FieldConvert error from calculating vorticity and Qcriterion
Hi nektar++ users and developers, I 'm using the latest nektar++ version. However when running the TurbChFl_3DH1D.xml example and using FieldConvert calculating vorticity or Qcriterion, I got errors as "ERROR: Element u_33 is 0 from dgetrf". I have checked the velocity and pressure field, it is correct, so if there is some inconsistency in FieldConvert for 1D homogeneous problem?
Hi Zhicheng, We have recently been trying to homogenise and update the field convert utility. Could I ask if you have tried it on the branch feature/FieldConvertCleanUp. Hopefully it will be working in this branch. We hope to get this branch into Master soon so it would also be good for us to know if you are experiencing problems here so we get things working as smoothly as possible. Cheers, Spencer. On 25 Apr 2015, at 14:23, Zhicheng Wang <wangzhicheng09@gmail.com<mailto:wangzhicheng09@gmail.com>> wrote: Hi nektar++ users and developers, I 'm using the latest nektar++ version. However when running the TurbChFl_3DH1D.xml example and using FieldConvert calculating vorticity or Qcriterion, I got errors as "ERROR: Element u_33 is 0 from dgetrf". I have checked the velocity and pressure field, it is correct, so if there is some inconsistency in FieldConvert for 1D homogeneous problem? _______________________________________________ Nektar-users mailing list Nektar-users@imperial.ac.uk<mailto:Nektar-users@imperial.ac.uk> https://mailman.ic.ac.uk/mailman/listinfo/nektar-users Spencer Sherwin McLaren Racing/Royal Academy of Engineering Research Chair, Professor of Computational Fluid Mechanics, Department of Aeronautics, Imperial College London South Kensington Campus London SW7 2AZ s.sherwin@imperial.ac.uk<mailto:s.sherwin@imperial.ac.uk> +44 (0) 20 759 45052
Hi Spencer, Thank you very much for your reply. but could you please tell me the link to branch feature/FieldConvertCleanUp ? I have browse the whole site, but have no idea on reaching this branch. zhicheng On Sat, Apr 25, 2015 at 11:19 PM Sherwin, Spencer J < s.sherwin@imperial.ac.uk> wrote:
Hi Zhicheng,
We have recently been trying to homogenise and update the field convert utility. Could I ask if you have tried it on the branch feature/FieldConvertCleanUp.
Hopefully it will be working in this branch. We hope to get this branch into Master soon so it would also be good for us to know if you are experiencing problems here so we get things working as smoothly as possible.
Cheers, Spencer.
On 25 Apr 2015, at 14:23, Zhicheng Wang <wangzhicheng09@gmail.com> wrote:
Hi nektar++ users and developers, I 'm using the latest nektar++ version. However when running the TurbChFl_3DH1D.xml example and using FieldConvert calculating vorticity or Qcriterion, I got errors as "ERROR: Element u_33 is 0 from dgetrf". I have checked the velocity and pressure field, it is correct, so if there is some inconsistency in FieldConvert for 1D homogeneous problem?
_______________________________________________ Nektar-users mailing list Nektar-users@imperial.ac.uk https://mailman.ic.ac.uk/mailman/listinfo/nektar-users
Spencer Sherwin McLaren Racing/Royal Academy of Engineering Research Chair, Professor of Computational Fluid Mechanics, Department of Aeronautics, Imperial College London South Kensington Campus London SW7 2AZ
s.sherwin@imperial.ac.uk +44 (0) 20 759 45052
Dear Zhicheng, When you download using the code from the gitlab repository you can then select a branch by typing git checkout feature/FieldConvertCleanUP if you type ‘git branch’ it tell you which branch you are on. Have you been using git or the tar ball distribution? Cheers Spencer. On 25 Apr 2015, at 17:48, Zhicheng Wang <wangzhicheng09@gmail.com<mailto:wangzhicheng09@gmail.com>> wrote: Hi Spencer, Thank you very much for your reply. but could you please tell me the link to branch feature/FieldConvertCleanUp ? I have browse the whole site, but have no idea on reaching this branch. zhicheng On Sat, Apr 25, 2015 at 11:19 PM Sherwin, Spencer J <s.sherwin@imperial.ac.uk<mailto:s.sherwin@imperial.ac.uk>> wrote: Hi Zhicheng, We have recently been trying to homogenise and update the field convert utility. Could I ask if you have tried it on the branch feature/FieldConvertCleanUp. Hopefully it will be working in this branch. We hope to get this branch into Master soon so it would also be good for us to know if