<html>
<head>
<meta content="text/html; charset=ISO-8859-1"
http-equiv="Content-Type">
</head>
<body bgcolor="#FFFFFF" text="#000000">
<font face="Ubuntu">Matt,<br>
<br>
thank you! I will try to reduce the number of levels and see how
it goes. <br>
I asked about the speed since CG + Block Jacobi with ICC in each
block runs faster then CG + MG, so I thought I was missing
something.<br>
Could you please tell me how to get rid of Chebichev?<br>
<br>
Best,<br>
Michele<br>
<br>
</font>
<div class="moz-cite-prefix">On 08/13/2013 05:51 M, Matthew Knepley
wrote:<br>
</div>
<blockquote
cite="mid:CAMYG4GmM2oCuPZ8odWv_EbuJuS0n9u+B9dFU-oM6wN5KwUGkZQ@mail.gmail.com"
type="cite">
<div dir="ltr">On Tue, Aug 13, 2013 at 7:05 PM, Michele Rosso <span
dir="ltr"><<a moz-do-not-send="true"
href="mailto:mrosso@uci.edu" target="_blank">mrosso@uci.edu</a>></span>
wrote:<br>
<div class="gmail_extra">
<div class="gmail_quote">
<blockquote class="gmail_quote" style="margin:0 0 0
.8ex;border-left:1px #ccc solid;padding-left:1ex">
<div bgcolor="#FFFFFF" text="#000000"> <font
face="Ubuntu">Hi Matt,<br>
<br>
I attached the output of the commands you suggested.<br>
The options I used are:<br>
<br>
-log_summary -ksp_monitor -ksp_view
-ksp_converged_reason -pc_type mg -pc_mg_galerkin
-pc_mg_levels 5 -options_left<br>
</font></div>
</blockquote>
<div><br>
</div>
<div>The convergence is great. I notice that your coarse
solve takes no time. You could probably use fewer levels
for</div>
<div>this problem. For this problem there is no easy things
left I think. We are currently debating how you can
squeeze</div>
<div>something extra out of the smoother. Here you could
probably get rid of Chebychev and use only SOR.</div>
<div><br>
</div>
<div> Matt</div>
<div> </div>
<blockquote class="gmail_quote" style="margin:0 0 0
.8ex;border-left:1px #ccc solid;padding-left:1ex">
<div bgcolor="#FFFFFF" text="#000000"><font face="Ubuntu">
and here are the lines of codes where I setup the
solution process:<br>
<br>
call DMDACreate3d( PETSC_COMM_WORLD
, &<br>
& DMDA_BOUNDARY_PERIODIC ,
DMDA_BOUNDARY_PERIODIC, &<br>
& DMDA_BOUNDARY_PERIODIC ,
DMDA_STENCIL_STAR, &<br>
& N_Z , N_Y , N_X , N_B3 ,
N_B2 , 1_ip, 1_ip , 1_ip , &<br>
& NNZ ,NNY , NNX, da , ierr)<br>
<br>
<br>
! Create Global Vectors <br>
call DMCreateGlobalVector(da,b,ierr)<br>
call VecDuplicate(b,x,ierr)<br>
<br>
! Set initial guess for first use of the module to
0<br>
call VecSet(x,0.0_rp,ierr) <br>
<br>
! Create matrix <br>
call DMCreateMatrix(da,MATAIJ,A,ierr)<br>
<br>
! Create solver<br>
call KSPCreate(PETSC_COMM_WORLD,ksp,ierr) <br>
call KSPSetDM(ksp,da,ierr)<br>
call KSPSetDMActive(ksp,PETSC_FALSE,ierr)<br>
call
KSPSetOperators(ksp,A,A,SAME_NONZERO_PATTERN,ierr)<br>
call KSPSetType(ksp,KSPCG,ierr)<br>
call
KSPSetNormType(ksp,KSP_NORM_UNPRECONDITIONED,ierr)<br>
call
KSPSetInitialGuessNonzero(ksp,PETSC_TRUE,ierr)<br>
