re Optimisation / Sparse Matrices

There is provision for extensions that handle such things.
It is through the csymbol element.  For example, a
sparse matrix representation could be defined so that
For example, you could write:

<apply><csymbol definitionURL=".../SparseMatrix">SparseMatrix</csymbol>
    <vector><cn>5</cn><cn>5</cn><cn>
    <vector><cn>1</cn><cn>1</cn><cn> 3.232432423</cn></vector>
    <vector><cn>1</cn><cn>2</cn><cn> 3.01</cn></vector>
    <vector><cn>1</cn><cn>5</cn><cn> 3.43543453</cn></vector>
</apply>

to mean a 5 by 5 sparse matrix with the a[i,j] specified for selected 
coordinates.  Column defined matrices could be handled in a similar
manner.

The presentation is handled via XSLT and can range from something
as simple as listing the non-zero table entries, c/w coordinates, through 
to generating an actual matrix.

What is needed is an agreed apon vocabulary and then
references to those definitions.  Building such a vocabulary 
is one of the goals of the OpenMath project.  The MathML content
definitions match (or will shortly) a specific collection of those
definitions.  Both groups are tracking changes closely.

Stan Devitt