On Oct 2, 2013, at 12:28 AM, Jed Brown wrote:
> "Mark F. Adams" writes:
>> run3.txt uses:
>>
>> -ksp_type richardson
>>
>> This is bad and I doubt anyone recommended it intentionally.
Hell this is normal multigrid without a Krylov accelerator. Under normal
circumstances with geometric
"Mark F. Adams" writes:
> run3.txt uses:
>
> -ksp_type richardson
>
> This is bad and I doubt anyone recommended it intentionally.
I would have expected FGMRES, but Barry likes Krylov smoothers and
Richardson is one of a few methods that can tolerate nonlinear
preconditioners.
> You also have,
Hi guys,
I just did something by pure guessing which seems to work and I want
to make sure its the right thing!
I have a specific layout for my vectors that look like this
--
| ghost values | local values | ghost values |
On Tue, 1 Oct 2013, Mengda Wu wrote:
> Hi Satish,
>
>Thanks a lot for your reply!
>
>Is there a way to use the mingw64 which is not shipped with cygwin? For
> example, use windows native paths when testing the compiler.
Nope - I don't think it will work. Our build tools are primarily wo
On Wed, 2 Oct 2013, at 13:38, Jed Brown wrote:
> Gryllida writes:
>
> > I have a parallel program which reads data and makes a call to a
> > subroutine that uses petsc. It only needs to read data once. Is it
> > possible to implement that using MPI and then initialise petsc in the
> > subroutine?
On Oct 1, 2013, at 7:02 PM, Michele Rosso wrote:
> Barry,
>
> I repeated the previous runs since I noticed that it was not using the options
>
> -mg_levels_ksp_max_it 3
>
> They are faster then before now but still slower than my initial test and
> anyway the solution time increases consider
Gryllida writes:
> I have a parallel program which reads data and makes a call to a
> subroutine that uses petsc. It only needs to read data once. Is it
> possible to implement that using MPI and then initialise petsc in the
> subroutine? Or do I need to initialise petsc first (and use MPI calls
I have a parallel program which reads data and makes a call to a subroutine
that uses petsc. It only needs to read data once. Is it possible to implement
that using MPI and then initialise petsc in the subroutine? Or do I need to
initialise petsc first (and use MPI calls with PETSC_COMM_WORLD)?
Thanks. Unacceptably poor convergence of multigrid here.
What kind of discretization are you using for your "projection method"? Is is
cell centered? What kind of boundary conditions?
If you are using cell-centered (piece-wise constant) then the interpolation
you are getting is all wro
S V N Vishwanathan writes:
>> You can use MatTransposeMatMult with an MPIAIJ and a MPIDENSE. (This
>> is the standard case mentioned earlier, and I don't understand what
>> Elemental is offering you in this situation. Do you have other dense
>> matrices running around?)
>
> Perhaps I should giv
Hi Satish,
Thanks a lot for your reply!
Is there a way to use the mingw64 which is not shipped with cygwin? For
example, use windows native paths when testing the compiler.
Thanks,
Mengda
On Tue, Oct 1, 2013 at 5:55 PM, Satish Balay wrote:
> Looks like this mingw install is not from cy
> You can use MatTransposeMatMult with an MPIAIJ and a MPIDENSE. (This
> is the standard case mentioned earlier, and I don't understand what
> Elemental is offering you in this situation. Do you have other dense
> matrices running around?)
Perhaps I should give a bit of background on what we ar
On Tue, Oct 1, 2013 at 7:21 PM, Gryllida wrote:
> >On Wed, 1 Sep 2010 20:54:54 +, "Li, Zhisong (lizs)" mail.uc.edu> wrote:
> >> Hi, Petsc Team,
> >>
> >> I wonder if Petsc has any function like MPI_Bcast() which can
> >> broadcast a value to all processes. If we directly add MPI functions
>
Thanks Matt.