you are experiencing problems here so we get things working as smoothly as possible. Cheers, Spencer. On 25 Apr 2015, at 14:23, Zhicheng Wang <wangzhicheng09@gmail.com<mailto:wangzhicheng09@gmail.com>> wrote: Hi nektar++ users and developers, I 'm using the latest nektar++ version. However when running the TurbChFl_3DH1D.xml example and using FieldConvert calculating vorticity or Qcriterion, I got errors as "ERROR: Element u_33 is 0 from dgetrf". I have checked the velocity and pressure field, it is correct, so if there is some inconsistency in FieldConvert for 1D homogeneous problem? _______________________________________________ Nektar-users mailing list Nektar-users@imperial.ac.uk<mailto:Nektar-users@imperial.ac.uk> https://mailman.ic.ac.uk/mailman/listinfo/nektar-users Spencer Sherwin McLaren Racing/Royal Academy of Engineering Research Chair, Professor of Computational Fluid Mechanics, Department of Aeronautics, Imperial College London South Kensington Campus London SW7 2AZ s.sherwin@imperial.ac.uk<mailto:s.sherwin@imperial.ac.uk> +44 (0) 20 759 45052 Spencer Sherwin McLaren Racing/Royal Academy of Engineering Research Chair, Professor of Computational Fluid Mechanics, Department of Aeronautics, Imperial College London South Kensington Campus London SW7 2AZ s.sherwin@imperial.ac.uk<mailto:s.sherwin@imperial.ac.uk> +44 (0) 20 759 45052
Hi Spencer, I have checked the latest branch feature/FieldConvertCleanUP, but unfortunately it still doesn't work. On Sun, Apr 26, 2015 at 1:03 AM Sherwin, Spencer J <s.sherwin@imperial.ac.uk> wrote:
Dear Zhicheng,
When you download using the code from the gitlab repository you can then select a branch by typing
git checkout feature/FieldConvertCleanUP
if you type ‘git branch’ it tell you which branch you are on.
Have you been using git or the tar ball distribution?
Cheers Spencer.
On 25 Apr 2015, at 17:48, Zhicheng Wang <wangzhicheng09@gmail.com> wrote:
Hi Spencer, Thank you very much for your reply. but could you please tell me the link to branch feature/FieldConvertCleanUp ? I have browse the whole site, but have no idea on reaching this branch.
zhicheng
On Sat, Apr 25, 2015 at 11:19 PM Sherwin, Spencer J < s.sherwin@imperial.ac.uk> wrote:
Hi Zhicheng,
We have recently been trying to homogenise and update the field convert utility. Could I ask if you have tried it on the branch feature/FieldConvertCleanUp.
Hopefully it will be working in this branch. We hope to get this branch into Master soon so it would also be good for us to know if you are experiencing problems here so we get things working as smoothly as possible.
Cheers, Spencer.
On 25 Apr 2015, at 14:23, Zhicheng Wang <wangzhicheng09@gmail.com> wrote:
Hi nektar++ users and developers, I 'm using the latest nektar++ version. However when running the TurbChFl_3DH1D.xml example and using FieldConvert calculating vorticity or Qcriterion, I got errors as "ERROR: Element u_33 is 0 from dgetrf". I have checked the velocity and pressure field, it is correct, so if there is some inconsistency in FieldConvert for 1D homogeneous problem?
_______________________________________________ Nektar-users mailing list Nektar-users@imperial.ac.uk https://mailman.ic.ac.uk/mailman/listinfo/nektar-users
Spencer Sherwin McLaren Racing/Royal Academy of Engineering Research Chair, Professor of Computational Fluid Mechanics, Department of Aeronautics, Imperial College London South Kensington Campus London SW7 2AZ
s.sherwin@imperial.ac.uk +44 (0) 20 759 45052
Spencer Sherwin McLaren Racing/Royal Academy of Engineering Research Chair, Professor of Computational Fluid Mechanics, Department of Aeronautics, Imperial College London South Kensington Campus London SW7 2AZ
s.sherwin@imperial.ac.uk +44 (0) 20 759 45052
Hi Zhicheng, I've taken a look this morning but was unable to reproduce the error on my machine. From the feature/FieldConvertCleanUp branch, I did the following: 1. Compiled with FFTW/MPI enabled. 2. Run solver: mpirun -n 6 IncNavierStokesSolver-g TurbChFl_3DH1D.xml 3. Run FieldConvert: FieldConvert-g -m QCriterion TurbChFl_3DH1D.xml TurbChFl_3DH1D_1.chk out.vtu This produces a field that I can visualise in paraview. What's the process that you're using which gives your the error message? Cheers, Dave On 2015-04-27 03:39, Zhicheng Wang wrote:
Hi Spencer, I have checked the latest branch feature/FieldConvertCleanUP, but unfortunately it still doesn't work.
