import z2pack error

Dominik Gresch greschd at phys.ethz.ch
Mon Feb 20 18:00:15 CET 2017


Hi Wei Li,

There is clearly something wrong with these results -- there are far too 
many Wannier charge centers. This has nothing to do with the input 
parameters to Z2Pack. Instead, the input files for VASP or Wannier90 are 
generating too many bands. The reason for this might be that VASP 
automatically adds more bands when the number of bands doesn't fit the 
number of cores on which the calculation is running. Since Wannier90 has 
no way of automatically ignoring these, they are included in the 
calculation.

There are two ways of solving this:

     1. Change the number of cores on which you run, such that only the 
15 NBANDS are actually calculated

     2. Check the number of bands that is actually used (you should find 
this in the VASP output file), and change the upper limit of 
"exclude_bands" in wannier90.win

Best regards,
Dominik

On 20.02.2017 17:50, Wei Li wrote:
> Hi Dominik,
>
> Thanks.
>
> I generated the plot for Bi calculation as attached.  But there are 
> very high dense KPOINTS. I thought I change optimize it by set the 
> parameter iterator=range as following:
>
>
> result_0 = z2pack.surface.run(
>   system=system,
>   surface=lambdas, t: [0, s / 2, t],
>   iterator=range(10,24,2),
>     save_file='./results/res_0.p',
>   load=True
>
> But it generate same result.
>
> Or do I need to set s limit by pos_tol?
>
> Regards
> Wei Li
>
>
>
>
>> On Feb 16, 2017, at 5:37 PM, Dominik Gresch <greschd at phys.ethz.ch 
>> <mailto:greschd at phys.ethz.ch>> wrote:
>>
>> Dear Wei Li,
>>
>> The surface is parametrized as a function of two parameters, as 
>> described here: 
>> http://z2pack.ethz.ch/doc/2.0/tutorial_surface.html#basic-syntax
>>
>> For VASP, it is important to note that the surface must be straight 
>> in the second parameter (that is, f(t1=const., t2) describes a 
>> straight line across the full BZ, parallel to one of the reciprocal 
>> axes). This is again due to a limitation in the interface between 
>> Wannier90 and VASP, where only automatic mesh generation is possible.
>>
>> For your second question about the progress of the calculation: To 
>> see the progress of your calculation, it is much more helpful to look 
>> at the output of Z2Pack itself instead of the build folder or the 
>> result file. Z2Pack reports which lines it is calculating, and what 
>> the status of the convergence is.
>>
>> You can improve the performance for example if you find that the 
>> iteration in a single line never converges before a specific value, 
>> by adjusting the 'iterator' parameter accordingly. The main reason it 
>> can sometimes take a long time is because of the adaptive convergence 
>> criteria. This is a tradeoff we made on purpose, because it 
>> significantly improves the reliability of the results. Of course you 
