I was actually using the geometric product.
The geometric product of a base vector with itself is commutative (it's what the inner product gives you, and in fact it
must be in the way I framed it above:
is easily shown to always be 1, i.e. changing order does not result in sign change, i.e. it's commutative.), but with another base vector it's anti-commutative (xy = -xy etc.), which is the wedge-part of the product.
And scalars are completely commutative.
That's what I was assuming here. Or at least I was trying to.
I still must have made some error of that kind somewhere in there. But I can't really spot precisely where.
Basically, all I did was generalizing two things in Geometric Algebra to support arbitrary scalars: Switching (geometric) multiplication orders of base-vectors
results in a swap-factor
and (geometrically) multiplying two vectors
results in another scalar factor
which is essentially the signature, except I have used it on all base-vectors, not just ones of equal index. (Normally, the signature would appear like this:
; where
simply is the signature for that particular vector. I have added the possibility for giving mixed vectors such a signature factor as well.)
Nowhere in this am I even using the scalar product or the wedge product, except for where they happen to align with the geometric product.
All I was trying to do is to see for which instances of
and
this is consistent; if, perhaps, this only even consistently works for
and
I.e. whether this only works for the conditions we already know about (swapping unequal vectors anticommutes, swapping equal ones commutes, only base 1-vectors get an additional factor through signature) or whether there are more general situations.
Now it would be no problem if what I showed was, that only what we already do is allowed, but if I made no mistake (and I probably did), it means that not even that case is allowed.
Edit: Tex doesn't work well with the notation I used up there in c. What I was trying to convey is that c could have an arbitrary list of indices in general, for products of arbitrary n-vectors with arbitrary k-vectors. I wasn't quite sure how else to depict that, but this method using products clearly doesn't work all too well.