## Glas :## Re: [glas] vector space, banach space and hilbert space concepts |

**From:** Anders Logg (*logg_at_[hidden]*)

**Date:** 2005-03-29 11:16:50

**Next message:**Matthias Troyer: "Re: [glas] vector space, banach space and hilbert space concepts"**Previous message:**Karl Meerbergen: "Re: [glas] vector space, banach space and hilbert space concepts"**In reply to:**Karl Meerbergen: "Re: [glas] vector space, banach space and hilbert space concepts"**Next in thread:**Matthias Troyer: "Re: [glas] vector space, banach space and hilbert space concepts"

On Tue, Mar 29, 2005 at 05:32:56PM +0200, Karl Meerbergen wrote:

> I think it is quite important that we agree the way we proceed. If we need

> another round of requirements, it should be done.

Here's a requirement/use case: For finite element applications, I

would like to see a sparse matrix with the following member function:

void add(const double block[],

const unsigned int rows[], unsigned int m,

const unsigned int cols[], unsigned int n);

This function would get called repeatedly many times (~10^6) to add a

block of values at the given indices to the sparse matrix. It needs to

be fast and should be clever enough to avoid having to allocate or

move data around excessively.

This is the key piece of functionality for finite element

applications. And of course also the matrix-vector product, but that's

about it.

/Anders

**Next message:**Matthias Troyer: "Re: [glas] vector space, banach space and hilbert space concepts"**Previous message:**Karl Meerbergen: "Re: [glas] vector space, banach space and hilbert space concepts"**In reply to:**Karl Meerbergen: "Re: [glas] vector space, banach space and hilbert space concepts"**Next in thread:**Matthias Troyer: "Re: [glas] vector space, banach space and hilbert space concepts"