>> can turn them off, but I would strongly advise against this. All in 
>> all, performance was not so much the focus in developing Z2Pack as 
>> reliability and flexibility in using it for different purposes.
>>
>> Best regards,
>>
>> Dominik
>>
>>
>> On 16.02.2017 22:15, Wei Li wrote:
>>> Dear Dominik,
>>>
>>> Thanks, I have complied wannier90 2.1 on top of VASP and I am doing 
>>> the test currently.
>>>
>>> I noticed that z2pack dynamically generate the KPOINTS files for WCC 
>>> calculations. Everything runs well except that I am not sure about 
>>> parameterizing the
>>> surface, which means that when the KPOINTS loop can end. So far my 
>>> KPOINTS files reach to like below:
>>>
>>> Automatic mesh
>>> 0              ! number of k-points = 0 ->automatic generation scheme
>>> Gamma          ! generate a Gamma centered grid
>>> 1 1 17        ! subdivisions
>>> 0.0 0.03125 0.0! shift
>>>
>>>
>>> Do I need to set a parameter to end the KPOINTS shifting, or I just 
>>> need to waiting for the calculation completed as the parameters in 
>>> the tutorial? So far, it realy
>>> consumes lots of computation time. Here is my results file information:
>>>
>>> [wli9 at br006 Z2]$ ls -artl results/
>>> total 1026
>>> -rw------- 1 wli9 mr4s90p 75853 Feb 16 16:13 res_0.p
>>> drwxr-xr-x 2 wli9 mr4s90p 33280 Feb 16 16:13 .
>>> drwxr-xr-x 6 wli9 mr4s90p 33280 Feb 16 16:13 ..
>>>
>>> Regards
>>> Wei Li
>>>
>>>
>>>> On Feb 13, 2017, at 5:44 PM, Dominik Gresch <greschd at phys.ethz.ch 
>>>> <mailto:greschd at phys.ethz.ch>> wrote:
>>>>
>>>> Dear Wei Li,
>>>>
>>>> The Bi example should run with the supplied input files. However, 
>>>> you need to install a modified version of Wannier90, as explained 
>>>> in the installation instructions: 
>>>> http://z2pack.ethz.ch/doc/2.0/tutorial_installation.html#setting-up-first-principles-codes
>>>>
>>>> For Wannier90 2.1 or newer this is no longer necessary because we 
>>>> added a specific interface to run with Z2Pack, but unfortunately 
>>>> VASP only supports Wannier90 version 1.x.
>>>>
>>>> Best regards,
>>>>
>>>> Dominik Gresch
>>>>
>>>>
>>>> On 13.02.2017 23:40, Wei Li wrote:
>>>>> Hi Dominik,
>>>>>
>>>>> Thanks, it runs well now.
>>>>>
>>>>> For the given example (Bi) calculation, I need to set up the 
>>>>> search_shells number. What’s the value you recommend?
>>>>> I have increased to 30 from default value 12, it still can not 
>>>>> satisfy the B1 condition:
>>>>>
>>>>>  Unable to satisfy B1 with any of the first 30 shells
>>>>>  Your cell might be very long, or you may have an irregular MP grid
>>>>>  Try increasing the parameter search_shells in the win file 
>>>>> (default=12)
>>>>>
>>>>> Or do I need to check other parameters?
>>>>>
>>>>> Here is my input:
>>>>>
>>>>> num_wann = 10
>>>>> num_bands = 10
>>>>> spinors : true
>>>>> search_shells=30
