← Back to team overview

dolfin team mailing list archive

Re: [Bug 706909] [NEW] [python] matrix multiplication fails in parallel

 

On Tuesday January 25 2011 10:28:07 Joachim Berdal Haga wrote:
> > To follow the current design, we should rather have a method that
> > returns whatever data is needed to create a vector, then use the
> > existing factory method to actually create the vector:
> > 
> > 1. get layout data from matrix
> > 2. get factory from matrix
> > 3. use factory to create vector
> 
> It looks like the handiest data container for (1) is the matrix itself,
> though. The actual method can still be put in the factory?
> (Is there a difference, design-wise, if the method resizes rather than
> creates the vector?)

As it is now the factory just gives you the correct Vector. You need to resize 
it after construction. 

But aren't you passing the return vector to the mult method, and then a resize 
is only required? You might have lost me on the way ;)

Johan



References