(P ∧ Q) = − 2 (convention 2).  (83)
c − b2
Rotor of a Simple Bivector 813

For the isometry (78) we get the following expression for the Lorentz scalar
factor,
c 2 + b2
P ∧ b = 0, and P · (Lb P ) ≡ γ P 2 =⇒ γ≡ ∈ K. (84)
c 2 − b2
Isometry (78) for c 6= 0 possesses a third-order minimal polynomial,
n  p  on  p  o
(L − id) L − γ + γ 2 − 1 id L − γ − γ 2 − 1 id = 0, (85)
tr L = (dim E) − 2 + 2γ. (86)
The above trace (86) is valid for any c ∈ K, and agrees with (26) for dimK E = 4.
We must distinguish the following two cases.
• c = 0, which implies γ = −1 (84). In this case, and only in this case, Lb is a
reflection from a simple bivector, (71).
• If c 6= 0, then the constant, 0 6= c22c
−b2 can be included in the simple bivector,
and we can put

2c = c2 − b2 6= 0, hence c = γ + 1, and γ 2 = 1 + b2 (87)

This gives a non-reflection isometry from a simple bivector (73)-(74) for γ 6=


−1.

9. Challenge: isometry from trivector


Consider the following endomorphism,
L ≡ idE −P ⊗ α − Q ⊗ β − R ⊗ γ ∈ EndK E,
t ≡ P ∧ Q ∧ R 6= 0, and α ∧ β ∧ γ 6= 0, (88)
∗ ∗ ∗
gt = e
e g(P ∧ Q ∧ R) = (Q ∧ R) ⊗ g P + (R ∧ P ) ⊗ g Q + (P ∧ Q) ⊗ g R. (89)

It is a challenge to find explicitly all (g ∗ = g)-isometries allowed by this


endomorphism, L ∈ Aut g. An analogy with (79) suggest the following hint where
a, b, c ∈ K are arbitrary constants,
    
gP P2 Q·P −a R·P −b α
2 gQ = P · Q + a Q2 R · Q − c  β  (90)
gR P ·R+b Q·R+c R2 γ
v ≡ aR − bQ + cP, v ∧ t = 0, (91)
∧ ∧ 2 2
8 g t = 8 g (P ∧ Q ∧ R) = (v − t ) α ∧ β ∧ γ, (92)
 
P2 Q·P −a R·P −b
det P · Q + a Q2 R · Q − c  = v 2 − t2 . (93)
P ·R+b Q·R+c R2
814 Zbigniew Oziewicz AACA

10. Comparison of rotors


We conclude that a simple bivector, b ∈ E ∧2 , generates an isometry Lc ∈
Aut(E, g) = Og , (78), in particular the reflection (71), and Lb given by (73). This
allows for the concept of a rotor of a simple bivector that a priori does not need
the Hestenes condition, X 2 Y 2 = 1, (44).
33. Definition (Rotor of a simple bivector). Let X ∈ E, and let Lc ∈ Aut(E, g),
be an isometry, given by (78). A rotor of a simple bivector b, Rc (b) ∈ Cℓ(E, g),
is an element of the even Clifford subalgebra, defined by the condition that the
isometry action, X 7−→ Lc X, is expressed within Clifford algebra as follows,
Lc X = Rc (b)XRc (−b). (94)
34. Lemma. The rotor of a simple bivector b is given by the following expression,
up to ±,
c+b 2 2c
Rc (b) = √ , (Rc (b)) = γ + b. (95)
c 2 − b2 c2 − b2
Here γ-factor is defined by (84). In particular for (87) we have
r  
1+γ+b γ+1 b
R(b) = p = 1+ , R(b)R(−b) = 1. (96)
2(1 + γ) 2 γ+1