>>>>> num_iter 0
>>>>> exclude_bands 11-15
>>>>>
>>>>> Regards
>>>>> Wei Li
>>>>>
>>>>>
>>>>>> On Feb 11, 2017, at 4:00 PM, Dominik Gresch <greschd at phys.ethz.ch 
>>>>>> <mailto:greschd at phys.ethz.ch>> wrote:
>>>>>>
>>>>>> Dear Wei Li,
>>>>>>
>>>>>> If everything work correctly, Z2Pack should automatically call 
>>>>>> VASP (via the command given) and read the .mmn file after VASP 
>>>>>> has generated it. This is because Z2Pack needs to iteratively 
>>>>>> change the k-point path with which VASP is called and the overlap 
>>>>>> file is calculated.
>>>>>>
>>>>>> In your case, it seems that you are submitting the VASP run to 
>>>>>> some scheduler -- this means the command returns immediately, 
>>>>>> before the .mmn file has been generated. As a result, Z2Pack 
>>>>>> tries to read the file, but it does not exist yet. You can solve 
>>>>>> this problem by submitting instead the Python process running 
>>>>>> Z2Pack, and then running VASP within that same job.
>>>>>>
>>>>>> Best regards,
>>>>>>
>>>>>> Dominik
>>>>>>
>>>>>>
>>>>>> On 11.02.2017 20:40, Wei Li wrote:
>>>>>>> Dear Dominik,
>>>>>>>
>>>>>>> From what I understand, after calling the first principle code 
>>>>>>> (in my case, it is VASP), the next step is to
>>>>>>> read the .mmn file in the top level of the build file. So I give 
>>>>>>> the mmn path like below:
>>>>>>>
>>>>>>> system = z2pack.fp.System(
>>>>>>>     input_files=["input/CHGCAR", "input/INCAR", "input/POSCAR", 
>>>>>>> "input/POTCAR", "input/wannier90.win"],
>>>>>>> kpt_fct=z2pack.fp.kpoint.vasp,
>>>>>>>     kpt_path="KPOINTS",
>>>>>>>     command="qsub /home/work/mtg/users/verali/Bi_Z2/Z2/paral.qs"
>>>>>>> mmn_path="wannier90.mmn"
>>>>>>> )
>>>>>>>
>>>>>>> But it does not work.
>>>>>>>
>>>>>>> Could you advise how to read the mmn file?
>>>>>>>
>>>>>>> Regards
>>>>>>> Wei Li
>>>>>>>
>>>>>>>
>>>>>>>> On Feb 10, 2017, at 2:59 PM, Wei Li <verali at udel.edu 
>>>>>>>> <mailto:verali at udel.edu>> wrote:
>>>>>>>>
>>>>>>>> Dear Dominik,
>>>>>>>>
>>>>>>>> Thanks for your reply. I solved this problem.
>>>>>>>>
>>>>>>>> I generated the build file as following:
>>>>>>>>
>>>>>>>> CHG CONTCAR  EIGENVAL INCAR    OSZICAR paral.qs.o889660 POSCAR  
>>>>>>>> REPORT wannier90.eig wannier90.win WAVECAR
>>>>>>>> CHGCAR DOSCAR   IBZKPT KPOINTS  OUTCAR PCDAT POTCAR  
>>>>>>>> vasprun.xml wannier90.mmn wannier90.wout XDATCAR
>>>>>>>>
>>>>>>>> There is no error.
>>>>>>>>
>>>>>>>> which script do I need to read the .mmn file to plot the 
>>>>>>>> result? So far, the result file is empty.
>>>>>>>>
>>>>>>>> Regards
>>>>>>>> Vera
>>>>>>>>
>>>>>>>>
>>>>>>>>> On Feb 8, 2017, at 11:16 AM, Dominik Gresch 
>>>>>>>>> <greschd at phys.ethz.ch <mailto:greschd at phys.ethz.ch>> wrote:
>>>>>>>>>
>>>>>>>>> Dear Wei Li,
>>>>>>>>>
>>>>>>>>> It seems you are running multiple instances of the Python 
