[QE-users] NEB fails

Denis E. Zavelev metalian at mail.ru
Thu Mar 15 16:44:59 CET 2018


Hello!

After finding 2 minima (coordinative and dissociative adsorption) I have to find transition state between them (or several transition states and intermediates), so I have to use NEB.
I am using QE 6.2.1 compiled on JSSC RAS MVS-10P supercomputer (I used Intel MKL and ifc to compile the binaries). After starting NEB it fails in less than an hour without any fair reason. 
So I installed QE locally (AMD FX-8150/32 Gb RAM (no swap!) under Linux Mint 17, I used gcc and gfortran, also blas/lapack/fftw installed in the system). And the same problems.

     initial path length           = 10.9146 bohr
     initial inter-image distance  =  1.5592 bohr

     string_method                 =    neb
     restart_mode                  =    from_scratch
     opt_scheme                    =    quick-min
     num_of_images                 =    8
     nstep_path                    =    150
     CI_scheme                     =    no-CI
     first_last_opt                =    F
     use_freezing                  =    F
     ds                            =    1.0000 a.u.
     k_max                         =    0.1000 a.u.
     k_min                         =    0.1000 a.u.
     suggested k_max               =    0.6169 a.u.
     suggested k_min               =    0.6169 a.u.
     path_thr                      =    0.0500 eV / A

     ------------------------------ iteration   1 ------------------------------

     tcpu =      2.1    self-consistency for image   1
--------------------------------------------------------------------------
mpirun noticed that process rank 3 with PID 19047 on node FX8150 exited on signal 9 (Killed).
--------------------------------------------------------------------------

PW computations of structures used as 1st and last images have no problems.
What can be possible reasons of such fails? 



Best regards,
Denis E. Zavelev
Research assistant, PhD Chem.
Laboratory of catalytic nanotechnologies
TIPS RAS
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.quantum-espresso.org/pipermail/users/attachments/20180315/718a7063/attachment.html>


More information about the users mailing list