Hi,
I just read this inspiring paper: TbGAL about implementation of high dimensions. The key idea is to factorises k-blades as the (outer / geometric) product of vectors, leading to a very performant memory storage. They conducted some test up to dimension 1536. This approach presents 2 drawbacks: handles only versors and k-blades but not general multivectors nor k-vectors that are not k-blades, and the algebra signature should be (p,q,0), but still, the approach is very interesting.
If you have in mind some applications that requires such very high dimensions, just discuss them here.
2 Likes
Cureently I work with geometric algebras with 10,000 dimensions and more to solve partial differential equations. Grassmann.jl has an API that lets you work over a very large basis manifold. Obviously you cant work with full multivectors at that scale, but you can still do things.
I am new to GA. Just learned about outer products and k-vectors.
I have a vector in very high dimensions say (N-D) usually in thousands. I have a vector (u) and I want to sample some points on N-1 dimensional plane which is orthogonal to (u) at the tip of (u) and the points should also be around the tip of the vector (u). I was thinking can I use GA for this purpose? Any suggestions are welcome.