Lounesto derived the particular form of expression (95) for the Clifford alge-
bra Cℓ(3, 1) [Lounesto 1997, 2001, Chapter 9.8, pages 127-128]. Lemma 34 holds
for any dimension and for any signature. Definition 33 with Lemma 34 is anal-
ogous to Proposition 9 in [Aragón González et al. 2009, page 9]. (The present
paper was submitted in May 2007 and rejected in February 2008 with the Secret
Referee’s opinion ”. . . although some pieces are interesting and perhaps new, the
whole content of the submitted paper is not heavy enough”.)
Let X and Y be vectors. Then a simple bivector b = X ∧ Y, generates an
isometry (73)-(72), if and only if
X 2 Y 2 ≤ 1 + (X · Y )2 . (97)
This must be compared with the more restrictive Hestenes condition (44) for a
rotor from the pair of reflections, X 2 Y 2 = 1. Note that, the extra condition,
γ = X · Y, implies X 2 Y 2 = 1, and the rotor of a bivector (95) collapses to the
Hestenes rotor (45). Therefore, for γ 6= X · Y, a rotor of a bivector generalizes a
rotor of reflections.
35. Example. Light-like vectors coplanar with a bivector b = X ∧ Y, generate an
isometry and rotor (95) for γ 2 = 1 + (X · Y )2 . This is not possible for isometries
from reflections.
36. Example. Let X 2 6= 0. Then a simple bivector
X ∧ Y = X ∧ {Y − (X · Y )X −1 },
Rotor of a Simple Bivector 815

can be presented in terms of orthogonal vectors. Let b = X ∧ Y with X · Y = 0.


In this case the Hestenes rotor is

R(XY ) = (1 + X ∧ Y )/ 2, with X 2 Y 2 = 1, (98)
(X ∧ Y )2 = −1, {R(XY )}2 = X ∧ Y. (99)
Whereas the isometry and rotor of a simple bivector needs a weaker condition.
For, γ 2 = 1 − X 2 Y 2 , we need −1 < γ. For example, X 2 Y 2 ≤ 1, always holds if
one vector is time-like, and the other is space-like, and again this is impossible for
a rotor from such reflections.
37. Corollary. Consider a simple bivector, b ≡ X ∧ Y, such that
X · Y = 0, γ 2 ≡ 1 + b2 = 1 − X 2 Y 2 , Y = γV, (100)
1
=⇒ γ = √ . (101)
1 + X 2V 2
The unital condition, X 2 Y 2 = 1, (44), cannot hold. Then,
LX∧Y X = γX − X 2 Y. (102)
Proof. The assertion follows from the following expressions,
{(X ∧ Y )∗ ◦ g}X = X 2 Y, {(X ∧ Y )∗ ◦ g}Y = −Y 2 X.  (103)

11. Public Referee. Section removed by Editor.


The present paper was honored by being rejected for publication, therefore the
reader must be aware that she/he is reading an illegal text. Why was it rejected?
René Descartes before published his ‘Meditations on First Philosophy: . . . ’,
he send copies of the manuscript to six University Professors of Theology – ‘eru-
dite statesmen’, and in 1641 published ‘Meditations . . . ’ including an Appendix
‘Reproaches of Erudite Statesmen’ with his answers. During the prosecution of
Galileo Galilei in 1636 the Holly Office (Saint Inquisition) asked twelve University
professors (probably all Galileo’s colleges) to write Secret Reports on Galileo’s
1632 book. There was complete consensus, Earth can not move, [Pagano 1984].
However, from 1636 until 1936, there were 300 years without Secret Referees (ex-
ception for Nobel prize). The first journal, in World history, that introduced Secret
Referees, was the Physical Review in 1936.
Albert Einstein published his first and only EPR-paper in the Physical Re-
view in 1935 - the last year without Secret Referees. In 1936 Einstein submitted
his second paper to the Physical Review and after some time received a negative
Secret Referee Report. Einstein wrote an indignant letter to J. T. Tate, Editor of
the Physical Review, asking how it be that a non-published paper of his was shown
to other persons before publication? J. T. Tate answered that starting in 1936 all
papers submitted to the Physical Review will be Secretly Reviewed. Since 1936
Einstein never submitted papers to the Physical Review, and published instead in
816 Zbigniew Oziewicz AACA

