The ABINIT Users Mailing List ( CLOSED )

["matthieu verstraete" <matthieu.jean.verstraete@gmail.com>]

Hi Vincent, I'm responding to forum, as I suspect others will benefit (and you have already solved the problem).


in the following, presume you have a files file as follows:
i.abinit               ascii input
o.abinit              ascii output
inp_abinit           binary input
out_abinit           binary output
tmp_abinit
psp
psp

On Sun, Nov 2, 2008 at 12:31 AM, Vincent Chevrier <vincent.chevrier@dahn.phys.dal.ca> wrote:

Hi,

I'm posting this to the developer mailing list because it's not a problem
but a comment regarding the use of irdwfk and getwfk.

irdwfk 1  implies reading in from inp_abinit_WFK, or if you  are in dataset "m" from inp_abinit_DSM_WFK

getwfk n implies reading _in_ from the _output_ of DS "n", ie from file out_abinit_DSN_WFK

Let's say we wanted to do the first example of tutorial RF2 (trf2_1.in) in
two separate calculations. Say, example_1.in containing ndtset 1 and
example_2.in containing ndtset 2-10. In order to use parallelisation over
k-points in example_1.in to generate the WFK file.

the file names will not change or care if you use several input files or runs, eg identical inputs apart from

file1: ndtset 1 jdtset 1
file2: ndtset 9 jdtset 2 3 4 5 6 7 8 9 10

or even
file1: ndtset 1 jdtset 1
file2: ndtset 1 jdtset 2
file3: ndtset 1 jdtset 3
...
providing you are always in the same directory, or have copied the needed files to the other directories.

The nomenclature required to reuse the "example_1o_WFK" for all the
datasets in example_2.in was unclear to me.

in this case, as it is not part of a series of datsets, you can only reuse it by moving it to "example_2i_WFK" and using irdwfk 1 in example 2.in
You can also pretend it was DS 1 of your run, and continue from dataset 2 onwards, setting getwfk 1, and moving the file to example_2o_DS1_WFK


 

The first thing I tried was to use "irdwfk 1" in example_2.in and
"example_1o" on the third line of example_2.files. The run died with the
following error:
"""
-P-0000         current calculation                         restart file
-P-0000         -------------------                         ------------
-P-0000
-P-0000   calculation expects a wf_planewave    |  input file contains a
wf_planewave
-P-0000 . ABINIT  code version 5.4.4            |  ABINIT  code version
^@^@^@^@^@^@
-P-0000
-P-0000  hdr_check: BUG -
-P-0000   input fform=         2 differs from disk file fform=         0.
-P-0000
-P-0000  leave_new : decision taken to exit ...

"""
I noticed that a new file of zero size had been created with the name
"example_1o_DS1_WFK". I found another post on the forum asking about this
"hdr_check: BUG - input fflorm=" and the only response was the possibility
of the file being corrupt (which isn't the case here, abinit is looking
for a file that doesn't exist).

If it is looking for something o_DS1 that means you have set getwfk... you should send in the input file (and the files file in this case) to be clear about what is happening. Normally you can not set getwf and irdwfk in a single dataset

-Shouldn't the user be told that the file wasn't found instead of this
error. Or maybe just state the filename of the restart file abinit will
attempt to use.

absolutely, abinit should be complaining it can't find the file instead of dying. If you look in the log it does tell you which file it will try to read from.

 

-Why was a "example_1o_DS1_WFK" file created?

an erroneous file open, when a file is expected to be there but isn't, sometimes creates an empty file in place of the one looked for.
 

-In reading the help I thought "_WFK" would be appended to "example_1o",
not "_DS1_WFK", "_DS2_WFK". Is that ever clearly stated in the help?
(Neither
http://www.abinit.org/Infos_v5.5/input_variables/varfil.html#irdwfk or
http://www.abinit.org/Infos_v5.5/users/abinis_help.html#4
mention it)

if you are in a dataset, _all_ output and input files have _DSN_ inserted between the radix and the suffix.
 

I realized I could address this in two different ways:
copying "example_1o_WFK" to "example_1o_DS1_WFK", "example_1o_DS2_WFK",
etc... and using the same parameters as stated previously.

if you have irdwfk 1

or

I could avoid making multiple copies by naming my old WFK
"example_2o_DS1_WFK" and stating "ndtset 10  jdtset 2 3 4 5 6 7 8 9 10 11"
in example_2.in.

exactly, but again this means you have set getwfk 1 in your input. This is definitely the preferred method (no file copying needed, and everything can be chained in the dtset calculations).
 

It's not a big issue but it was confusing to me. An extra sentence in the
help or the log file could easily clarify this.

excellent idea, this email is a start, but integrating it in the doc is even better. For the moment the corresponding comments are more likely in the section on datasets, but should be mentioned also in the particular case of getwfk irdwfk (no other input file has this dichotomy).

Ciao


Matthieu

-- 
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
Dr. Matthieu Verstraete

European Theoretical Spectroscopy Facility (ETSF) 
Dpto. Fisica de Materiales, 
U. del Pais Vasco,            
Centro Joxe Mari Korta, Av. de Tolosa, 72,   Phone: +34-943018393
E-20018 Donostia-San Sebastian, Spain        Fax  : +34-943018390

Mail : matthieu.jean.verstraete@gmail.com
http://www-users.york.ac.uk/~mjv500