From antonios at mech.kth.se Wed Jun 3 03:31:00 2009 From: antonios at mech.kth.se (Antonios Monokrousos) Date: Wed, 03 Jun 2009 10:31:00 +0200 Subject: [Nek5000-users] Nek5000 pertubation mode Message-ID: <4A2634C4.5000504@mech.kth.se> Dear Paul, I finally found what the problem with my case was. It was the de-aliasing. The problem appears first in the subroutine advabp (file perturb.f) where subroutine convop is called (file navier1.f). It seems that de-aliasing is not working for the pertubation/linear mode (?) When I deactivate the de-aliasing (from the rea file) the result looks correct. I tested it also with channel flow and it was the same problem: In pertubation mode, with de-aliasing a forcing appears in a few specific elements. I haven't been able to understand exactly why this happens but I suspect it has to do with the size of the arrays. Best regards Antonios -- --------------------------------------------------------------------------------------------- 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 Wed Jun 3 04:49:38 2009 From: fischer at mcs.anl.gov (Paul Fischer) Date: Wed, 3 Jun 2009 04:49:38 -0500 (CDT) Subject: [Nek5000-users] Nek5000 pertubation mode In-Reply-To: <4A2634C4.5000504@mech.kth.se> References: <4A2634C4.5000504@mech.kth.se> Message-ID: Dear Antonio, Thanks for bringing this to my attention. I had similarly found this as the issue last week and had started to investigate when I got derailed by other projects. I'll try to sort this out and hopefully repair the svn repo version of the code today. Best regards, Paul On Wed, 3 Jun 2009, Antonios Monokrousos wrote: > Dear Paul, > I finally found what the problem with my case was. It was the > de-aliasing. The problem appears first in the subroutine advabp (file > perturb.f) where subroutine convop is called (file navier1.f). It seems > that de-aliasing is not working for the pertubation/linear mode (?) When > I deactivate the de-aliasing (from the rea file) the result looks correct. > I tested it also with channel flow and it was the same problem: In > pertubation mode, with de-aliasing a forcing appears in a few specific > elements. I haven't been able to understand exactly why this happens but > I suspect it has to do with the size of the arrays. > Best regards > Antonios > > > -- > --------------------------------------------------------------------------------------------- > 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 Wed Jun 3 07:20:18 2009 From: fischer at mcs.anl.gov (Paul Fischer) Date: Wed, 3 Jun 2009 07:20:18 -0500 (CDT) Subject: [Nek5000-users] Nek5000 pertubation mode In-Reply-To: <4A2634C4.5000504@mech.kth.se> References: <4A2634C4.5000504@mech.kth.se> Message-ID: Dear Antonios, If you can try the current version of the svn repo, the perturbation code should now work with dealiasing turned on. Moreover, you can automatically turn off the base-flow evolution by setting the number of perturbation modes (p31 in the .rea file) to be < 0. Thus, p31=-1 would give one perturbation mode, no base flow evolution. The base flow would be defined, e.g., in the useric. I include an example case in the attached tarfile, Orr-Sommerfeld evolution. Paul On Wed, 3 Jun 2009, Antonios Monokrousos wrote: > Dear Paul, > I finally found what the problem with my case was. It was the > de-aliasing. The problem appears first in the subroutine advabp (file > perturb.f) where subroutine convop is called (file navier1.f). It seems > that de-aliasing is not working for the pertubation/linear mode (?) When > I deactivate the de-aliasing (from the rea file) the result looks correct. > I tested it also with channel flow and it was the same problem: In > pertubation mode, with de-aliasing a forcing appears in a few specific > elements. I haven't been able to understand exactly why this happens but > I suspect it has to do with the size of the arrays. > Best regards > Antonios > > > -- > --------------------------------------------------------------------------------------------- > 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 > -------------- next part -------------- A non-text attachment was scrubbed... Name: os2d.tar Type: application/x-tar Size: 71680 bytes Desc: URL: From antonios at mech.kth.se Wed Jun 3 10:02:55 2009 From: