Error message for double definition of tolXXX  [SOLVED]

Total energy, geometry optimization, DFT+U, spin....

Moderator: bguster

Locked
mverstra
Posts: 655
Joined: Wed Aug 19, 2009 12:01 pm

Error message for double definition of tolXXX  [SOLVED]

Post by mverstra » Mon Apr 01, 2019 2:05 pm

This is copied from Wofferman - please make separate posts for each topic.

The short answer is, this is not a bug, simply the error message is not easy to see or find if you run in parallel. This is inevitable - the test is done at the right time and the error message is clear, but in parallel on a large cluster it gets lost in the mpi junk and your many possible output files. Also, due to buffering, it is not even guaranteed that it will be saved to disk.

Matthieu

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

Dear Eric and ABINIT friends,

Before I respond to your suggestion, I would like to report another error.

If one specifies toldfe and tolvrs in the input file, then this is not very handy.

$ grep tol pd.in
toldfe 1.000000e-06 eV
tolvrs 1.0E-16

However one would expect a descent error message about this mistake.
But this is not the case if one runs the job in parallel. It took me at least an hour to discover the mistake and it was only after I decided to check if the job runs in serial. Only then a human readable error message appeared.

In the output files of the parallel run, no single hint to the mistake can be found.

The tails of output files of the parallel run:

$ tail -f pd.txt
.Starting date : Tue 18 Dec 2018.
- ( at 09h49 )

- input file -> pd.in
- output file -> pd.txt
- root for input files -> pdi
- root for output files -> pdo

- inpspheads : Reading pseudopotential header in XML form from ./Pd.GGA_PBE-JTH.xml
- inpspheads : Reading pseudopotential header in XML form from ./O.GGA_PBE-JTH.xml

$ tail -f pd.log
vectors, bravais(1)= 3, has a lower symmetry than before,
but is still more symmetric than the real one, iholohedry= 1
obtained by taking into account the atomic positions.
...


symlatt : the Bravais lattice is aP (primitive triclinic)
symspgr : spgroup= 2 P-1 (=Ci^1)
ingeo: angdeg(1:3)= 90.00000000 90.00000000 60.00000000
invars2: take the default value of fband= 5.00000000E-01

$ cat stderr.111217
forrtl: No such file or directory
forrtl: No such file or directory
application called MPI_Abort(MPI_COMM_WORLD, 13) - process 11
Matthieu Verstraete
University of Liege, Belgium

Locked