Thanks for those contributions
At this point I am just experimenting.
I'm not even concerned with the outer product right now. As said, I only experimented with the geometric product.
However, you are right about the outer product being changed quite significantly.
As I already showed in my previous post, in using a
, you actually don't even get a pure antisymmetric part. You get both a scalar (which is symmetric) AND a(n antisymmetric) bi-vector component if you use the standard definition of a wedge product.
I did not define the inner product as anti-commutative. I already showed how
, i.e. the factor you get from exchanging a base-vector with itself, must always be 1. Hence, it is symmetric.
, that is correct. It's a nice alternate method to denote it.
Your orthogonality remark is topical: That's actually why I am investigating this direction.
You do not actually need an orthonormal basis of vectors. You can just have n completely arbitrary base-vectors for an n-dimensional space. They can lie at completely odd angles to each other. The only limitation is that they must not be linearly dependent on each other.
That
would just be a generalization. Nothing unheard of. In fact, in certain cases, like, for instance, in
triclinic crystals, it's actually more natural to describe your coordinates in such a non-orthognonal and perhaps even non-normal base.
However, I have found the problem with my problem now. I did nothing wrong in my calculations, but I interpreted it wrong:
The offending equation was the following:
So that means:
I previously simply assumed, that that means,
could
only be 1 to fullfill this equation. However, all that is required, in accordance with a different equation I already got before, is this:
This is required for sake of consistency.
Thus, the whole system reduces again to precisely what we already had before, as far as commutativity goes.
All that is left is our
, which now are restricted a slight bit further too.
In particular:
From there we could get:
This is definitely consistent if we just set
. I'm not absolutely certain right now, that this is the only consistent concept, but it might be.
If it turns out that this is the unique solution, we'd essentially have some symmetric matrix
of which each
and
stands for the additional factor of a geometric product of the
and
base vectors. - This would essentially be the metric tensor of our given system. The main diagonal would give the usual signature of the system, while all the other values
correspond to rotated components. They only occur, if the system is not orthogonal.
As said, I'm not yet fully convinced, that this is the only solution here, though it's not unlikely.
But if it is, we'll still have to look into what happens for other vector multiplications. For instance, what happens if we multiply a vector and a bi-vector or, essentially equivalently, three vectors? Things can become really complicated really quickly, but most certainly, a pattern for multiplying
with
vectors should appear sooner rather than later.