Boost logo

Ublas :

Subject: Re: [ublas] New uBLAS maintainer
From: Matwey V. Kornilov (matwey.kornilov_at_[hidden])
Date: 2010-03-16 13:08:06


David Bellot wrote:

> (2) after reading last week emails, I think we could provide basic
> implementations of a few standard algorithms like inversion, solvers,
> etc...

I am agree. It would be useful. The lack of the solvers made me implement
SVD decomposition for solving linear equation sets(and least squares
problem).

> (4) another hot topic which is a recurrent complain about uBLAS: the
> product of 2 matrices. Do we want prod(A,B) or A*B. Let's think about it
> because other libraries implemented A*B in a very efficient manner too.

In my opinion the behavior of operator*() is a tensorial product, i.e
v*u=outer_prod(v,u) for vectors.

> Please everybody contribute with your own ideas and desiderata.
> Let's work and make uBLAS simply the best.

a) In my opinion there is a lack of concatenation function for matrix and
vectors:

vector<double> v(4);
vector<double> u(6);

matrix<double> a(4,1);
matrix<double> b(4,2);

cat(v,u) // vector of 10
cat(a,b) // matrix of 4x3

b) ties as alternative of sub-ranges:

vector<double> v(10);
vector<double> a(3);
vector<double> b(7);

tie(a,b) = v;

c) I posted ( http://old.nabble.com/vector-generator-p26014314.html ) my
implementation of vector_generator class. It would be useful to have such
vector_generator and matrix_generator in order to implement special
matrices(like Frobenius) without allocating a memory.