ABINIT Discussion Forums

Dear all,
i'm finding an error when running “optics” calculations with Abinit.

I'm following the general scheme shown in Abinit Tutorials, i.e.
1) first a run which performs an SCF calculation and writes (among others) a DEN and a WFK file
2) then a run doing a non SCF calc. with more bands, reading DEN and WFK from previous run
3) then a run doing ddk for the 3 directions
4) and lastly, what in the tutorials is called toptic_4 if I remember well, which generates the files containing the spectrum

Of course this is just one of the possible ways to group all these runs into 1 or more multidataset runs, but i'm choosing this particular scheme in order to separate those parts which can be run in parallel from those which cannot (which is useful for big systems, although now i'm doing only a test on a small one)


The first run (SCF) works ok, but the second run stops here:


- hdr_check: checking restart file header for consistency -





current calculation restart file

------------------- ------------



calculation expects a wf_planewave | input file contains a wf_planewave

. ABINIT code version 6.0.4 | ABINIT code version °Ï^_<9f>ÿ^?



hdr_check: BUG -

input fform= 2 differs from disk file fform= 0.

Action : contact ABINIT group.



.Delivered 0 WARNINGs and 0 COMMENTs to log file.



leave_new : decision taken to exit ...




Regarding the suggestions given in previous posts/topics on this kind of error, I checked that my DEN and WFK files created by the first run are non-empty.
I also ran cut3d on one of these WFK files, and it runs without errors; but when I open the resulting file with XcrysDen, it seems to get stuck in trying to visualize the wavefunction (I used in cut3d the option “analize 1 wf”), which is quite strange, since i'm doing this test on something so simple as a CH4 molecule! (on the same machine xcrysden takes only few seconds to visualize 1 wf of a 175 atom molecule...).


I repeated the same test with different abinit versions (6.0.4, 6.6.2, 5.8.4) and on different machines (our local linux cluster and CINECA SP6), and I always get the same error.

For reading during the 2nd run (2) the DEN and WFK files created by the 1st run (1), I tried two ways:

a) irdden, irdwfk after renaming the DEN and WFK files of the 1st run as ...2ds2to3i... (ds2to3 is my root name for the second run, since it is a 2 dataset run, where I have called the 2 datasets “2” and “3”)
b) getden and getwfk after renaming the DEN and WFK files of the 1st run as ...o_DS1_... i.e. as if produced by a previous “dataset 1” in the same run

The only way to make the thing work seems to be that of doing what I called steps (1) (2) and (3) in a single multidataset run. In this case there are no problems in reading DEN and WFK files from previous datasets. I don't see where the difference can be with this approach: the DEN, WFK etc files are written to disk in both cases (multidataset or separate runs), with the same names (and the same content, hopefully).

This single file multidataset scheme doesn't solve the problem, though, since, also in this case, I then have to run the final “toptic_4” run,
the one which has an input like this:

toptic_3o_DS4_1WF526

toptic_3o_DS5_1WF527

toptic_3o_DS6_1WF528

toptic_3o_DS3_WFK

0.002 ! Value of the smearing factor, in Hartree

0.0003 0.5 ! Difference between frequency values (in Hartree), and maximum frequency (

1 Ha is about 27.211 eV)

0.000 ! Scissor shift if needed, in Hartree

0.002 ! Tolerance on closeness of singularities (in Hartree)

3 ! Number of components of linear optic tensor to be computed

11 22 33 ! Linear coefficients to be computed (x=1, y=2, z=3)

0 ! Number of components of nonlinear optic tensor to be computed

! Non-linear coefficients to be computed





and I really don't know how to include an input like this in the previous multidataset run (steps (1) (2) (3) in my initial scheme). And, if I run it separately, it cannot read the ...WF... files it should read. (so again the same problem in reading WF files written by previous runs)

What is really strange is that up to 2 or 3 months ago I ran lots of these optics runs, with the same separate-runs procedure, on the same machines, with (some of) the same versions of abinit.. and on a much bigger system, and I had no errors of this kind.

I also thought it might be a trivial problem of introducing a “tab” character or something similar when preparing the input files by “copy-paste- modify keywords, atom coords etc “ from old input files, so I ran again the test writing the input files from scratch, but it gave the same error.


Does anyone have suggestions? Can it be due to some libraries being updated in the meantime ( = since I ran the last correct run) ?
Many thanks in advance

Elena

Elena Molteni
Department of Physics
University of Milan
via Celoria, 16
I-20133, Milan, Italy
and European Theoretical Spectroscopy Facility (ETSF)
http://www.etsf.eu

...a correction to what i said in my previous post:

indeed, after running cut3d on my DEN or WFK files (without errors), xcrysden *does* read the resulting files and visualize the wavefunctions.

This might suggest these DEN and WFK files are not corrupted?

In fact, these same DEN and WFK files are read correctly in the "single run multidataset" case
(the only difference is that they are written, and later read, during a single run, instead of being written by a run and read by another one)

cheers
Elena

Elena Molteni
Department of Physics
University of Milan
via Celoria, 16
I-20133, Milan, Italy
and European Theoretical Spectroscopy Facility (ETSF)
http://www.etsf.eu