>>>>>>>>> interpreter, due to the 'mpirun' command. There seems to be a 
>>>>>>>>> bit of a race condition when they all try to create the 
>>>>>>>>> 'results' folder.
>>>>>>>>>
>>>>>>>>> In general, you shouldn't run Python itself with mpirun, only 
>>>>>>>>> the first-principles command (command="mpirun 
>>>>>>>>> YOUR_VASP_COMMAND" in the z2pack.fp.System options).
>>>>>>>>>
>>>>>>>>> Best regards,
>>>>>>>>>
>>>>>>>>> Dominik Gresch
>>>>>>>>>
>>>>>>>>>
>>>>>>>>> On 08.02.2017 17:08, Wei Li wrote:
>>>>>>>>>> Dear Dominik Gresch,
>>>>>>>>>>
>>>>>>>>>> Thanks a lot.
>>>>>>>>>>
>>>>>>>>>> I activated py35 and it works.
>>>>>>>>>>
>>>>>>>>>> Do I need to change any parameters if I want to run the 
>>>>>>>>>> example given here:
>>>>>>>>>> https://github.com/Z2PackDev/Z2Pack/tree/master/2.0.x/examples/fp/Bi_vasp
>>>>>>>>>>
>>>>>>>>>> I run it directly and I got the attached output, I did not 
>>>>>>>>>> find any error in output file or log file, but the process is 
>>>>>>>>>> terminated.
>>>>>>>>>>
>>>>>>>>>>
>>>>>>>>>> Regards
>>>>>>>>>> Wei Li
>>>>>>>>>>
>>>>>>>>>>
>>>>>>>>>>> On Feb 8, 2017, at 4:53 AM, Dominik Gresch 
>>>>>>>>>>> <greschd at phys.ethz.ch <mailto:greschd at phys.ethz.ch>> wrote:
>>>>>>>>>>>
>>>>>>>>>>> Dear Wei Li,
>>>>>>>>>>>
>>>>>>>>>>> The python version on which you installed Z2Pack (3.5.2) is 
>>>>>>>>>>> probably not the same as you are using to run the script. 
>>>>>>>>>>> Since it seems you are using a virtualenv (py35) when 
>>>>>>>>>>> running Python from the command line, you should make sure 
>>>>>>>>>>> the same virtualenv is also active in your job (i.e., 
>>>>>>>>>>> "source py35/bin/activate" in your script).
>>>>>>>>>>>
>>>>>>>>>>> Best regards,
>>>>>>>>>>>
>>>>>>>>>>> Dominik Gresch
>>>>>>>>>>>
>>>>>>>>>>> On 08.02.2017 10:31, Wei Li wrote:
>>>>>>>>>>>> Dear Z2pack team,
>>>>>>>>>>>>
>>>>>>>>>>>> I have complied Wannier90 1.2 on top of VASP, and then I 
>>>>>>>>>>>> downloaded and installed z2pack like below:
>>>>>>>>>>>>
>>>>>>>>>>>> (py35) [verali at farber input]$ python
>>>>>>>>>>>> Python 3.5.2 |Anaconda 4.2.0 (64-bit)| (default, Jul 2 
>>>>>>>>>>>> 2016, 17:53:06)
>>>>>>>>>>>> [GCC 4.4.7 20120313 (Red Hat 4.4.7-1)] on linux
>>>>>>>>>>>> Type "help", "copyright", "credits" or "license" for more 
>>>>>>>>>>>> information.
>>>>>>>>>>>> >>> import z2pack
>>>>>>>>>>>> >>> z2pack.__version__
>>>>>>>>>>>> '2.0.3'
>>>>>>>>>>>> >>>
>>>>>>>>>>>>
>>>>>>>>>>>> I think this means I have successfully installed z2pack, right?
>>>>>>>>>>>>
>>>>>>>>>>>> But when I run run.py script, it had the error:
>>>>>>>>>>>>
>>>>>>>>>>>>   File "run.py", line 14, in <module>
>>>>>>>>>>>> Traceback (most recent call last):
>>>>>>>>>>>>   File "run.py", line 14, in <module>
>>>>>>>>>>>>     import z2pack
>>>>>>>>>>>> ImportError: No module named z2pack
>>>>>>>>>>>> Traceback (most recent call last):