On Sun, Apr 26, 2015 at 1:03 AM Sherwin, Spencer J <s.sherwin@imperial.ac.uk> wrote:
Dear Zhicheng,
When you download using the code from the gitlab repository you can then select a branch by typing
git checkout feature/FieldConvertCleanUP
if you type ‘git branch’ it tell you which branch you are on.
Have you been using git or the tar ball distribution?
Cheers
Spencer.
On 25 Apr 2015, at 17:48, Zhicheng Wang <wangzhicheng09@gmail.com> wrote:
Hi Spencer, Thank you very much for your reply. but could you please tell me the link to branch feature/FieldConvertCleanUp ? I have browse the whole site, but have no idea on reaching this branch.
zhicheng
On Sat, Apr 25, 2015 at 11:19 PM Sherwin, Spencer J <s.sherwin@imperial.ac.uk> wrote:
Hi Zhicheng,
We have recently been trying to homogenise and update the field convert utility. Could I ask if you have tried it on the branch feature/FieldConvertCleanUp.
Hopefully it will be working in this branch. We hope to get this branch into Master soon so it would also be good for us to know if you are experiencing problems here so we get things working as smoothly as possible.
Cheers, Spencer.
On 25 Apr 2015, at 14:23, Zhicheng Wang <wangzhicheng09@gmail.com> wrote:
Hi nektar++ users and developers, I 'm using the latest nektar++ version. However when running the TurbChFl_3DH1D.xml example and using FieldConvert calculating vorticity or Qcriterion, I got errors as "ERROR: Element u_33 is 0 from dgetrf". I have checked the velocity and pressure field, it is correct, so if there is some inconsistency in FieldConvert for 1D homogeneous problem?
_______________________________________________ Nektar-users mailing list Nektar-users@imperial.ac.uk https://mailman.ic.ac.uk/mailman/listinfo/nektar-users [1]
Spencer Sherwin McLaren Racing/Royal Academy of Engineering Research Chair, Professor of Computational Fluid Mechanics, Department of Aeronautics, Imperial College London South Kensington Campus London SW7 2AZ
s.sherwin@imperial.ac.uk +44 (0) 20 759 45052
Spencer Sherwin McLaren Racing/Royal Academy of Engineering Research Chair, Professor of Computational Fluid Mechanics, Department of Aeronautics, Imperial College London South Kensington Campus London SW7 2AZ
s.sherwin@imperial.ac.uk +44 (0) 20 759 45052
Links: ------ [1] https://mailman.ic.ac.uk/mailman/listinfo/nektar-users
_______________________________________________ Nektar-users mailing list Nektar-users@imperial.ac.uk https://mailman.ic.ac.uk/mailman/listinfo/nektar-users
-- David Moxey (Research Associate) d.moxey@imperial.ac.uk | www.imperial.ac.uk/people/d.moxey Room 363, Department of Aeronautics, Imperial College London, London, SW7 2AZ, UK.
Hi Dave, Thanks for your help. I should have state the problem more clearly. actually I did a minor change in the TurbChFl_3DH1D.xml. I increased the HomoModesZ parameter to 128, then ./FieldConvert -m QCriterion TurbChFl_3DH1D.xml TurbChFl_3DH1D_1.chk TurbChFl_3DH1D_Q.vtu reports errors. But after using command ./FieldConvert TurbChFl_3DH1D.xml TurbChFl_3DH1D_1.chk TurbChFl_3DH1D_1.vtu, and check the *.vtu file, I find both of the velocity and pressure looks OK. So it seems the errors come from the conditions that HomoModes>=128. zhicheng On Mon, Apr 27, 2015 at 7:06 PM David Moxey <d.moxey@imperial.ac.uk> wrote:
Hi Zhicheng,
I've taken a look this morning but was unable to reproduce the error on my machine. From the feature/FieldConvertCleanUp branch, I did the following:
1. Compiled with FFTW/MPI enabled. 2. Run solver: mpirun -n 6 IncNavierStokesSolver-g TurbChFl_3DH1D.xml 3. Run FieldConvert: FieldConvert-g -m QCriterion TurbChFl_3DH1D.xml TurbChFl_3DH1D_1.chk out.vtu
This produces a field that I can visualise in paraview. What's the process that you're using which gives your the error message?