the Journal of the Franklin Institute, without Secret Referees, [Pais 1982, Chapter
29].
I consider that every publication is the sole responsibility of its Authors,
and not of any Secret Referee, and not of the Editor. I have no doubts that
the concept of Secret Referee was unfortunate for Science. The Secret Referee
is an obstacle for open scientific discussion, it is against scientific collaborations,
against friendship, creating injustice and envy. Why is discussing science openly
and friendly so difficult? I wish to discuss openly with my Secret Referee. I am
interested in his research, however I prefer the Referee to be public.
Some journals introduced Public Referees: where the author has a right to
chose one or two Referees, that are obliged to publish a report jointly with the
submitted paper. For example, this is the rule in the journal the Concepts of
Physics, edited by Edward Kapuścik in Poland.
A member of the Editorial Board of Advances in Applied Clifford Algebras,
and a member of Bourbaki’s team, considers that a Referee must take full, uncon-
ditional responsibility for publication, even more responsibility than the author.
And further that the duty of the Secret Referee is to be an anonymous co-author
of the submitted but not-yet-published paper by means of his own strong sugges-
tion of ideas from his own research, and based on his personal subjective interest.
Citing: ‘is interesting but not wonderful’ or ‘it is not interesting because it affords
nothing new’, etc. He considers that a Secret Referee must act exactly as an Au-
thor of the paper, and that he has, as the Secret Referee, the full right to develop
his own research inspired by unpublished ideas from the submitted paper. Why
is that an ‘anonymous referee’ who greatly improve a submitted paper can not
be public? What does ‘acknowledgment to an anonymous Referee’ mean if not an
official confirmation of using not author’s ideas? I do not wish to use my Referee’s
‘perhaps new’ ideas, because he is secret, therefore his six page long research report
is totally ignored in the present paper, but it should be published. The six page
long Referee Report I received in February 2008 is not about my submitted paper.
This Report is exclusively about Referee’s own new research inspired by ideas from
my submitted paper. My Secret Referee considers, for example, that only he un-
derstands what Clifford Algebra is, and that he must be convinced (not being the
author) that the paper can be published. But this Secret Referee used submitted
paper for developing his own research, his own Theorems. In my opinion that can
be morally and ethically allowed only after the submitted paper is published. Ev-
erybody reading a paper can invent some generalization, find another applications,
simplify a proof, remove mistakes, or point out seeming errors, however to write a
Report about his own ideas is nothing but trying to be an anonymous co-author.
Everybody has the right to present his subjective ideas, his personal subjec-
tive understanding, but some Secret Referee might consider his private, subjective
idea as official and as the only objective, ‘correct official Science’ such as in George
Orwell’s ‘1984’, because he is the holy Secret Referee.
The problem of ‘grave errors’. At the beginning of the XVIII century, the
members of the Petersburg Academy of Sciences asked Tzar Peter I the Great
Rotor of a Simple Bivector 817

to prohibit publication to some author motivating this by ‘low quality and grave
errors’. The Tzar responded – ‘not at all! This author must publish in order that
everybody knows about his grave errors’. We learn from published errors more
than from forbidden publications. A Referee is also not perfect. Often, what a
Referee considers ‘a grave error’ of the author, could be instead an error of the
Referee. If the author admits that he made an error, and if error is corrected, then
the Referee in fact must be included as co-author, or at least his full name should
be cited in the paper, otherwise, the author are stealing the Referee’s ideas.
A Secret Referee could be jealous, and that happens among scientists. Isaac
Newton spent four years in Court accusing Gottfried Wilhelm Leibniz of plagia-
rism of Newton’s idea. Leibniz wished to discuss with Newton. Newton made his
accusation officially in the London Court. Prosecution was canceled after four
long years because the innocent Leibniz died. Newton also destroyed Hook. About
David Hestenes most delicate Secret Referee’s statements are about the ‘Hestenes
gang’ or ‘Hestenes Witnesses’, although David Hestenes invented a lot of beautiful
ideas. Bernd Schmeikal is also honored to be prohibited for publishing.
Some journals have a list of questions that it is expected that a secret-Referee
must answer.
1. Is the contents and subject of the submitted paper appropriate, or it is not
within the scope of Advances in Applied Clifford Algebras?
2. Is the length of submitted paper appropriate?
3. And this very subjective question: Do you consider that this paper is, or it
is not, interesting?
In science there must not be voting as in politics. The main inquisitor, the
Secret Referee, might consider subjectively that the subject is not interesting per-
sonally for Him, or it might be a subject that the Referee does not understand, so
therefore it must not be published.

Acknowledgements
My friend Bernd Schmeikal since December 24, 2004, is in a jail at Garsten
in Austria. I am greatly thankful to Bernd Schmeikal for highly inspiring snail
correspondence during the years, 2003-2009. In spite that Bernd is isolated from
scientific community, from libraries, from internet, being ecancerded in Austria for
already almost six years, depraved of using email, his ‘thousand’ snail jail-letters to
me are full of the deepest philosophical, mathematical and physical observations,
helped me a lot in re-understanding, and inspired my research greatly.
I am also greatly thankful to Jerzy Kocik, at Southern Illinois University at
Carbondale, who since 1979 in Poland patiently explained me his brilliant ideas,
mostly never published.
818 Zbigniew Oziewicz AACA

I am grateful to Bill Page, at Kingston, Canada, for very stimulating corre-


spondence, for his critics, useful email and Skype discussions, and for important in-
formations. I am specially indebted to Bill for his highest expertise of symbolic cal-
culation with AXIOM ad MAPLE programs, see http://algebraist.crowdvine.com/.
Also many thanks for infinite corrections of English.
It is my pleasure to thank Hilda Maria Colin Garcı́a from my University
UNAM Campus for her love and support.