On Tue, Oct 1, 2013 at 5:26 PM, Matthew Knepley wrote:
> On Tue, Oct 1, 2013 at 7:02 PM, Mohammad Mirzadeh
> wrote:
>>
>> > It would be half the messages. Why are these vectors separate in the
>> > first place?
>>
>> I had not worked with block vectors before so just convenience i
On Tue, Oct 1, 2013 at 7:02 PM, Mohammad Mirzadeh wrote:
> > It would be half the messages. Why are these vectors separate in the
> > first place?
>
> I had not worked with block vectors before so just convenience i
> guess! When working with blocked vectors, how are items stored? Is it
> like fx
>On Wed, 1 Sep 2010 20:54:54 +, "Li, Zhisong (lizs)"
>wrote:
>> Hi, Petsc Team,
>>
>> I wonder if Petsc has any function like MPI_Bcast() which can
>> broadcast a value to all processes. If we directly add MPI functions
>> into the Petsc program, the MPI datatype may be incompatible with th
> It would be half the messages. Why are these vectors separate in the
> first place?
I had not worked with block vectors before so just convenience i
guess! When working with blocked vectors, how are items stored? Is it
like fxx[0], fyy[0], fxx[1], fyy[1], ... or is it like
fxx[0],fxx[1],...,fyy
Joon Hee Choi writes:
> - Original Message -
> From: "Jack Poulson"
> To: d...@libelemental.org, "Jed Brown"
> Cc: "Joon Hee Choi" , "Hong Zhang" ,
> "PETSc users list"
> Sent: Tuesday, October 1, 2013 6:15:23 PM
> Subject: Re: [d...@libelemental.org] Re: [petsc-users] The product of
S V N Vishwanathan writes:
> Hi
>
>> > I am wondering whether I can multiply MATELEMENTAL and MATAIJ
>> > matrices. If I cannot multiply those, could you build the
>> > multiplication?
>>
>> What are the dimensions of the matrices?
>>
>> Jack, does Clique support multiplication of sparse matric
Mohammad Mirzadeh writes:
> I assume I'm getting this since they somehow share a VecScatter
> internally because of VecDuplicate? If so how can I fix this?
Hmm, there is no VecScatterDuplicate so I think you'd have to make the
vectors independently.
> Also as a side question, if I stored fxx and
- Original Message -
From: "Jack Poulson"
To: d...@libelemental.org, "Jed Brown"
Cc: "Joon Hee Choi" , "Hong Zhang" ,
"PETSc users list"
Sent: Tuesday, October 1, 2013 6:15:23 PM
Subject: Re: [d...@libelemental.org] Re: [petsc-users] The product of two
MPIDENSE matrices
On 10/01/201
Hi guys,
I have two ghosted vectors fxx, fyy which should store derivatives of
a quantity f. They are both obtained via calling VecDuplicate on f. To
minimize communications when computing both fxx and fyy I do the
following:
// loop over boundary points and compute fxx and fyy
VecGhostUpdateBeg
On 10/01/2013 03:04 PM, S V N Vishwanathan wrote:
> Hi
>
>>> I am wondering whether I can multiply MATELEMENTAL and MATAIJ
>>> matrices. If I cannot multiply those, could you build the
>>> multiplication?
>>
>> What are the dimensions of the matrices?
>>
>> Jack, does Clique support multiplication
Hi
> > I am wondering whether I can multiply MATELEMENTAL and MATAIJ
> > matrices. If I cannot multiply those, could you build the
> > multiplication?
>
> What are the dimensions of the matrices?
>
> Jack, does Clique support multiplication of sparse matrices with
> [MC,MR] dense matrices?
The
mingw64 from cygwin[32] appears to work fine with /tmp.
Satish
balay@msnehalem2 ~/junk
$ which x86_64-w64-mingw32-gcc.exe
/usr/bin/x86_64-w64-mingw32-gcc.exe
balay@msnehalem2 ~/junk
$ cygcheck -f /usr/bin/x86_64-w64-mingw32-gcc.exe
mingw64-x86_64-gcc-core-4.7.3-1
balay@msnehalem2 ~/jun
Looks like this mingw install is not from cygwin - and does not work with
cygwin paths.