Cheers,
Dave
On 2015-04-27 03:39, Zhicheng Wang wrote:
Hi Spencer, I have checked the latest branch feature/FieldConvertCleanUP, but unfortunately it still doesn't work.
On Sun, Apr 26, 2015 at 1:03 AM Sherwin, Spencer J <s.sherwin@imperial.ac.uk> wrote:
Dear Zhicheng,
When you download using the code from the gitlab repository you can then select a branch by typing
git checkout feature/FieldConvertCleanUP
if you type ‘git branch’ it tell you which branch you are on.
Have you been using git or the tar ball distribution?
Cheers
Spencer.
On 25 Apr 2015, at 17:48, Zhicheng Wang <wangzhicheng09@gmail.com> wrote:
Hi Spencer, Thank you very much for your reply. but could you please tell me the link to branch feature/FieldConvertCleanUp ? I have browse the whole site, but have no idea on reaching this branch.
zhicheng
On Sat, Apr 25, 2015 at 11:19 PM Sherwin, Spencer J <s.sherwin@imperial.ac.uk> wrote:
Hi Zhicheng,
We have recently been trying to homogenise and update the field convert utility. Could I ask if you have tried it on the branch feature/FieldConvertCleanUp.
Hopefully it will be working in this branch. We hope to get this branch into Master soon so it would also be good for us to know if you are experiencing problems here so we get things working as smoothly as possible.
Cheers, Spencer.
On 25 Apr 2015, at 14:23, Zhicheng Wang <wangzhicheng09@gmail.com> wrote:
Hi nektar++ users and developers, I 'm using the latest nektar++ version. However when running the TurbChFl_3DH1D.xml example and using FieldConvert calculating vorticity or Qcriterion, I got errors as "ERROR: Element u_33 is 0 from dgetrf". I have checked the velocity and pressure field, it is correct, so if there is some inconsistency in FieldConvert for 1D homogeneous problem?
_______________________________________________ Nektar-users mailing list Nektar-users@imperial.ac.uk https://mailman.ic.ac.uk/mailman/listinfo/nektar-users [1]
Spencer Sherwin McLaren Racing/Royal Academy of Engineering Research Chair, Professor of Computational Fluid Mechanics, Department of Aeronautics, Imperial College London South Kensington Campus London SW7 2AZ
s.sherwin@imperial.ac.uk +44 (0) 20 759 45052
Spencer Sherwin McLaren Racing/Royal Academy of Engineering Research Chair, Professor of Computational Fluid Mechanics, Department of Aeronautics, Imperial College London South Kensington Campus London SW7 2AZ
s.sherwin@imperial.ac.uk +44 (0) 20 759 45052
Links: ------ [1] https://mailman.ic.ac.uk/mailman/listinfo/nektar-users
_______________________________________________ Nektar-users mailing list Nektar-users@imperial.ac.uk https://mailman.ic.ac.uk/mailman/listinfo/nektar-users
-- David Moxey (Research Associate) d.moxey@imperial.ac.uk | www.imperial.ac.uk/people/d.moxey
Room 363, Department of Aeronautics, Imperial College London, London, SW7 2AZ, UK.
Hi Zhicheng, Okay, so now I can confirm there's a problem -- changing to 128 causes me to get all sorts of problems. It looks like there's a memory leak somewhere. Compiling in debug mode I've got malloc and free errors. Oddly, the code runs fine if I select the default BLAS/LAPACK reference implementation vs. OpenBLAS, which has all sorts of problems. However I get a field full of NaNs, so something is clearly wrong. I'll try to track it down and get back to you. Cheers, Dave On 2015-04-27 14:18, Zhicheng Wang wrote:
Hi Dave, Thanks for your help. I should have state the problem more clearly. actually I did a minor change in the TurbChFl_3DH1D.xml. I increased the HomoModesZ parameter to 128, then ./FieldConvert -m QCriterion TurbChFl_3DH1D.xml TurbChFl_3DH1D_1.chk TurbChFl_3DH1D_Q.vtu reports errors. But after using command ./FieldConvert TurbChFl_3DH1D.xml TurbChFl_3DH1D_1.chk TurbChFl_3DH1D_1.vtu, and check the *.vtu file, I find both of the velocity and pressure looks OK. So it seems the errors come from the conditions that HomoModes>=128.
