[petsc-users] minres and gmres difference

[Barry Smith]

On Mar 1, 2011, at 4:12 PM, Umut Tabak wrote:

> On 03/01/2011 10:17 PM, Matthew Knepley wrote:
>> On Tue, Mar 1, 2011 at 2:45 PM, Umut Tabak <u.tabak at tudelft.nl> wrote:
>> 
>> Henk did not count on different orhogonalization strategies. Try -ksp_gmres_modifiedgramschmidt
>> 
> It is better at the end, with 10000 iterations though, which makes it far from useful for this matrix.

  That 10000 is the default maxium that we allow, so I think your run with the other orgonalization is  meaningless. Change the restart like I suggest in my email.

   Barry

> Umut