[Libmesh-users] nonlinear Newton solver - it was old API

2012-01-23 Thread Michael Povolotskyi
Dear Libmesh developers/users, with your help I fixed the problem with the nonlinear Newton solver. The problem was I was using the old style APIs where one has to define the matvec functor. Apparently this does not work now. The workaround was to define the residual and the jacobian functors. I

Re: [Libmesh-users] nonlinear solver and petsc 3.2

2012-01-23 Thread Michael Povolotskyi
On 01/23/2012 12:26 PM, John Peterson wrote: > On Sun, Jan 22, 2012 at 10:13 PM, Michael Povolotskyi > wrote: >> Dear libmesh developers and users, >> I'd like to upgrade my code to use libmesh-0.7.2 with petsc3.2. > Hi Michael, > > Can you let us know what version of libmesh you are upgrading fr

Re: [Libmesh-users] nonlinear solver and petsc 3.2

2012-01-23 Thread John Peterson
On Sun, Jan 22, 2012 at 10:13 PM, Michael Povolotskyi wrote: > Dear libmesh developers and users, > I'd like to upgrade my code to use libmesh-0.7.2 with petsc3.2. Hi Michael, Can you let us know what version of libmesh you are upgrading from? Either an SVN revision or a tagged version would hel

Re: [Libmesh-users] nonlinear solver and petsc 3.2

2012-01-23 Thread Derek Gaston
Also, it looks like you might be using a libMesh release... you might try grabbing the current SVN version. We utilize the libMesh NonlinearSystem interface to PETSc all day long... so the likelihood of just direct bugs is fairly low... but one may have made it into the "release" that has subseque

Re: [Libmesh-users] nonlinear solver and petsc 3.2

2012-01-23 Thread Ataollah Mesgarnejad
Michael, It may seem obvious, but do you call Jacobian.close() at the end of your jacobian assembly function? Best, Ata -- Try before you buy = See our experts in action! The most comprehensive online learning library f

Re: [Libmesh-users] Periodic boundary conditions in 1D

2012-01-23 Thread Derek Gaston
I can't think of why it wouldn't work... we'll look into it here. Derek On Fri, Jan 20, 2012 at 11:56 AM, Dmitry Karpeev wrote: > Hi, > I've been trying to build a simple example using periodic boundary > conditions in 1D, but to no avail -- the system seems to ignore > periodicity altogether. A

Re: [Libmesh-users] nonlinear solver and petsc 3.2

2012-01-23 Thread Jed Brown
On Mon, Jan 23, 2012 at 10:11, Michael Povolotskyi wrote: > Sorry, I made a mistake in my previous e-mail. > The situation is as follows: > when I use flag -snes_mf the code does not crash. > But if I want to use my jacobian routine, then it crashes. > Does this happen with libmesh examples? Insi

Re: [Libmesh-users] nonlinear solver and petsc 3.2

2012-01-23 Thread Michael Povolotskyi
Sorry, I made a mistake in my previous e-mail. The situation is as follows: when I use flag -snes_mf the code does not crash. But if I want to use my jacobian routine, then it crashes. Thank you, Michael. On 01/23/2012 12:56 AM, Michael Povolotskyi wrote: > On 1/23/2012 12:53 AM, Jed Brown wrote:

Re: [Libmesh-users] HOW TO COMPILE LIBMESH and

2012-01-23 Thread Roy Stogner
On Sun, 22 Jan 2012, Dmitry Karpeev wrote: > What about cygwin? PETSc can be built on cygwin (with some effort, > sometimes). > Can libmesh just use PETSc's configuration on cygwin? As of 2008 or so, libMesh worked okay on Cygwin. I don't think it has any current users, though, and it's not in