>>>>>>>>>>>>   File "run.py", line 14, in <module>
>>>>>>>>>>>>
>>>>>>>>>>>>
>>>>>>>>>>>> *Below is my job file:*
>>>>>>>>>>>>
>>>>>>>>>>>>
>>>>>>>>>>>>
>>>>>>>>>>>> vpkg_require intel/2015.3.187
>>>>>>>>>>>> vpkg_require openmpi/1.8.5
>>>>>>>>>>>>
>>>>>>>>>>>> #$ -pe mpi 40
>>>>>>>>>>>> #$ -l m_mem_free=3G
>>>>>>>>>>>> #$ -l h_cpu=09:00:00
>>>>>>>>>>>> ##$ -l exclusive=1
>>>>>>>>>>>>
>>>>>>>>>>>> MY_EXE="/home/work/mtg/users/bin/vasp_ncl"
>>>>>>>>>>>>
>>>>>>>>>>>> MY_EXE_ARGS=()
>>>>>>>>>>>>
>>>>>>>>>>>> #
>>>>>>>>>>>> # Uncomment to enable lots of additional information as 
>>>>>>>>>>>> OpenMPI executes
>>>>>>>>>>>> # your job:
>>>>>>>>>>>> #
>>>>>>>>>>>> #SHOW_MPI_DEBUGGING=YES
>>>>>>>>>>>>
>>>>>>>>>>>> ##
>>>>>>>>>>>> ## You should NOT need to change anything after this comment.
>>>>>>>>>>>> ##
>>>>>>>>>>>> OPENMPI_FLAGS="--display-map --mca btl ^openib --mca 
>>>>>>>>>>>> oob_tcp_if_exclude lo,ib0 --mca btl_tcp_if_exclude lo,ib0"
>>>>>>>>>>>> if [ "${WANT_CPU_AFFINITY:-NO}"= "YES"]; then
>>>>>>>>>>>>   OPENMPI_FLAGS="${OPENMPI_FLAGS} --bind-to core"
>>>>>>>>>>>> fi
>>>>>>>>>>>> if [ "${WANT_NPROC:-0}"-gt 0 ]; then
>>>>>>>>>>>>   OPENMPI_FLAGS="${OPENMPI_FLAGS} --np ${WANT_NPROC} 
>>>>>>>>>>>> --map-by node"
>>>>>>>>>>>> fi
>>>>>>>>>>>> if [ "${SHOW_MPI_DEBUGGING:-NO}"= "YES"]; then
>>>>>>>>>>>>   OPENMPI_FLAGS="${OPENMPI_FLAGS} --debug-devel 
>>>>>>>>>>>> --debug-daemons --display-devel-map 
>>>>>>>>>>>> --display-devel-allocation --mca mca_verbose 1 --mca 
>>>>>>>>>>>> coll_base_verbose 1 --mca ras_base_verbose 1 --mca 
>>>>>>>>>>>> ras_gridengine_debug 1 --mca ras_gridengine_verbose 1 --mca 
>>>>>>>>>>>> btl_base_verbose 1 --mca mtl_base_verbose 1 --mca 
>>>>>>>>>>>> plm_base_verbose 1 --mca pls_rsh_debug 1"
>>>>>>>>>>>>   if [ "${WANT_CPU_AFFINITY:-NO}"= "YES"]; then
>>>>>>>>>>>>     OPENMPI_FLAGS="${OPENMPI_FLAGS} --report-bindings"
>>>>>>>>>>>>   fi
>>>>>>>>>>>> fi
>>>>>>>>>>>>
>>>>>>>>>>>> echo "GridEngine parameters:"
>>>>>>>>>>>> echo "  mpirun        = "`which mpirun`
>>>>>>>>>>>> echo "  nhosts        = $NHOSTS”
>>>>>>>>>>>> echo "  nproc       = $NSLOTS"
>>>>>>>>>>>> echo " executable = $MY_EXE"
>>>>>>>>>>>> echo "  MPI flags     = $OPENMPI_FLAGS"
>>>>>>>>>>>> echo "-- begin OPENMPI run -- on "
>>>>>>>>>>>> mpirun ${OPENMPI_FLAGS} python run.py
>>>>>>>>>>>> echo "-- end OPENMPI run -- on “
>>>>>>>>>>>>
>>>>>>>>>>>> Pls help to advise how can I solve this problem.
>>>>>>>>>>>>
>>>>>>>>>>>> Thanks in advance.
>>>>>>>>>>>>
>>>>>>>>>>>> Wei Li
>>>>>>>>>>>> Ph.D student
>>>>>>>>>>>> University of Delaware
>>>>>>>>>>>>
>>>>>>>>>>>>
>>>>>>>>>>>
>>>>>>>>>>
>>>>>>>>>
>>>>>>>>
>>>>>>>
>>>>>>
>>>>>
>>>>
>>>
>>
>

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://lists.phys.ethz.ch/private/z2pack/attachments/20170220/7b6c3895/attachment.html>


More information about the Z2Pack mailing list