Suggest using mingw from cygwin - or install and use cygwin64 [which has
non-mingw gcc-64]
You might need the following patch
https://bitbucket.org/petsc/petsc/commits/e4773d96ef73f1e0b315c5b0194cae499d71e207
Hello all,
I am trying to compile 64 bit version of petsc-3.4.2 using MinGW-w64
(latest version 4.8.1 installed with
http://sourceforge.net/projects/mingwbuilds/files/mingw-builds-install/mingw-builds-install.exe/download)
on Windows 7 64 bit. I have cygwin, python (under cygwin) installed.
Joon Hee Choi writes:
> Dear Petsc Developers,
>
> Hello.
> I am wondering whether I can multiply MATELEMENTAL and MATAIJ matrices. If I
> cannot multiply those, could you build the multiplication?
What are the dimensions of the matrices?
Jack, does Clique support multiplication of sparse mat
Dear Petsc Developers,
Hello.
I am wondering whether I can multiply MATELEMENTAL and MATAIJ matrices. If I
cannot multiply those, could you build the multiplication?
Thank you.
Joon
- Original Message -
From: "Joon Hee Choi"
To: "Jack Poulson"
Cc: "PETSc users list" , "dev"
, "Jed
On 10/1/13 9:45 PM, Jed Brown wrote:
Anton Popov writes:
Dear all,
I guess MatCreateSchurComplement is somehow broken in PETSc 3.4.2
You can fix it by calling KSPMatRegisterAll() before
MatCreateSchurComplement,
Thanks, Jed. It solves the problem.
and the issue is now fixed in 'maint' (i
Anton Popov writes:
> Dear all,
>
> I guess MatCreateSchurComplement is somehow broken in PETSc 3.4.2
You can fix it by calling KSPMatRegisterAll() before
MatCreateSchurComplement, and the issue is now fixed in 'maint' (it was
accidentally only fixed in 'master').
This problem came from simplif
Dear all,
I guess MatCreateSchurComplement is somehow broken in PETSc 3.4.2
Attached is the modified ex23 from src/ksp/ksp/examples/tutorials
I just copied system matrix three times and created a Schur complement
object.
However mpirun -n 4 ./ex23 gives: Unknown Mat type given: schurcompleme
I doubt very much that the ISO C bindings allow one to pass a Fortran
pointer through a C function which is what we need to do so I'd be surprised if
the ISO C bindings will save the day. But you are welcome to try.
Barry
On Oct 1, 2013, at 7:01 AM, Jed Brown wrote:
> Barry Smith wr
On Tue, Oct 1, 2013 at 2:01 PM, Jed Brown wrote:
> Barry Smith writes:
> >I put the example with VecGetArray() into petsc-dev and if you send
> >me one with DMDAVecGetArray() I'll add that as well. Since the
> >derived type has to appear explicitly in the code I can't put
> >som
Barry Smith writes:
>I put the example with VecGetArray() into petsc-dev and if you send
>me one with DMDAVecGetArray() I'll add that as well. Since the
>derived type has to appear explicitly in the code I can't put
>something in the library only in examples that people can copy a
On Oct 1, 2013, at 2:18 AM, Christophe Ortiz wrote:
> Hi Barry,
>
> Thanks for the piece of code. I will try to implement it. My problem is 1D,
> so should be ok. If I have some pb I'll let you know.
>
> If this work, do you plan to include it in next version of PETSc ?
I put the example
Hi Barry,
Thanks for the piece of code. I will try to implement it. My problem is 1D,
so should be ok. If I have some pb I'll let you know.
If this work, do you plan to include it in next version of PETSc ?
Christophe
On Mon, Sep 30, 2013 at 2:04 PM, Barry Smith wrote:
>
> On Sep 30, 2013, at
37 matches
Mail list logo