References
Rafal Ablamowicz, Zbigniew Oziewicz, and Jan Rzewuski, Clifford algebra approach to
twistors. Journal of Mathematical Physics 23 (2) (1982) 231–242.
G. Aragón González, J. L. Aragón, M. A. Rodrı́guez Andrade, and Luis Verde Star,
Reflections, rotations, and Pythagorean numbers. Advances in Applied Clifford Algebras
19 (1) (2009) 1–14.
Emil Artin, Geometric Algebra. Wiley-Interscience New York 1957, 1988.
Élie Joseph Cartan, Leçons sur les Invariants Intégraux. Paris 1922.
Élie Joseph Cartan, Leçons sur la théorie des spineurs I, II. Paris 1937, 1938; The Theory
of Spinors. Hermann et Cie, Paris 1966; MIT Press, Cambridge 1966; Dover 1981.
Claude Chevalley, The algebraic theory of spinors. New York 1954, Columbia University
Press, Morningside Heights.
Jean Dieudonné, Sur les groupes classiques. Hermann, Paris 1948. Actualités Scien-
tifiques et Industrieles, No. 1040.
Hermann Grassmann, Die Ausdehnungslehre. Berlin 1862; Extension Theory. American
Mathematical Society 2000, Chapter 3.4.
Werner H. Greub, Lineare Algebra. Springer-Verlag Berlin 1958, 1976.
Werner H. Greub, Multilinear Algebra. Springer-Verlag, New York, Heidelberg 1967,
1978.
David Hestenes, Space-Time Algebra. Gordon and Breach, New York 1966.
David Hestenes and Garret Sobczyk, Clifford Algebra to Geometric Calculus. D. Reidel
Publishing Company, Dordrecht 1984, 1986.
David Hestenes, New Foundations for Classical Mechanics. D. Reidel Publishing Com-
pany, Dordrecht 1986, 1987, 1999.
David Hestenes, Spacetime physics with geometric algebra. American Journal of Physics
71 (6) (2003).
John David Jackson, Classical Electrodynamics. John Wiley & Sons New York 1962,
1975, 1999.
Jerzy Kocik, http://www.math.siu.edu/jkocik.htm
Jerzy Kocik, Lie Tensors. Ph. Doctoral Thesis, Southern Illinois University at Carbon-
dale 1989.
Jerzy Kocik, Clifford algebras and Euclid’s parameterization of Pythagorean triples. Ad-
vances in Applied Clifford Algebras 17 (1) (January 2007) 71–93.
Pertti Lounesto, Clifford Algebras and Spinors. Cambridge University Press 1997, 2001.
Rotor of a Simple Bivector 819