antonios at mech.kth.se (Antonios Monokrousos) Date: Wed, 03 Jun 2009 17:02:55 +0200 Subject: [Nek5000-users] Nek5000 pertubation mode In-Reply-To: References: <4A2634C4.5000504@mech.kth.se> Message-ID: <4A26909F.1010500@mech.kth.se> Dear Paul, I tested the new version also with my case (blasius BL) and it works fine (with de-aliasing)! Many thanks! Antonios Paul Fischer wrote: > > > Dear Antonios, > > If you can try the current version of the svn repo, > the perturbation code should now work with dealiasing > turned on. > > Moreover, you can automatically turn off the base-flow > evolution by setting the number of perturbation modes > (p31 in the .rea file) to be < 0. Thus, p31=-1 would > give one perturbation mode, no base flow evolution. > The base flow would be defined, e.g., in the useric. > > I include an example case in the attached tarfile, > Orr-Sommerfeld evolution. > > Paul > > > On Wed, 3 Jun 2009, Antonios Monokrousos wrote: > >> Dear Paul, >> I finally found what the problem with my case was. It was the >> de-aliasing. The problem appears first in the subroutine advabp (file >> perturb.f) where subroutine convop is called (file navier1.f). It seems >> that de-aliasing is not working for the pertubation/linear mode (?) When >> I deactivate the de-aliasing (from the rea file) the result looks >> correct. >> I tested it also with channel flow and it was the same problem: In >> pertubation mode, with de-aliasing a forcing appears in a few specific >> elements. I haven't been able to understand exactly why this happens but >> I suspect it has to do with the size of the arrays. >> Best regards >> Antonios >> >> >> -- From fischer at mcs.anl.gov Wed Jun 3 10:27:41 2009 From: fischer at mcs.anl.gov (Paul Fischer) Date: Wed, 3 Jun 2009 10:27:41 -0500 (CDT) Subject: [Nek5000-users] Nek5000 pertubation mode In-Reply-To: <4A26909F.1010500@mech.kth.se> References: <4A2634C4.5000504@mech.kth.se> <4A26909F.1010500@mech.kth.se> Message-ID: great! On Wed, 3 Jun 2009, Antonios Monokrousos wrote: > Dear Paul, > I tested the new version also with my case (blasius BL) and it works > fine (with de-aliasing)! > Many thanks! > Antonios > > Paul Fischer wrote: >> >> >> Dear Antonios, >> >> If you can try the current version of the svn repo, >> the perturbation code should now work with dealiasing >> turned on. >> >> Moreover, you can automatically turn off the base-flow >> evolution by setting the number of perturbation modes >> (p31 in the .rea file) to be < 0. Thus, p31=-1 would >> give one perturbation mode, no base flow evolution. >> The base flow would be defined, e.g., in the useric. >> >> I include an example case in the attached tarfile, >> Orr-Sommerfeld evolution. >> >> Paul >> >> >> On Wed, 3 Jun 2009, Antonios Monokrousos wrote: >> >>> Dear Paul, >>> I finally found what the problem with my case was. It was the >>> de-aliasing. The problem appears first in the subroutine advabp (file >>> perturb.f) where subroutine convop is called (file navier1.f). It seems >>> that de-aliasing is not working for the pertubation/linear mode (?) When >>> I deactivate the de-aliasing (from the rea file) the result looks >>> correct. >>> I tested it also with channel flow and it was the same problem: In >>> pertubation mode, with de-aliasing a forcing appears in a few specific >>> elements. I haven't been able to understand exactly why this happens but >>> I suspect it has to do with the size of the arrays. >>> Best regards >>> Antonios >>> >>> >>> -- > From vanchisen at gmail.com Thu Jun 4 09:24:18 2009 From: vanchisen at gmail.com (=?GB2312?B?zfXWvrPJ?=) Date: Thu, 4 Jun 2009 22:24:18 +0800 Subject: [Nek5000-users] Fortran runtime error Message-ID: <32be724c0906040724x2915e5a7xc04f124182aaeff6@mail.gmail.com> Dear Paul, I have compile NEK5000 successful, But when I try to run the example turbChannel, it give me this error "* At line 505 of file drive2.f Fortran runtime error: No such file or directory*". I would like some one give me some advice to solve this problem. BTW, I'm a PhD student working on large eddy simulation of turbine blade heat transfer. Now I'm using a FVM code in unstructured grids and only have the basic knowledge about Spectral Element Method. Best regards Zhicheng Wang -------------- next part -------------- An HTML attachment was scrubbed... URL: From vanchisen at gmail.com Sat Jun 20 11:19:05 2009 From: vanchisen at gmail.com (=?GB2312?B?zfXWvrPJ?