Re: [boost] uBLAS vs MTL for sparse matrices

From: Paul C. Leopardi (leopardi_at_[hidden])
Date: 2002-05-03 09:38:58

Joerg,
Thanks for your reply. Comments below.
> But these matrices aren't permutation matrices, i.e. the non zero matrix
elements aren't always 1?
No, they are not always 1.
The matrices form an orthogonal basis for a Clifford algebra. eg.
[1 0] [0 -1] [0 1]  [1  0]
[0 1],[1  0],[1 0], [0 -1] form a basis for R_{1,1}
See Ian R. Porteous, "Clifford algebras and the classical groups",  Cambridge 
University Press, 1995.

> For O(n) complexity you better use sparse_vector_of_sparse_vector or
compressed_matrix (both currently undocumented ;-(
OK, I will try compressed_matrix, although I am worried that it is not 
documented. I may not be able to do any coding until June.

> P.S.: do we need to consider a class permutation_matrix<>?
Not on account of GluCat.

My own PS: a while ago, it was suggested that uBLAS and MTL be examined to 
see what the commonalities and differences are. Has this been done?
Is MTL3 still being developed? Will it have a uBLAS adaptor?
See http://groups.yahoo.com/group/boost/message/24374
Thanks