Scribd
Upload
23 views3 pages

A Minimalist Construction of The Geometric Algebra: Lemma 1 Ltd. 2nd Floor, 31A Chain Street, Reading UK. RG1 2HX

This document constructs the geometric algebra from minimal assumptions. It defines functions to represent products of basis elements and proves they satisfy an identity, allowing the algebra to be defined. The algebra is shown to be associative with the desired properties. Specific geometric algebras GA(p,q) are defined as subalgebras, shown to have the correct dimension and satisfy the Clifford algebra equations.

Uploaded by

James Chapel
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
23 views3 pages

A Minimalist Construction of The Geometric Algebra: Lemma 1 Ltd. 2nd Floor, 31A Chain Street, Reading UK. RG1 2HX

This document constructs the geometric algebra from minimal assumptions. It defines functions to represent products of basis elements and proves they satisfy an identity, allowing the algebra to be defined. The algebra is shown to be associative with the desired properties. Specific geometric algebras GA(p,q) are defined as subalgebras, shown to have the correct dimension and satisfy the Clifford algebra equations.

Uploaded by

James Chapel
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 3

A Minimalist Construction of the Geometric

Algebra

R.D. Arthan
arXiv:math/0607190v2 [math.RA] 9 Jul 2006

Lemma 1 Ltd. 2nd Floor, 31A Chain Street, Reading UK. RG1 2HX
[email protected]

Abstract. The geometric algebra is constructed from minimal raw ma-


terials.

1. Introduction. Macdonald [2] has noted that descriptions of the geomet-


ric algebra in the literature have tended either to use advanced concepts such
as tensor algebra or to skip over or skimp on the proof of existence of the al-
gebra. He remedies this with a simple construction of the algebra based on a
solution to the word problem for the defining equations of a Clifford algebra. In
this brief note, I give a construction along similar lines with a slightly different
take on the solution to the word problem. This simplifies Macdonald’s approach
in some respects, in particular, by using canonical representatives rather than
equivalence classes. As a technical convenience, I also adopt conventions that
make the GA(p, q) for different p and q fit together so they can be constructed
all at once. As with Macdonald’s construction, only an elementary knowledge of
discrete mathematics and real vector spaces is required.
2. The Construction. In the following, the variables I, J and K are always
finite sets of integers, FZ denotes the set of all finite sets of integers, #A denotes
the number of elements of a finite set A, and A ⊖ B denotes the symmetric
difference, (A \ B) ∪ (B \ A), of any sets A and B.

Definition 1 Define functions α, β : FZ×FZ → N and σ : FZ×FZ → {+1, −1}


as follows:

α(I, J) = #{(i, j) | i ∈ I ∧ j ∈ J ∧ j < i}


β(I, J) = #{i | i ∈ I ∧ i ∈ J ∧ i < 0}
σ(I, J) = (−1)α(I,J)+β(I,J)

Lemma 1 For any I, J, K ∈ FZ, σ(I, J)σ(I ⊖ J, K) = σ(I, J ⊖ K)σ(J, K)

Proof: From the definition of σ, it suffices to prove the following congruences


modulo 2:

α(I, J) + α(I ⊖ J, K) ≡ α(I, J ⊖ K) + α(J, K)(mod 2) (1)


β(I, J) + β(I ⊖ J, K) ≡ β(I, J ⊖ K) + β(J, K)(mod 2) (2)
From the definition of α and the inclusion/exclusion principle, α(I ⊖ J, K) =
α(I, K) + α(J, K) − 2α(I ∩ J, K). So α(I ⊖ J, K) ≡ α(I, K) + α(J, K)(mod 2).
Similarly, α(I, J ⊖ K) ≡ α(I, J) + α(I, K)(mod 2). Adding these congruences
and rearranging gives (1). Replacing α by β in the same argument gives (2).
Definition 2 Define GA(∞, ∞) to be the real vector space with a basis of sym-
bols eI , one for each I P∈ FZ. Thus each element of GA(∞, ∞) has a unique
representation as a sum I∈FZ λI eI , with real coefficients λI , where λI = 0 for
all but finitely many I. Define a multiplication on GA(∞, ∞) as follows:
X X X
( λI eI )( µJ eJ ) = λI µJ σ(I, J)eI⊖J
I J I,J

Define 1 = e{} and ei = e{i} for each i ∈ Z.


