Here is yet another development of some basic Clifford stuff without too many formulas or programs—sort of a short-cut to the original motivations of Clifford algebras, but via modern notions.
We start with an n dimensional vector space V over the reals, R; Q is a negative definite quadratic form from V to R and Q(x) = –(the length of x)2. The axiomatic definition gives us the Clifford algebra C over V. If B = {e1 ... en} is an orthogonal basis for V then the elements of B generate C. Recall the definition of the Clifford group: those clifford numbers x with an inverse such that for all y if y∈V then φx(y) = xyα(x–1) ∈ V. Thus V determines C which includes the Clifford group Γ. We take C to also include V and R. If x, y ∈ V then <x, y> is the unique symmetric bilinear form associated with Q such that for all x ∈ V Q(x) = <x, x>.
Our notation here matches the Wikipedia article for V, Q, α, Γ and <∙, ∙>.
φx(y) is the transformation of y for which the Clifford numbers were invented. Lets see how they transform V.
Clearly 1 ∈ Γ and φ1 leaves V unmoved. Calling Γ a group is due partly to φxy(z) = φx(φy(z)) for all x, y ∈ Γ and z ∈ V.
If {p, q} ⊂ V and <p, q> = 0 then φq(p) =
qpα(q–1) = –pqα(q–1) = –pqα(–q) = –pq(q) = –p(qq) = p
while φq(q) = qqα(q–1) = qqα(–q) = qq(q) = –q.
Thus φq is a reflection in the plane in V which is orthogonal to q. If you understand how to build an arbitrary orthogonal transformation D by composing reflections from planes thru the origin you can now find a Clifford number d such that φd = D. We conclude that R ∪ V ⊂ Γ ∪ {0}.
There are some non-trivial jumps in the above that I will try to fill in.
If x ∈ Γ and x is even then φx preserves orientation.
If x ∈ Γ and x is odd then φx reverses orientation.
if x ∈ Γ then x is even or odd.