[petsc-users] ODE solving

[Xuan Yu]

On Oct 21, 2010, at 6:07 PM, Matthew Knepley wrote:

> On Thu, Oct 21, 2010 at 5:02 PM, Xuan Yu <xxy113 at psu.edu> wrote:
> 
> On Oct 21, 2010, at 5:56 PM, Matthew Knepley wrote:
> 
>> On Thu, Oct 21, 2010 at 4:46 PM, Xuan Yu <xxy113 at psu.edu> wrote:
>> Hi
>> 
>> 
>> I am using Petsc solve ODE nonlinear problem.
>> 
>> I was told to use direct method: umfpackage because of the jacobian matrix is 1860 by 1860.
>> 
>> But the time consumption is still too big:
>> 
>> TSSetType(ts,TSBEULER)
>> 
>> time ./pihm -log_summary -ksp_type preonly -pc_type lu -pc_factor_mat_solver_package umfpack
>> 
>> I got the result:
>>  
>> 
>> When I use TSSetType(ts,TSSUNDIALS), it is fast:
>> 
>> 
>> 
>> How can I find the reason why BEULER is so slow?
>> 
>> You SNES is not converging. You solve 163 systems, but do 4300 function evaluations.
> 
> Could you please tell me more solution options or possible mistakes in my code?
> 
> Jed, is right. You are using coloring with a lot of colors. Sundials is doing much better
> than backwards-Euler, why not just use it?
> 
>    Matt

Thanks,  at first we are using sundials. Now, we found Petsc is more popular, and powerful of parallel and precondition. So, do you think it is the best way to use sundials option under Petsc framework?

Xuan



>  
>>    Matt
>>  
>> Thanks
>> 
>> Xuan
>> 
>> 
>> 
>> 
>> 
>> 
>> -- 
>> What most experimenters take for granted before they begin their experiments is infinitely more interesting than any results to which their experiments lead.
>> -- Norbert Wiener
> 
> 
> 
> 
> -- 
> What most experimenters take for granted before they begin their experiments is infinitely more interesting than any results to which their experiments lead.
> -- Norbert Wiener

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.mcs.anl.gov/pipermail/petsc-users/attachments/20101021/a077c228/attachment.htm>