[gmx-users] problem to execute the parallelized CPMD in Gromacs/CPMD interface

[Yun-an Yan]

Dear all,

I have problems to execute the parallelized CPMD in Gromacs/CPMD  
interface.
The Gromacs is build without support of parallelization and CPMD is  
parallelized.
mdrun exits after displaying the error message:

yunan at baldwin$ cat output_long.md
..........
.........

****************************************************************
*   ATOM       MASS   RAGGIO NLCC              PSEUDOPOTENTIAL *
*     H      1.0080        1.2000   NO              VANDERBILT      *
*     O     15.9994      1.2000   NO              VANDERBILT      *
*     C     12.0112      1.2000   NO              VANDERBILT      *
****************************************************************

PARAPARAPARAPARAPARAPARAPARAPARAPARAPARAPARAPARAPARAPARAPARAPARA
mpiexec_baldwin.mpg.uni-rostock.de: mpd_uncaught_except_tb handling:
  <type 'exceptions.IOError'>: [Errno 9] Bad file descriptor
    /usr/local/mpich64/bin/mpirun  1065  handle_stdin_input
        sys.stdin.flush()  # probably does nothing
    /usr/local/mpich64/bin/mpdlib.py  762  handle_active_streams
        handler(stream,*args)
    /usr/local/mpich64/bin/mpirun  515  mpiexec
        rv = streamHandler.handle_active_streams(timeout=1.0)
    /usr/local/mpich64/bin/mpirun  1423  <module>
        mpiexec()


The content of the runcpmd is like that:
yunan at baldwin$ cat runcpmd
#! /bin/sh

/usr/local/mpich64/bin/mpirun -np 4 /usr/local/bin/cpmd_mpi.x  
CPMD_inp.run .

It runs well when I touch the file QMSTART and execute runcpmd directly.

Does anyone have some ideas to figure it out?

Thank you so much for your consideration.

Best wishes,
Yun-an Yan