Artibano Micali and Mustapha Rachidi, On meson algebra. Advances in Applied Clifford
Algebras 18 (3-4) (2008) 875–889.
Juan Alberto Mignaco and Cesar Augusto Linhares, Lie algebras for the Dirac-Clifford
ring. Journal of Mathematical Physics 34 (5) (1993) 2066–2074.
M. Z. Nashed, Generalized Inverses and Applications. Academic Press, New York 1976.
Zbigniew Oziewicz, From Grassmann to Clifford. In: J. S. R. Chisholm and A. K. Com-
mon, Editors, Clifford Algebras and Their Applications in Mathematical Physics, D.
Reidel Publishing Company, Dordrecht 1986.
Zbigniew Oziewicz, Clifford algebra of multivectors (and the tensor product of the Dirac
matrices). Advances in Applied Clifford Algebras 7 (S) (1997) 467–486.
Zbigniew Oziewicz, How do you add relative velocities? In: Pogosyan George S., Luis
Edgar Vicent and Kurt Bernardo Wolf, Editors, Group Theoretical Methods in Physics,
Institute of Physics U.K., Conference Series Number 185, Bristol 2005, ISBN 0-7503-
1008-1, pages 439–444. Available at http://www.worldnpa.org/pdf/group25.pdf.
Zbigniew Oziewicz, The Lorentz boost-link is not unique. math-ph/0608062
Zbigniew Oziewicz, Relativity groupoid instead of relativity group. International Journal
of Geometric Methods in Modern Physics 4 (4) (2007) 1-11; http://www.arxiv.org, arXiv
math.CT/0608770, http://www.worldnpa.org/pdf/ijgmmpOziewicz.pdf
Zbigniew Oziewicz Ternary relative velocity: astonishing conflict of the Lorentz relativity
group with the relativity principle? In: M. C. Duffy and V. O. Gladyshev, Editors,
Physical Interpretations of Relativity Theory, Moscow: Bauman Moscow State Technical
University 2007, pp. 292–303.http://www.worldnpa.org/pdf/ternaryvelocity.pdf
Zbigniew Oziewicz, Electric field and magnetic field in a moving reference system. Review
Bulletin of the Calcutta Mathematical Society 16 (1) (2008) 49–66
Zbigniew Oziewicz, and Cynthia Kolb Whitney, Electric and magnetic fields according
to Hermann Minkowski. Proceedings of the Natural Philosophy Alliance NPA, Albu-
querque 2008, Volume 5 No. 2, pages 183–194, http://www.worldnpa.org/php/
M. Pagano, Editor, I documenti del processo di Galileo Galilei. Roma, Citta del Vaticano,
Archivo Vaticano 1984.
Abraham Pais, SUBTLE IS THE LORD. . . , The Science and the Life of Albert Einstein.
Oxford University Press 1982.
Ian R. Porteous, Topological Geometry. Cambridge 1969, Cambridge University Press.
Ian R. Porteous, Clifford algebras and the classical groups. Cambridge 1995, Cambridge
University Press.
Waldyr Alves Rodrigues, Jr., Q. A. G. de Souza, Jayme Vaz, Jr. and, Pertti Lounesto.,
Dirac-Hestenes spinor fields on Riemann-Cartan manifolds, International Journal of
Theoretical Physics 35 (1996) 1849–1900.
Bernd Schmeikal, The generative process of space-time and strong interaction quantum
numbers of orientation, in: Rafal Ablamowicz, Pertti Lounesto and Joseph M. Parra,
Editors, Clifford Algebras with Numeric and Symbolic Computations, Boston, Basel,
Berlin: Birkhäuser 1996, pages 83–100.
Bernd Schmeikal, Algebra of matter. Advances in Applied Clifford Algebras 15 (2) (2005)
271–290.
820 Zbigniew Oziewicz AACA

Bernd Schmeikal, Symmetric spaces of matter and real fermion manifolds. Advances in
Applied Clifford Algebras 16 (1) (2006) 69–83.
Bernd Schmeikal, The idempotent manifold in a Clifford algebra. Advances in Applied
Clifford Algebras 16 (2) (2006) 159–165.
Bernd Schmeikal, The surabale of space-time and its algebra signature split. Advances in
Applied Clifford Algebras 17 (1) (2007) 107–135.
Bernd Schmeikal, Space-time matter. Bookbound in the penitentiary of Garsten, Austria,
2007. Five offprints.
Bernd Schmeikal, Lie group guide to the universe, in: Lie groups: New Research, Frank
Columbus, Editor, New York: Nova Science Publishers, to be published in 2009. Sub-
mitted June 2008.
Bernd Schmeikal, Rotation roots and eigenforms. Jail at Garsten, Austria, June 2008.
Bernd Schmeikal, Primordial Space, New York: Nova Science Publishers 2009
Helmuth K. Ubantke, Lorentz transformations from reflections: some applications. Foun-
dations of Physics Letters 16 (2) (2003) 111–117.
Hermann Weyl, Annals of Mathematics 44 (1945) 1–6.
C. B. van Wyk, Nuoevo Cimento 10 (1958) 854
C. B. van Wyk, Lorentz transformations in terms of initial and final vectors. Journal of
Mathematical Physics 27 (5) (1986) 1311–1314.
http: //algebraist.crowdvine.com/
http: //en.wikipedia.org/wiki/Gyrovector space

Zbigniew Oziewicz
Universidad Nacional Autónoma de México
Facultad Estudios Superiores Cuautitlán
C.P. 54714 Cuautitlán Izcalli, Estado de México
e-mail: oziewicz@unam.mx

Received: May, 2007 (Rejected: February, 2008)


Accepted: July, 2008 (without change, no appealed)

You might also like