From fischer at mcs.anl.gov Thu Jan 8 10:42:01 2009 From: fischer at mcs.anl.gov (Paul Fischer) Date: Thu, 8 Jan 2009 10:42:01 -0600 (CST) Subject: [Nek5000-users] test Message-ID: test From johan at mech.kth.se Fri Jan 9 09:07:35 2009 From: johan at mech.kth.se (Johan Ohlsson) Date: Fri, 09 Jan 2009 16:07:35 +0100 Subject: [Nek5000-users] Reset Eig in hsmg setup fast Message-ID: <49676837.4030903@mech.kth.se> Hi Nek5000 users, Have you encountered this message: 'Reset Eig in hsmg setup fast' which appears in hsmg.f ? I am trying to run a problem with quite many elements and I get this message when N is greater than a critical number. As soon as N is below this number, everything is fine. It seems as some arrays get too big, which make the mg not working correctly and eventually I will have NaNs from the hmholtz. What exactly is going on ? Johan Ohlsson From fischer at mcs.anl.gov Fri Jan 9 09:10:40 2009 From: fischer at mcs.anl.gov (Paul Fischer) Date: Fri, 9 Jan 2009 09:10:40 -0600 (CST) Subject: [Nek5000-users] Reset Eig in hsmg setup fast In-Reply-To: <49676837.4030903@mech.kth.se> References: <49676837.4030903@mech.kth.se> Message-ID: Hi Johan, I've encountered that message in the past, but it's generally just informational. It's interesting that you're finding a correlation with code failure -- I certainly want to take a look at that. If you're using the current repo version of the code, can you email a tarfile with SIZEu, .rea, and .usr file directly to me? Thanks, Paul On Fri, 9 Jan 2009, Johan Ohlsson wrote: > Hi Nek5000 users, > > Have you encountered this message: 'Reset Eig in hsmg setup fast' which > appears in hsmg.f ? I am trying to run a problem with quite many > elements and I get this message when N is greater than a critical > number. As soon as N is below this number, everything is fine. It seems > as some arrays get too big, which make the mg not working correctly and > eventually I will have NaNs from the hmholtz. What exactly is going on ? > > Johan Ohlsson > > _______________________________________________ > Nek5000-users mailing list > Nek5000-users at lists.mcs.anl.gov > https://lists.mcs.anl.gov/mailman/listinfo/nek5000-users > From johan at mech.kth.se Fri Jan 9 08:08:30 2009 From: johan at mech.kth.se (Johan Ohlsson) Date: Fri, 09 Jan 2009 15:08:30 +0100 Subject: [Nek5000-users] Reset Eig in hsmg setup fast Message-ID: <49675A5E.5080501@mech.kth.se> Hi Nek5000 users, Have you encountered this message: 'Reset Eig in hsmg setup fast' which appears in hsmg.f ? I am trying to run a problem with quite many elements and I get this message when N is greater than a critical number. As soon as N is below this number, everything is fine. It seems as some arrays get too big, which make the mg not working correctly and eventually I will have NaNs from the hmholtz. What exactly is going on ? Johan Ohlsson From fischer at mcs.anl.gov Sat Jan 17 11:32:22 2009 From: fischer at mcs.anl.gov (Paul Fischer) Date: Sat, 17 Jan 2009 11:32:22 -0600 (CST) Subject: [Nek5000-users] coarse solver iteration count Message-ID: I think we should lower the default coarse grid solve iteration count to max of 100. In general, it's not too bad if it fails to converge. The code will typically get the velocity projected onto a div-free space within a few steps anyway. However, if it takes interminably long to solve the pressure than one can't know if the code has hung or is just iterating away. Thoughts? From johan at mech.kth.se Sat Jan 17 11:37:22 2009 From: johan at mech.kth.se (johan at mech.kth.se) Date: Sat, 17 Jan 2009 18:37:22 +0100 Subject: [Nek5000-users] coarse solver iteration count In-Reply-To: References: Message-ID: <20090117183722.xttt262uu4gw44co@www.mech.kth.se> I agree on this. At least if one can be sure that the not fully converged pressure will not cause a blow up later on. Johan Quoting Paul Fischer : > > I think we should lower the default coarse grid solve iteration > count to max of 100. > > In general, it's not too bad if it fails to converge. The code will > typically get the velocity projected onto a div-free space within a few > steps anyway. However, if it takes interminably long to solve > the pressure than one can't know if the code has hung or is just > iterating away. > > Thoughts? > > _______________________________________________ > Nek5000-users mailing list > Nek5000-users at lists.mcs.anl.gov > https://lists.mcs.anl.gov/mailman/listinfo/nek5000-users ---------------------------------------------------------------- This message was sent using IMP, the Internet Messaging Program. From antonios at mech.kth.se Thu Jan 22 07:30:56 2009 From: antonios at mech.kth.se (Antonios Monokrousos) Date: Thu, 22 Jan 2009 14:30:56 +0100 Subject: [Nek5000-users] path mentioned on wiki page wrong Message-ID: <49787510.3030503@mech.kth.se> Hello fellow nek5000 users! I just thought to make my small contribution: The path in the wiki page is wrong: svn checkout https://svn.mcs.anl.gov/repos/nek5/tags/latest The forlder latest does not exist. Regards to all! --------------------------------------------------------------------------------------------- Antonios Monokrousos Department of Mechanics email: antonios at mech.kth.se Royal Institute of Technology (KTH) phone: +46 8 790 68 76 Osquars Backe 18 telefax: +46 8 796 98 50 SE-100 44 Stockholm, SWEDEN From fischer at mcs.anl.gov Thu Jan 22 09:51:18 2009 From: fischer at mcs.anl.gov (Paul Fischer) Date: Thu, 22 Jan 2009 09:51:18 -0600 (CST) Subject: [Nek5000-users] path mentioned on wiki page wrong In-Reply-To: <49787510.3030503@mech.kth.se> References: <49787510.3030503@mech.kth.se> Message-ID: Thanks Antonios. I see this also. In the meantime, if you'd like to check out the code, the following svn command will work: svn co https://svn.mcs.anl.gov/repos/nek5 nek5_svn Best regards, Paul On Thu, 22 Jan 2009, Antonios Monokrousos wrote: > Hello fellow nek5000 users! > I just thought to make my small contribution: > The path in the wiki page is wrong: > svn checkout https://svn.mcs.anl.gov/repos/nek5/tags/latest > The forlder latest does not exist. > Regards to all! > > --------------------------------------------------------------------------------------------- > Antonios Monokrousos > Department of Mechanics email: antonios at mech.kth.se > Royal Institute of Technology (KTH) phone: +46 8 790 68 76 > Osquars Backe 18 telefax: +46 8 796 98 50 > SE-100 44 Stockholm, SWEDEN > _______________________________________________ > Nek5000-users mailing list > Nek5000-users at lists.mcs.anl.gov > https://lists.mcs.anl.gov/mailman/listinfo/nek5000-users >