Theorem 1 GA(∞, ∞) is an associative algebra over the field R of real num-
bers, having 1 as a two-sided unit. It is generated as an algebra by the elements
ei , which are linearly independent and satisfy:
e2i = 1 if i ≥ 0
e2i = −1 if i < 0
ei ej = −ej ei if i 6= j
Proof: To say that GA(∞, ∞) is an algebra over R is to say that it is
a real vector space and that its multiplication satisfies the bilinearity condi-
tions (λx)(µy) = (λµ)(xy), (x + y)z = xz + yz and x(y + z) = xy + xz
for any λ, µ ∈ R and x, y ∈ GA(∞, ∞). It is a real vector space by construc-
tion and bilinearity is easily checked from the definitions. The multiplication
will be associative if (eI eJ )eK = eI (eJ eK ) for every I, J, K ∈ FZ, i.e., if
σ(I, J)σ(I ⊖J, K)e(I⊖J)⊖K = σ(I, J ⊖K)σ(J, K)eI⊖(J⊖K) , but as (I ⊖J)⊖K =
I ⊖ (J ⊖ K), this reduces to lemma 1. It is easy to check that 1x = x1 = x for
any x ∈ GA(∞, ∞), so 1 is indeed a two-sided unit for the algebra. The set of
elements of the form ei is a subset of the basis we have used to define GA(∞, ∞)
and so must be linearly independent. By induction on m and the definitions, one
has that if I = {i1 , i2 , . . . , im }, with i1 < i2 < . . . < im then eI = ei1 ei2 . . . eim .
This means that the ei do indeed generate GA(∞, ∞) as a real algebra. Finally,
the formulae for the various cases of e2i and ei ej are easy to verify from the
definitions (for example, if i < j, then σ({i}, {j}) = 1, and σ({j}, {i}) = −1, so
ei ej = e{i,j} = −ej ei ).
Definition 3 For p, q ∈ N, define GA(p, q) to be the subalgebra of GA(∞, ∞)
generated by the ei , −q ≤ i < p.
Theorem 2 GA(p, q) is an associative algebra over R with 1 as a two-sided unit.
It has dimension 2p+q as a real vector space and is generated as an algebra by
elements ei , −q ≤ i < p, which are linearly independent and satisfy:
e2i = 1 if i ≥ 0
e2i = −1 if i < 0
ei ej = −ej ei if i 6= j

2
Proof: GA(p, q) has a basis comprising those eI where I ⊆ {−q, −q +
1, . . . , p − 1}. As there are 2p+q such I, GA(p, q) has dimension 2p+q . The rest is
immediate from theorem 1.
3. Concluding Remarks. No special properties of the field R have been
used in this construction. Any other field could have been used for the scalars.
Lundholm [1] gives a generalised construction with some similarities to the one
presented here.
The equations in our two theorems define what is known as a Clifford algebra
of type (p, q). Together with the condition on the dimension in theorem 2, the
equations characterise the GA(p, q) as what are called the universal Clifford
algebras. The real numbers themselves provide an example of a non-universal
Clifford algebra of type (1, 0): they satisfy the equations if one takes e0 = 1 = 1,
but in GA(1, 0), e0 6= 1. For more information see theorem 13.10 in [3]. The
method of proof of that theorem shows that GA(∞, ∞) is characterised by the
conditions of theorem 1.
The functions α and β give the number of sign changes involved in putting a
product of the elements ei in a normal form, see [2]. It is perhaps surprising that
this intuition is scarcely relevant to the proofs given here. The condition i < j
in the definition of the function α was only used above to derive the equations
e2i = ±1 and ei ej = −ej ei . Varying the condition gives similar algebras in
which different pairs of basis elements can be made to commute or anticommute
as desired. However, none of these variants is a Clifford algebra.
It is traditional to give the ei that generate GA(p, q) positive indices, say
with e2i = 1 for 1 ≤ i ≤ p and e2i = −1 for p < i ≤ p + q. This has the
disadvantage that GA(p, q) 6⊆ GA(p + r, q + s) unless r = 0. Our convention
of using negative indices for the basis elements with negative squares means
that the different GA(p, q) all fit together into the one large algebra GA(∞, ∞).
GA(∞, ∞) is perhaps a better candidate for the name “the geometric algebra”
than any of the other (N × N)-indexed family of contestants for the title.

References
1. Douglas Lundholm. Geometric (Clifford) algebra and its applications.
arXiv:math.RA/00605280 v1, 10 May 2006.
2. Alan Macdonald. An Elementary Construction of the Geometric Algebra, 2002.
Adv. Appl. Cliff. Alg. 12, 1-6 (2002).
3. Ian R. Porteous. Topological Geometry. Van Nostrand Reinhold, 1969.

You might also like