zhicheng
On Mon, Apr 27, 2015 at 7:06 PM David Moxey <d.moxey@imperial.ac.uk> wrote:
Hi Zhicheng,
I've taken a look this morning but was unable to reproduce the error on my machine. From the feature/FieldConvertCleanUp branch, I did the following:
1. Compiled with FFTW/MPI enabled. 2. Run solver: mpirun -n 6 IncNavierStokesSolver-g TurbChFl_3DH1D.xml 3. Run FieldConvert: FieldConvert-g -m QCriterion TurbChFl_3DH1D.xml TurbChFl_3DH1D_1.chk out.vtu
This produces a field that I can visualise in paraview. What's the process that you're using which gives your the error message?
Cheers,
Dave
On 2015-04-27 03:39, Zhicheng Wang wrote:
Hi Spencer, I have checked the latest branch feature/FieldConvertCleanUP, but unfortunately it still doesn't work.
On Sun, Apr 26, 2015 at 1:03 AM Sherwin, Spencer J <s.sherwin@imperial.ac.uk> wrote:
Dear Zhicheng,
When you download using the code from the gitlab repository you can then select a branch by typing
git checkout feature/FieldConvertCleanUP
if you type ‘git branch’ it tell you which branch you are on.
Have you been using git or the tar ball distribution?
Cheers
Spencer.
On 25 Apr 2015, at 17:48, Zhicheng Wang <wangzhicheng09@gmail.com> wrote:
Hi Spencer, Thank you very much for your reply. but could you please tell me the link to branch feature/FieldConvertCleanUp ? I have browse the whole site, but have no idea on reaching this branch.
zhicheng
On Sat, Apr 25, 2015 at 11:19 PM Sherwin, Spencer J <s.sherwin@imperial.ac.uk> wrote:
Hi Zhicheng,
We have recently been trying to homogenise and update the field convert utility. Could I ask if you have tried it on the branch feature/FieldConvertCleanUp.
Hopefully it will be working in this branch. We hope to get this branch into Master soon so it would also be good for us to know if you are experiencing problems here so we get things working as smoothly as possible.
Cheers, Spencer.
On 25 Apr 2015, at 14:23, Zhicheng Wang <wangzhicheng09@gmail.com> wrote:
Hi nektar++ users and developers, I 'm using the latest nektar++ version. However when running the TurbChFl_3DH1D.xml example and using FieldConvert calculating vorticity or Qcriterion, I got errors as "ERROR: Element u_33 is 0 from dgetrf". I have checked the velocity and pressure field, it is correct, so if there is some inconsistency in FieldConvert for 1D homogeneous problem?
_______________________________________________ Nektar-users mailing list Nektar-users@imperial.ac.uk https://mailman.ic.ac.uk/mailman/listinfo/nektar-users [1] [1]
Spencer Sherwin McLaren Racing/Royal Academy of Engineering Research Chair, Professor of Computational Fluid Mechanics, Department of Aeronautics, Imperial College London South Kensington Campus London SW7 2AZ
s.sherwin@imperial.ac.uk +44 (0) 20 759 45052
Spencer Sherwin McLaren Racing/Royal Academy of Engineering Research Chair, Professor of Computational Fluid Mechanics, Department of Aeronautics, Imperial College London South Kensington Campus London SW7 2AZ
s.sherwin@imperial.ac.uk +44 (0) 20 759 45052
Links: ------ [1] https://mailman.ic.ac.uk/mailman/listinfo/nektar-users [1]
_______________________________________________ Nektar-users mailing list Nektar-users@imperial.ac.uk https://mailman.ic.ac.uk/mailman/listinfo/nektar-users [1]
-- David Moxey (Research Associate) d.moxey@imperial.ac.uk | www.imperial.ac.uk/people/d.moxey [2]
Room 363, Department of Aeronautics, Imperial College London, London, SW7 2AZ, UK.
Links: ------ [1] https://mailman.ic.ac.uk/mailman/listinfo/nektar-users [2] http://www.imperial.ac.uk/people/d.moxey
-- David Moxey (Research Associate) d.moxey@imperial.ac.uk | www.imperial.ac.uk/people/d.moxey Room 363, Department of Aeronautics, Imperial College London, London, SW7 2AZ, UK.
