[CP2K:384] Block after " Adding QM/MM electrostatic....."

[Teodoro Laino]

Just a summary: Luca tried on another machine always using 8 cpus,  
the qmmm job runs smoothly without any problem..
Luca can for sure comment more about that..

Anyway here there are few tips about the input Luca sent to the  
google group:

> &FORCE_EVAL
>
>     &SCF
>       MAX_SCF 150
>       EPS_SCF 1.0E-6
>       SCF_GUESS ATOMIC
>     &END SCF
>
This SCF setup is going to use the normal diagonalization/mixing  
scheme, which is ok bu maybe not so stable and efficient.
One may want to try also the OT (see reference at the end of cp2k  
run). For large scale simulations this is the way to go:

     &SCF
       EPS_SCF 1.0E-6
       SCF_GUESS RESTART
       MAX_SCF 30
       &OT
         PRECONDITIONER FULL_SINGLE_INVERSE
         MINIMIZER DIIS
         N_DIIS 7
       &END
       &OUTER_SCF
         MAX_SCF 5
         EPS_SCF 1.0E-6
       &END
       &PRINT
         &RESTART
            FILENAME =RESTART
            EACH 1 1 0
            ADD_LAST NUMERIC
            BACKUP_COPIES 0
         &END
         &RESTART_HISTORY OFF
         &END
       &END
     &END SCF

Use OT in conjuction with an OUTER SCF loop..
RESTART files (for IO reasons) should be printed only when the SCF is  
converged.
A lot of trials can be done on the choice of the PRECONDITIONER..

>
>    &POISSON
>      &EWALD
>        EWALD_TYPE NONE
>        ALPHA 0.44
>        GMAX   64
>        O_SPLINE 6
>      &END EWALD
>      PERIODIC NONE
>    &END POISSON
>
For classical run PERIODIC NONE has never been tested.. I just tried  
and it gives wrong numbers.. so for the time being (while I work on  
the bug, which has anyway a very low priority) calculations asking  
for periodic NONE in the MM module will abort.. you can only get full  
periodicity in MM runs.

>
>  &QMMM
>     MM_POTENTIAL_FILE_NAME MM_POTENTIAL
If you specify USE_GEEP_LIB you don't need to specify any  
MM_POTENTIAL_FILE_NAME..
Unless you don't want to do something really exotic use always  
GEEP_LIB..

>     &CELL
>      UNITS ANGSTROM
>      ABC 12.0 12.0 18.0
>      PERIODIC NONE
this periodic NONE affects the PBS for QM calculations (neighbor  
lists, etc..)
If you use this one you SHOULD use also to use a proper decoupling  
for the electrostatic term.

>     &END CELL
>     NOCOMPATIBILITY T
>     E_COUPL GAUSS
>     USE_GEEP_LIB 10
>
>   &SUBSYS
>
>     &CELL
>       ABC  64.000 64.0 57.000
>       UNIT ANGSTROM
>       PERIODIC NONE

Again in a QMMM run PERIODIC none is affecting the PBC only for MM  
energy/force evaluations..
>     &END CELL
>
>     &TOPOLOGY
>       COORD_FILE_NAME   As_THI_T2.pdb
>       COORDINATE PDB
>       CONN_FILE_NAME     As.psf
>       CONNECTIVITY PSF
>     &END TOPOLOGY
>
PSF: If the psf file is just one of the many format available please use

CONNECTIVITY UPSF

instead of

CONNECTIVITY PSF

the PSF keywords supports only the EXTended PSF format (which has a  
peculiar format compared to for example to the PSF written by XPLOR!!)
For example "normal" PSF (like the one of XPLOR or NAMD) should be  
used with the keyword UPSF..

Teo