some issues with postprocessing

[ferroteam]

I have an issue with cryspy, when stage 1 is completed. I got the following message:

cryspy
[2023-10-15 19:15:21,315][cryspy_restart][INFO]

Restart CrySPY 1.2.1

[2023-10-15 19:15:22,933][ctrl_job][INFO] # ---------- job status
[2023-10-15 19:15:22,933][ctrl_job][INFO] ID 0: Stage 1 Done!
Traceback (most recent call last):
File "/p/home/sergeyl/miniconda3/bin/cryspy", line 8, in
sys.exit(main())
File "/p/home/sergeyl/miniconda3/lib/python3.10/site-packages/cryspy/scripts/cryspy.py", line 114, in main
jobs.handle_job()
File "/p/home/sergeyl/miniconda3/lib/python3.10/site-packages/cryspy/job/ctrl_job.py", line 157, in handle_job
self.ctrl_done()
File "/p/home/sergeyl/miniconda3/lib/python3.10/site-packages/cryspy/job/ctrl_job.py", line 173, in ctrl_done
self.ctrl_next_stage()
File "/p/home/sergeyl/miniconda3/lib/python3.10/site-packages/cryspy/job/ctrl_job.py", line 201, in ctrl_next_stage
skip_flag, self.kpt_data = select_code.next_stage(
File "/p/home/sergeyl/miniconda3/lib/python3.10/site-packages/cryspy/interface/select_code.py", line 57, in next_stage
skip_flag, kpt_data = ctrl_job_vasp.next_stage_vasp(stage, work_path,
File "/p/home/sergeyl/miniconda3/lib/python3.10/site-packages/cryspy/interface/VASP/ctrl_job_vasp.py", line 42, in next_stage_vasp
structure = Structure.from_file(work_path+'POSCAR')
File "/p/home/sergeyl/miniconda3/lib/python3.10/site-packages/pymatgen/core/structure.py", line 2852, in from_file
struct = cls.from_str(contents, fmt="poscar", primitive=primitive, sort=sort, merge_tol=merge_tol, **kwargs)
File "/p/home/sergeyl/miniconda3/lib/python3.10/site-packages/pymatgen/core/structure.py", line 2770, in from_str
struct = Poscar.from_str(input_string, default_names=False, read_velocities=False, **kwargs).structure
File "/p/home/sergeyl/miniconda3/lib/python3.10/site-packages/pymatgen/io/vasp/inputs.py", line 411, in from_str
struct = Structure(
File "/p/home/sergeyl/miniconda3/lib/python3.10/site-packages/pymatgen/core/structure.py", line 3658, in init
super().init(
File "/p/home/sergeyl/miniconda3/lib/python3.10/site-packages/pymatgen/core/structure.py", line 921, in init
site = PeriodicSite(
File "/p/home/sergeyl/miniconda3/lib/python3.10/site-packages/pymatgen/core/sites.py", line 334, in init
species = Composition({get_el_sp(species): 1}) # type: ignore
File "/p/home/sergeyl/miniconda3/lib/python3.10/site-packages/pymatgen/core/periodic_table.py", line 1409, in get_el_sp
return Species.from_str(obj) # type: ignore
File "/p/home/sergeyl/miniconda3/lib/python3.10/site-packages/pymatgen/core/periodic_table.py", line 1052, in from_str
properties = {tokens[0]: ast.literal_eval(tokens[1])}
IndexError: list index out of range

I cannot figure it out what is wrong. Jobs finished correctly.

[Tomoki-YAMASHITA]

It seems pymatgen is causing this problem.
Can you tell me the version of pymatgen and what element you used?
And paste cryspy.in too.

[ferroteam]

Since I did "pip install csp-cryspy", it did install pymatgen (whatever version was required).
cryspy.in file is here:

[basic]
algo = RS
calc_code = VASP
tot_struc = 10
nstage = 5
njob = 10
jobcmd = sbatch
jobfile = wrapper.slm

[structure]
natot = 20
atype = Cs Ge F
nat = 4 4 12

[VASP]
kppvol = 20 40 40 40 60

[option]

I looked and see that pymatgen's version installed is pymatgen-2023.10.11

[ferroteam]

those files are correct, however the format of stage1_CONTCAR is different (new form of writing of atomic species from VASP as of version 6.4.X):

ID_0
1.00000000000000
5.7669656560892433 -3.3295591739504493 0.0000000000000004
0.0000000000000000 6.6591183479009013 0.0000000000000004
0.0000000000000000 0.0000000000000000 16.9213109569694637
Cs_sv/c1463496 Ge_d/944b26c40 F/53c630871ac6
4 4 12
Direct
-0.0000000000000000 -0.0000000000000000 0.2500000000000000
-0.0000000000000000 -0.0000000000000000 0.7500000000000000
0.6666666666666643 0.3333333333333357 0.2500000000000000
0.3333333333333357 0.6666666666666643 0.7500000000000000
-0.0000000000000000 -0.0000000000000000 -0.0000000000000000
-0.0000000000000000 -0.0000000000000000 0.5000000000000000
0.3333333333333357 0.6666666666666643 0.2500000000000000
0.6666666666666643 0.3333333333333357 0.7500000000000000
0.7469807024486543 0.8685159571105212 0.4331276778580619
0.1314840428894784 0.8784647453381325 0.4331276778580619
0.1215352546618673 0.2530192975513456 0.4331276778580619
0.7469807024486543 0.8685159571105212 0.0668723221419380
0.1314840428894784 0.8784647453381323 0.0668723221419381
0.1215352546618673 0.2530192975513456 0.0668723221419381
0.8685159571105212 0.7469807024486543 0.5668723221419381
0.8784647453381325 0.1314840428894785 0.5668723221419379
0.2530192975513456 0.1215352546618673 0.5668723221419379
0.8685159571105212 0.7469807024486543 0.9331276778580621
0.8784647453381325 0.1314840428894784 0.9331276778580619
0.2530192975513456 0.1215352546618673 0.9331276778580619

Is this causing the trouble?

[ferroteam]

I just found this thread on VASP forum:

https://vasp.at/forum/viewtopic.php?f=3&t=19113

It seems their attempt to change how CONTCAR is written, is causing trouble with pymatgen (and other software). I have to recompile VASP without HDF5 support to avoid this issue.

[Tomoki-YAMASHITA]

Thank you for the helpful information.
The only way to avoid problems now is to avoid using hashes in CONTCAR.

[Tomoki-YAMASHITA]

Another first aid measure is to rewrite line 6 of CONTCAR with job script (wrapper.slm).

#!/bin/sh
#$ -cwd
#$ -V -S /bin/bash
####$ -V -S /bin/zsh
#$ -N N_CrySPY_ID
#$ -pe smp 64
####$ -q ibis1.q
#####$ -q ibis2.q

mpirun -np $NSLOTS vasp_std

# for hash
sed -i -e '6 s/^.*$/Cs Ge F/' CONTCAR

sed -i -e '3 s/^.*$/done/' stat_job

[USFmatscilab]

Thank you for suggestions. It seems VASP's team offers a patch how to avoid this printing of hashtags in CONTCAR. It seems there are many people are affected by this new deveopment.