Hi Zhicheng, I think I have a fix in place in the latest commit of feature/FieldConvertCleanUp. You need to have USEFFT enabled in your SOLVERINFO (this is the default option). Essentially, the issue is that, in the absence of FFTW, we will attempt to set up a block matrix vector multiplication to evaluate the linear DFT. Unfortunately, for large numbers of nodes, we have issues with the evaluation of the interpolation weights at the set of equally spaced points, leading to rather nasty problems in the mass/Laplacian matrices. So for a quick fix I have just enabled the FFT module in FieldConvert. This has one, possibly troublesome, side effect, in that you won't have the plane at z = L_Z in your output vtu, due to the periodicity of the Fourier points. I will try to enable this in the homogeneous 1D output at some point soon. Cheers, Dave On 27 Apr 2015, at 16:23, David Moxey <d.moxey@imperial.ac.uk> wrote:
Hi Zhicheng,
Okay, so now I can confirm there's a problem -- changing to 128 causes me to get all sorts of problems.
It looks like there's a memory leak somewhere. Compiling in debug mode I've got malloc and free errors. Oddly, the code runs fine if I select the default BLAS/LAPACK reference implementation vs. OpenBLAS, which has all sorts of problems. However I get a field full of NaNs, so something is clearly wrong.
I'll try to track it down and get back to you.
Cheers,
Dave
On 2015-04-27 14:18, Zhicheng Wang wrote:
Hi Dave, Thanks for your help. I should have state the problem more clearly. actually I did a minor change in the TurbChFl_3DH1D.xml. I increased the HomoModesZ parameter to 128, then ./FieldConvert -m QCriterion TurbChFl_3DH1D.xml TurbChFl_3DH1D_1.chk TurbChFl_3DH1D_Q.vtu reports errors. But after using command ./FieldConvert TurbChFl_3DH1D.xml TurbChFl_3DH1D_1.chk TurbChFl_3DH1D_1.vtu, and check the *.vtu file, I find both of the velocity and pressure looks OK. So it seems the errors come from the conditions that HomoModes>=128. zhicheng On Mon, Apr 27, 2015 at 7:06 PM David Moxey <d.moxey@imperial.ac.uk> wrote:
Hi Zhicheng, I've taken a look this morning but was unable to reproduce the error on my machine. From the feature/FieldConvertCleanUp branch, I did the following: 1. Compiled with FFTW/MPI enabled. 2. Run solver: mpirun -n 6 IncNavierStokesSolver-g TurbChFl_3DH1D.xml 3. Run FieldConvert: FieldConvert-g -m QCriterion TurbChFl_3DH1D.xml TurbChFl_3DH1D_1.chk out.vtu This produces a field that I can visualise in paraview. What's the process that you're using which gives your the error message? Cheers, Dave On 2015-04-27 03:39, Zhicheng Wang wrote:
Hi Spencer, I have checked the latest branch feature/FieldConvertCleanUP, but unfortunately it still doesn't work. On Sun, Apr 26, 2015 at 1:03 AM Sherwin, Spencer J <s.sherwin@imperial.ac.uk> wrote:
Dear Zhicheng, When you download using the code from the gitlab repository you can then select a branch by typing git checkout feature/FieldConvertCleanUP if you type ‘git branch’ it tell you which branch you are on. Have you been using git or the tar ball distribution? Cheers Spencer. On 25 Apr 2015, at 17:48, Zhicheng Wang <wangzhicheng09@gmail.com> wrote: Hi Spencer, Thank you very much for your reply. but could you please tell me the link to branch feature/FieldConvertCleanUp ? I have browse the whole site, but have no idea on reaching this branch. zhicheng On Sat, Apr 25, 2015 at 11:19 PM Sherwin, Spencer J <s.sherwin@imperial.ac.uk> wrote: Hi Zhicheng, We have recently been trying to homogenise and update the field convert utility. Could I ask if you have tried it on the branch feature/FieldConvertCleanUp. Hopefully it will be working in this branch. We hope to get this branch into Master soon so it would also be good for us to know if you are experiencing problems here