[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
## Re: Static storage of huge sparce matrices

**From**: |
c. |

**Subject**: |
Re: Static storage of huge sparce matrices |

**Date**: |
Fri, 2 Nov 2012 23:31:48 +0100 |

On 2 Nov 2012, at 21:10, Joza wrote:
>* In light of that, is it possible to access and make use of the storage*
>* structure?*
>* *
>* For example, if I am implementing SOR, normally I would iterate through all*
>* columns (1000). There is a lot of needless computation here as most of the*
>* elements are zero. Will Octave automatically ignore such computations, since*
>* it knows my matrix is sparse and has stored it as such?*
Yes.
Still, if your homemwork assignement is to construct a data structure
for sparse matrices and then use it in an implementation of the SOR method,
I think you should do everything from scratch.
You should ignore the fact that Octave has a special sparse matrix class
and overloaded operators for sparse matrix times vector multiplication and
mldivide (matrix left division).
Your implementation in Octave's interpreted language will probably be less
efficient than what can be done by using Octave's built-ins, yet more
instructive.
c.