=) Date: Sun, 21 Jun 2009 00:19:05 +0800 Subject: [Nek5000-users] How to visualize the result by VisIt Message-ID: <32be724c0906200919p40adfb57j663ce13d839aeb2f@mail.gmail.com> Hi, I run the case 2D eddy_uv and have writtten the metadata file vis.nek5000 for visualization. But the metadata file can't be loaded into VisIt which gives error like this: VisIt could not read from the file /home/wzc/nek5/examples/eddy_uv/vis.nek5000. PS: The name of result file are: eddy_uv.fld01 ........ eddy_uv.fld11 The content in my vis.nek5000 is : NEK3D version: 1.0 filetemplate: eddy_uv.fld%11d firsttimestep: 1 numtimesteps: 11 and I use executable version of VisIt 1.11.2. Regards, Zhicheng Wang -------------- next part -------------- An HTML attachment was scrubbed... URL: From kerkemeier at lav.mavt.ethz.ch Sat Jun 20 12:54:15 2009 From: kerkemeier at lav.mavt.ethz.ch (Stefan Kerkemeier) Date: Sat, 20 Jun 2009 19:54:15 +0200 Subject: [Nek5000-users] How to visualize the result by VisIt In-Reply-To: <32be724c0906200919p40adfb57j663ce13d839aeb2f@mail.gmail.com> References: <32be724c0906200919p40adfb57j663ce13d839aeb2f@mail.gmail.com> Message-ID: <4A3D2247.2080106@lav.mavt.ethz.ch> Hi, are you using the ASCII output format (param(66) < 0)? I would recommend to switch to the binary format (param(66) > 0). For more details see here: https://nek5000.mcs.anl.gov/index.php/Parameters Anyway your filetemplate in metadata file seems to be wrong. Try the following: filetemplate: eddy_uv.fld%02d btw: the '%02d' means that two digits will follow. hth, Stefan ??? wrote: > Hi, > > I run the case 2D eddy_uv and have writtten the metadata file > vis.nek5000 for visualization. > But the metadata file can't be loaded into VisIt which gives error > like this: VisIt could not read from the file > /home/wzc/nek5/examples/eddy_uv/vis.nek5000. > > > PS: > The name of result file are: eddy_uv.fld01 ........ eddy_uv.fld11 > The content in my vis.nek5000 is : > NEK3D > version: 1.0 > filetemplate: eddy_uv.fld%11d > firsttimestep: 1 > numtimesteps: 11 > > and I use executable version of VisIt 1.11.2. > > > > Regards, > > Zhicheng Wang > > ------------------------------------------------------------------------ > > _______________________________________________ > Nek5000-users mailing list > Nek5000-users at lists.mcs.anl.gov > https://lists.mcs.anl.gov/mailman/listinfo/nek5000-users > From vanchisen at gmail.com Mon Jun 22 06:37:15 2009 From: vanchisen at gmail.com (=?GB2312?B?zfXWvrPJ?=) Date: Mon, 22 Jun 2009 19:37:15 +0800 Subject: [Nek5000-users] How to create mesh? Message-ID: <32be724c0906220437w3a8f5413u24de7069537d44e6@mail.gmail.com> Hi, I'm still don't understand how to create a mesh for nek5000, although I have read the user guide page for many times. The first problem is how to compile the tool such as genmap and genbox ? Secondly, can the mesh created by Gmsh(an opensource FEM mesh generator) be used by nek5000? Could anyone please give me some hints on these problems? Thanks! Regards, Zhicheng -------------- next part -------------- An HTML attachment was scrubbed... URL: From kerkemeier at lav.mavt.ethz.ch Mon Jun 22 06:53:54 2009 From: kerkemeier at lav.mavt.ethz.ch (Kerkemeier Stefan) Date: Mon, 22 Jun 2009 13:53:54 +0200 Subject: [Nek5000-users] How to create mesh? References: <32be724c0906220437w3a8f5413u24de7069537d44e6@mail.gmail.com> Message-ID: Hi, > The first problem is how to compile the tool such as genmap and genbox ? You can compile the tools using the maketools script (similar to makenek). > Secondly, can the mesh created by Gmsh(an opensource FEM mesh generator) be used by nek5000? No we do not support GMSH. Please note that GMSH is a tet-mesher and NEK5000 does only support quad/hexahedral elements. Currently we have the following options to generate a mesh: - genbox: simple box mesh - prenek: GUI which can be used to create more complicated meshes - by hand basically you have to provide the corner points of the elements + curved sides (check Wiki for more details) - MOAB: we are currently working on a MOAB interface to read meshes from Cubit (http://cubit.sandia.gov) hth, -Stefan