call KSPSetTolerances(ksp, tol
,PETSC_DEFAULT_DOUBLE_PRECISION,&<br>
&
PETSC_DEFAULT_DOUBLE_PRECISION,PETSC_DEFAULT_INTEGER,ierr)
<br>
<br>
! Nullspace removal <br>
call MatNullSpaceCreate(
PETSC_COMM_WORLD,PETSC_TRUE,PETSC_NULL_INTEGER,&<br>
&
PETSC_NULL_INTEGER,nullspace,ierr)<br>
call KSPSetNullspace(ksp,nullspace,ierr)<br>
call MatNullSpaceDestroy(nullspace,ierr) <br>
<br>
! To allow using option from command line<br>
call KSPSetFromOptions(ksp,ierr) <br>
<br>
<br>
Hope I did not omit anything useful.<br>
Thank you for your time.<br>
<br>
Best,<br>
Michele<br>
<br>
<br>
<br>
<br>
</font>
<div>On 08/13/2013 04:26 PM, Matthew Knepley wrote:<br>
</div>
<blockquote type="cite">
<div dir="ltr">On Tue, Aug 13, 2013 at 6:09 PM,
Michele Rosso <span dir="ltr"><<a
moz-do-not-send="true"
href="mailto:mrosso@uci.edu" target="_blank">mrosso@uci.edu</a>></span>
wrote:<br>
<div class="gmail_extra">
<div class="gmail_quote">
<blockquote class="gmail_quote" style="margin:0
0 0 .8ex;border-left:1px #ccc
solid;padding-left:1ex">
<div bgcolor="#FFFFFF" text="#000000"> <font
face="Ubuntu">Hi Karli,<br>
<br>
thank you for your hint: now it works. <br>
Now I would like to speed up the solution:
I was counting on increasing the number of
levels/the number of processors used, but
now I see I cannot do that.<br>
Do you have any hint to achieve better
speed?<br>
Thanks!<br>
</font></div>
</blockquote>
<div><br>
</div>
<div>"Better speed" is not very helpful for us,
and thus we cannot offer much help. You could</div>
<div><br>
</div>
<div> 1) Send the output of -log_summary
-ksp_monitor -ksp_view</div>
<div><br>
</div>
<div> 2) Describe the operator succintly</div>
<div><br>
</div>
<div> Matt</div>
<div> </div>
<blockquote class="gmail_quote" style="margin:0
0 0 .8ex;border-left:1px #ccc
solid;padding-left:1ex">
<div bgcolor="#FFFFFF" text="#000000"> <font
face="Ubuntu"> Best,<br>
Michele<br>
</font>
<blockquote type="cite">
<blockquote type="cite">
<blockquote type="cite">
<blockquote type="cite">
<blockquote type="cite">
<blockquote type="cite">
<blockquote type="cite">
<blockquote type="cite">
<blockquote type="cite">
<blockquote type="cite">
<blockquote type="cite">
<blockquote
type="cite">
<blockquote
type="cite">
<blockquote
type="cite">
<blockquote
type="cite">
<blockquote
type="cite"><br>
</blockquote>
</blockquote>
</blockquote>
</blockquote>
</blockquote>
</blockquote>
</blockquote>
</blockquote>
</blockquote>
</blockquote>
</blockquote>
</blockquote>
</blockquote>
</blockquote>
</blockquote>
</blockquote>
</div>
</blockquote>
</div>
</div>
</div>
</blockquote>
<br>
</div>
</blockquote>
</div>
<br>
<br clear="all">
<div><br>
</div>
-- <br>
What most experimenters take for granted before they begin
their experiments is infinitely more interesting than any
results to which their experiments lead.<br>
-- Norbert Wiener
</div>
</div>
</blockquote>
<br>
</body>
</html>