[petsc-users] Load distributed matrices from directory

[Barry Smith]

> On Sep 29, 2017, at 8:23 AM, Matthieu Vitse <vitse at lmt.ens-cachan.fr> wrote:
> 
> Hi all, 
> 
> I wanted to have some advices from advanced users about what I want to do. I would like to load in parallel a set of distributed matrices into petsc to build an ASM preconditioner, of course without having to handle the full matrix.

   Why, this is likely a painful, yet fruitless optimization. 

   Just generate the entire matrix in parallel and use -pc_type asm

   Or if it is some other program generating the matrix save it to disk with MatView() and then load it with MatLoad(). 

   Or is your matrix generator code sequential and cannot generate the full matrix so you want to generate chunks at a time and save to disk then load them? Better for you to refactor your code to work in parallel in generating the whole thing (since you can already generate parts the refactoring shouldn't be terribly difficult).

  Barry

> Those matrices are generated by my FE software, exported using petsc4py. Is there a simple way to load all the matrices at the same time ? (by targeting a given folder for example, using MatLoad ? I haven’t been able to succeed in doing that yet). I guess it then be directly fed to PCASM? (maybe by provided user-defined subdomains). 
> 
> Thanks, 
> 
> — 
> Matt