so we get things working as smoothly as possible. Cheers, Spencer. On 25 Apr 2015, at 14:23, Zhicheng Wang <wangzhicheng09@gmail.com> wrote: Hi nektar++ users and developers, I 'm using the latest nektar++ version. However when running the TurbChFl_3DH1D.xml example and using FieldConvert calculating vorticity or Qcriterion, I got errors as "ERROR: Element u_33 is 0 from dgetrf". I have checked the velocity and pressure field, it is correct, so if there is some inconsistency in FieldConvert for 1D homogeneous problem? _______________________________________________ Nektar-users mailing list Nektar-users@imperial.ac.uk https://mailman.ic.ac.uk/mailman/listinfo/nektar-users [1] [1] Spencer Sherwin McLaren Racing/Royal Academy of Engineering Research Chair, Professor of Computational Fluid Mechanics, Department of Aeronautics, Imperial College London South Kensington Campus London SW7 2AZ s.sherwin@imperial.ac.uk +44 (0) 20 759 45052 Spencer Sherwin McLaren Racing/Royal Academy of Engineering Research Chair, Professor of Computational Fluid Mechanics, Department of Aeronautics, Imperial College London South Kensington Campus London SW7 2AZ s.sherwin@imperial.ac.uk +44 (0) 20 759 45052 Links:
[1] https://mailman.ic.ac.uk/mailman/listinfo/nektar-users [1] _______________________________________________ Nektar-users mailing list Nektar-users@imperial.ac.uk https://mailman.ic.ac.uk/mailman/listinfo/nektar-users [1] -- David Moxey (Research Associate) d.moxey@imperial.ac.uk | www.imperial.ac.uk/people/d.moxey [2] Room 363, Department of Aeronautics, Imperial College London, London, SW7 2AZ, UK. Links:
[1] https://mailman.ic.ac.uk/mailman/listinfo/nektar-users [2] http://www.imperial.ac.uk/people/d.moxey
-- David Moxey (Research Associate) d.moxey@imperial.ac.uk | www.imperial.ac.uk/people/d.moxey
Room 363, Department of Aeronautics, Imperial College London, London, SW7 2AZ, UK.
_______________________________________________ Nektar-users mailing list Nektar-users@imperial.ac.uk https://mailman.ic.ac.uk/mailman/listinfo/nektar-users
-- David Moxey (Research Associate) d.moxey@imperial.ac.uk | www.imperial.ac.uk/people/d.moxey Room 363, Department of Aeronautics, Imperial College London, London, SW7 2AZ, UK.
Hi Dave, Great jobs! I have tried latest feature/FieldConvertCleanUp It works now. Thank you very much. On Wed, Apr 29, 2015 at 7:49 PM David Moxey <d.moxey@imperial.ac.uk> wrote:
Hi Zhicheng,
I think I have a fix in place in the latest commit of feature/FieldConvertCleanUp. You need to have USEFFT enabled in your SOLVERINFO (this is the default option).
Essentially, the issue is that, in the absence of FFTW, we will attempt to set up a block matrix vector multiplication to evaluate the linear DFT. Unfortunately, for large numbers of nodes, we have issues with the evaluation of the interpolation weights at the set of equally spaced points, leading to rather nasty problems in the mass/Laplacian matrices.
So for a quick fix I have just enabled the FFT module in FieldConvert. This has one, possibly troublesome, side effect, in that you won't have the plane at z = L_Z in your output vtu, due to the periodicity of the Fourier points. I will try to enable this in the homogeneous 1D output at some point soon.
Cheers,
Dave
On 27 Apr 2015, at 16:23, David Moxey <d.moxey@imperial.ac.uk> wrote:
Hi Zhicheng,
Okay, so now I can confirm there's a problem -- changing to 128 causes me to get all sorts of problems.
It looks like there's a memory leak somewhere. Compiling in debug mode I've got malloc and free errors. Oddly, the code runs fine if I select the default BLAS/LAPACK reference implementation vs. OpenBLAS, which has all sorts of problems. However I get a field full of NaNs, so something is clearly wrong.
I'll try to track it down and get back to you.
