QuantumATK Forum

QuantumATK => General Questions and Answers => Topic started by: fanjiaping on November 19, 2010, 07:01

Title: the problem about index out of bound
Post by: fanjiaping on November 19, 2010, 07:01
Dear professor
       My calculation appeared a problem. I don't know what happened ! So I copy the message to here.

+------------------------------------------------------------------------------+
|  12 E = -189.203 dE =  2.329186e-05 dH =  3.725734e-07                       |
+------------------------------------------------------------------------------+
| Calculation Converged in 12 steps                                            |
|                                                                              |
| Fermi Level  = -0.068148 Ha                                                  |
+------------------------------------------------------------------------------+
+------------------------------------------------------------------------------+
|                                                                              |
| Right Electrode Calculation  [Finished Thu Nov 18 10:06:56 2010]             |
|                                                                              |
+------------------------------------------------------------------------------+
+------------------------------------------------------------------------------+
|                                                                              |
| Device Density Matrix Calculation   [Started Thu Nov 18 10:06:56 2010]       |
|                                                                              |
+------------------------------------------------------------------------------+
| Left electrode chemical potential  = -0.068148 Ha                            |
| Right electrode chemical potential = -0.068148 Ha                            |
+------------------------------------------------------------------------------+
+------------------------------------------------------------------------------+
|                                                                              |
| Equivalent Bulk  [Started Thu Nov 18 10:07:39 2010]                          |
|                                                                              |
+------------------------------------------------------------------------------+
0 : ML_CommInfoOP_GenUsingGIDExternals warning : index out of bound              = 827(1) 324395 268450802(0)
rank 11 in job 1  master_36287   caused collective abort of all ranks
  exit status of rank 11: killed by signal 9



Above all is the imformation the output file given. Could you kindly tell me how to solve the problem?

my second question:
 The ATK software installed in my computer unnormally run. If I clicked the modules of  builder or custom, the main window will close! I uninstall the software and then install again,but the problem still exist! the bug appeared like:
  /opt/QuantumWise/atk-10.8.2/vnl/bin/vnl: line 3: 13080 Segmentation fault      VNL_PATH=$EXEC_DIR/../.. PSEUDOPOTENTIALS_PATH=$EXEC_DIR/../../atkpython/share/pseudopotentials PYTHONHOME=$EXEC_DIR/.. PYTHONPATH= LD_LIBRARY_PATH=$EXEC_DIR/../lib $EXEC_DIR/vnl_exec $*



Thank you for you help!
Title: Re: the problem about index out of bound
Post by: zh on November 19, 2010, 09:45
For the first problem, i.e., "0 : ML_CommInfoOP_GenUsingGIDExternals warning : index out of bound", it may be caused by something wrong in your input file. In particular, the index of atoms to align the electrodes and the center region are not specified properly. If possible, please show your input file to us.

The possible reason for the second problem (i.e., "Segmentation fault of VNL"): some of libraries required for VNL are not installed properly into your system.
Title: Re: the problem about index out of bound
Post by: zh on November 19, 2010, 11:46
Sorry, I cannot find out your input file in this forum.

The problem cannot be fixed by reinstalling the VNL. First, please the check the user-guide of the installment of VNL and find out the required libraries, and then confirm that all these libraries are installed.
Title: Re: the problem about index out of bound
Post by: fanjiaping on November 19, 2010, 11:52
  I am sorry ! I forgot to add the file after refresh the website.( :'( Always ,I need to refresh so many times to browse this forum   ;D)
Title: Re: the problem about index out of bound
Post by: fanjiaping on November 19, 2010, 11:56
     But the problem(second question) appeared recently. In the last month ,it can work well! If some libraries were absent, Why it can run in last month?
Title: Re: the problem about index out of bound
Post by: zh on November 19, 2010, 14:10
Do you change the setup of your computer system?
I have tried to run your input file with 8 CPUs. It also failed because the setup in your input file needs a large amount of memory.  Maybe, both of your problems are due to the insufficient memory for the job of running your input file.
Title: Re: the problem about index out of bound
Post by: fanjiaping on November 19, 2010, 15:56
   change the setup of my system means minimize the electrode or others?
   But I am sure my second problem wasn't caused by setup of my system! the main window of VNL closed was caused just by click on the module of builder or others !
Title: Re: the problem about index out of bound
Post by: Anders Blom on November 19, 2010, 23:00
Seems you've been quite unlucky! :)

I agree the first problem is most likely due to lack of memory. If you are running in parallel, make sure to only assign 1 MPI process per physical node, not to overload the memory.

The second problem is caused by your video driver. Somehow it does not fully support 3D OpenGL; most likely you are using a MESA driver. Try to upgrade to the latest vendor drivers, that should solve the issue.

Hope it works out easily!
Title: Re: the problem about index out of bound
Post by: mgsternberg on December 7, 2010, 23:14
Quote from: Anders Blom on November 19, 2010, 23:00
Seems you've been quite unlucky! :)

The [segfault] is caused by your video driver. Somehow it does not fully support 3D OpenGL; most likely you are using a MESA driver. Try to upgrade to the latest vendor drivers, that should solve the issue.

I'm afraid we've been hit by this issue as well, using ATK-10.8.2 on Redhat/CentOS-5.5, and displaying on Macs which run the latest Xquartz-2.5.3 (http://xquartz.macosforge.org/trac/wiki) X11 server on Mac OS X-10.6.5.  /opt/X11/bin/glxgears works, and we have:

Code
$ xdpyinfo | grep GL
    GLX
    SGI-GLX

vnl used to work  prior to Mac OS 10.6.5 (which was released  2010-11-18). /opt/X11/lib is all-new in this release, dated 2010-08-15;


Regards, Michael
Title: Re: the problem about index out of bound
Post by: mgsternberg on December 8, 2010, 01:08
Quote from: mgsternberg on December  7, 2010, 23:14
vnl used to work  prior to Mac OS 10.6.5 (which was released  2010-11-18). /opt/X11/lib is all-new in this release, dated 2010-08-15;

Correction: I could not, in fact, corroborate my statement.  I see the same segfault on Mac OS X-10.5.x, and using Apple's X11.app.  Alas, checking previous versions I found a curious workaround:


Regards,
Michael
Title: Re: the problem about index out of bound
Post by: Anders Blom on December 8, 2010, 07:31
Curiouser and curiouser, as Alice said :)

It was neat you found the workaround! We have a machine that exhibits the same problem, I'll try your workaround there, and then we'll see how to fix it more properly, this trick might help us a lot, thanks!