pressure shur complement with PETSC

[Barry Smith]

     We need to come up with a good general PETSc infrastructure for  
solving these types of systems,
that supports a variety of solving approaches.

     Barry


On Jan 10, 2008, at 10:36 AM, filippo wrote:

> Dear Professor Barry,
> I'm a PHD student at University of Rome 'La Sapienza' and I'm  
> developing a
> fluid dynamic Finite Element code in a LINUX environment.
>
> After reading the book you have published, we decided to use PETSC  
> and we
> downloaded all the package from the web site.
>
> To resolve the Navier-Stokes system I have to resolve the following  
> linear equation
> system
>
> |A  B| |u|   |f|
> |    | | | = | |
> |BT  0| |p|   |g|
>
> I want to solve the previous system with the pressure Shur  
> complement technique.
>
> Given an initial guess for pressure p I would like to solve the  
> following system
>
>  (BA-1BT)*p=BA-1f-g
>
> Then I calculate the variable u:
>
>  u=A-1(f-BTp)
>
> I’m wondering if it is possible, with PETSC, to solve the shur  
> complement
> equation system without allocating and calculating the global matrix  
> (BA-1BT)
> i.e. if it is possible to give to PETSC the single matrix B and A-1  
> in order to save allocating memory.
>
> Please could you give me any suggestions in order to solve this  
> problem with PETSC ?
> Please in case it is possible, could you give some examples showing  
> the solution of the problem ?
>
> Thanks in advance for your time replay.
>
> Filippo Menichini
>
>
> No virus found in this outgoing message.
> Checked by AVG Free Edition.
> Version: 7.5.516 / Virus Database: 269.19.0/1216 - Release Date:  
> 09/01/2008 10.16
>

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.mcs.anl.gov/pipermail/petsc-dev/attachments/20080110/37c5e436/attachment.html>