Cheers,
Dave
On 2015-04-27 14:18, Zhicheng Wang wrote:
Hi Dave, Thanks for your help. I should have state the problem more clearly. actually I did a minor change in the TurbChFl_3DH1D.xml. I increased the HomoModesZ parameter to 128, then ./FieldConvert -m QCriterion TurbChFl_3DH1D.xml TurbChFl_3DH1D_1.chk TurbChFl_3DH1D_Q.vtu reports errors. But after using command ./FieldConvert TurbChFl_3DH1D.xml TurbChFl_3DH1D_1.chk TurbChFl_3DH1D_1.vtu, and check the *.vtu file, I find both of the velocity and pressure looks OK. So it seems the errors come from the conditions that HomoModes>=128. zhicheng On Mon, Apr 27, 2015 at 7:06 PM David Moxey <d.moxey@imperial.ac.uk> wrote:
Hi Zhicheng, I've taken a look this morning but was unable to reproduce the error on my machine. From the feature/FieldConvertCleanUp branch, I did the following: 1. Compiled with FFTW/MPI enabled. 2. Run solver: mpirun -n 6 IncNavierStokesSolver-g TurbChFl_3DH1D.xml 3. Run FieldConvert: FieldConvert-g -m QCriterion TurbChFl_3DH1D.xml TurbChFl_3DH1D_1.chk out.vtu This produces a field that I can visualise in paraview. What's the process that you're using which gives your the error message? Cheers, Dave On 2015-04-27 03:39, Zhicheng Wang wrote:
Hi Spencer, I have checked the latest branch feature/FieldConvertCleanUP, but unfortunately it still doesn't work. On Sun, Apr 26, 2015 at 1:03 AM Sherwin, Spencer J <s.sherwin@imperial.ac.uk> wrote:
Dear Zhicheng, When you download using the code from the gitlab repository you can then select a branch by typing git checkout feature/FieldConvertCleanUP if you type ‘git branch’ it tell you which branch you are on. Have you been using git or the tar ball distribution? Cheers Spencer. On 25 Apr 2015, at 17:48, Zhicheng Wang <wangzhicheng09@gmail.com> wrote: Hi Spencer, Thank you very much for your reply. but could you please tell me the link to branch feature/FieldConvertCleanUp ? I have browse the whole site, but have no idea on reaching this branch. zhicheng On Sat, Apr 25, 2015 at 11:19 PM Sherwin, Spencer J <s.sherwin@imperial.ac.uk> wrote: Hi Zhicheng, We have recently been trying to homogenise and update the field convert utility. Could I ask if you have tried it on the branch feature/FieldConvertCleanUp. Hopefully it will be working in this branch. We hope to get this branch into Master soon so it would also be good for us to know if you are experiencing problems here so we get things working as smoothly as possible. Cheers, Spencer. On 25 Apr 2015, at 14:23, Zhicheng Wang <wangzhicheng09@gmail.com> wrote: Hi nektar++ users and developers, I 'm using the latest nektar++ version. However when running the TurbChFl_3DH1D.xml example and using FieldConvert calculating vorticity or Qcriterion, I got errors as "ERROR: Element u_33 is 0 from dgetrf". I have checked the velocity and pressure field, it is correct, so if there is some inconsistency in FieldConvert for 1D homogeneous problem? _______________________________________________ Nektar-users mailing list Nektar-users@imperial.ac.uk https://mailman.ic.ac.uk/mailman/listinfo/nektar-users [1] [1] Spencer Sherwin McLaren Racing/Royal Academy of Engineering Research Chair, Professor of Computational Fluid Mechanics, Department of Aeronautics, Imperial College London South Kensington Campus London SW7 2AZ s.sherwin@imperial.ac.uk +44 (0) 20 759 45052 Spencer Sherwin McLaren Racing/Royal Academy of Engineering Research Chair, Professor of Computational Fluid Mechanics, Department of Aeronautics, Imperial College London South Kensington Campus London SW7 2AZ s.sherwin@imperial.ac.uk +44 (0) 20 759 45052 Links:
[1] https://mailman.ic.ac.uk/mailman/listinfo/nektar-users [1] _______________________________________________ Nektar-users mailing list Nektar-users@imperial.ac.uk https://mailman.ic.ac.uk/mailman/listinfo/nektar-users [1] -- David Moxey (Research Associate) d.moxey@imperial.ac.uk | www.imperial.ac.uk/people/d.moxey [2] Room 363, Department of Aeronautics, Imperial College London, London, SW7 2AZ, UK. Links:
[1] https://mailman.ic.ac.uk/mailman/listinfo/nektar-users [2] http://www.imperial.ac.uk/people/d.moxey
-- David Moxey (Research Associate) d.moxey@imperial.ac.uk | www.imperial.ac.uk/people/d.moxey
Room 363, Department of Aeronautics, Imperial College London, London, SW7 2AZ, UK.
_______________________________________________ Nektar-users mailing list Nektar-users@imperial.ac.uk https://mailman.ic.ac.uk/mailman/listinfo/nektar-users
-- David Moxey (Research Associate) d.moxey@imperial.ac.uk | www.imperial.ac.uk/people/d.moxey
Room 363, Department of Aeronautics, Imperial College London, London, SW7 2AZ, UK.
participants (3)
-
David Moxey
-
Sherwin, Spencer J
-
Zhicheng Wang