[PD] complex network

Mathieu Bouchard matju at artengine.ca
Fri Apr 3 21:11:33 CEST 2009


On Fri, 3 Apr 2009, Ricardo Brazileiro wrote:

> i'm refering a graphs, for study models in complex networks. it's 
> necessary uses matrix....

I don't think anything in Pd handles sparse matrices of any kind, but I 
could be wrong. In that case, you will have to use ordinary matrices. If 
you have very very large numbers of nodes, it can take a lot of room 
(quadratic room) but especially a lot of time (cubic time).

AFAIK, the only tool that has int matrices and byte matrices is GridFlow, 
so, anything else would use t_float for that (?), but that only saves on 
space.

Probably computation time is the biggest issue. There are some upsides to 
using ordinary matrices: for example, you can compute some fabulous 
shortcuts using eigendecomposition (valór e vector proprio, aka 
autovalores). But if you have some thousand vertices, the computation time 
is still quite prohibitive.

So if you need something better, you will need some kind of sparse matrix 
or some non-matrix graph representation. If I were you, I'd try to find 
something that can be processed as a matrix, so that I have a greater 
variety of operations to pick from.

  _ _ __ ___ _____ ________ _____________ _____________________ ...
| Mathieu Bouchard - tél:+1.514.383.3801, Montréal, Québec


More information about the Pd-list mailing list