TURBOMOLE Users Forum
TURBOMOLE Modules => Jobex: Structure Optimization and Molecular Dynamics => Topic started by: 175116 on March 10, 2009, 04:32:11 PM
-
Hello, this may be quite a silly questions, but need to be sure.
1/ My input coordinates for optimization were in coord file. During jobex (during optimization) the former coordinates in coord file are replaced in every single step by the new ones?
2/ If the optimization was successful and I need only the last and optimized geometry in xyz format (in Angstrom), do I obtain it using " t2x -c > optimized_geometry.xyz " in the working directory?
Thank you very much indeed!
-
Hi,
0. there are no silly questions
1. yes
2. yes
the coordinates of all steps of a geometry optimization can be found in the gradient file - that is also the source of information if you call t2x without the -c option.
Uwe
-
Hello everyone:
I am a new comer for turbomole forum, and I don't know where should I suggest my topic, so I reply them here. Is there anyone who can help me about my problem encountering in the usage of the program?
There is an error in optimization by ri-dft that always confused me, and I don't know how to handle it, please help me about. Thank you very much! The input and the output of the operation have been added in the following section, and these results have been obtained under the order of the "jobex -ri" without the single point calculation by ridft. Is there something wrong with my input files or any problem with our program? Thanks again for your help!
Input:
$title
benzene,hf
$operating system unix
$symmetry d6h
$redundant file=coord
$coord file=coord
$intdef file=coord
$user-defined bonds file=coord
$atoms
c 1-6 \
basis =c def-SV(P) \
jbas =c def-SV(P)
h 7-12 \
basis =h def-SV(P) \
jbas =h def-SV(P)
$basis file=basis
$rundimensions
dim(fock,dens)=5379
natoms=12
nshell=48
nbf(CAO)=102
nbf(AO)=96
dim(trafo[SAO<-->AO/CAO])=808
rhfshells=1
$scfmo file=mos
$scfiterlimit 30
$thize 0.10000000E-04
$thime 5
$scfdump
$scfintunit
unit=30 size=0 file=twoint
$scfdiis
$scforbitalshift automatic=.1
$drvopt
cartesian on
basis off
global off
hessian on
dipole on
nuclear polarizability
$interconversion off
qconv=1.d-7
maxiter=25
$optimize
internal on
redundant off
cartesian off
global off
basis off logarithm
$coordinateupdate
dqmax=0.3
interpolate on
statistics 5
$forceupdate
ahlrichs numgeo=0 mingeo=3 maxgeo=4 modus=<g|dq> dynamic fail=0.3
threig=0.005 reseig=0.005 thrbig=3.0 scale=1.00 damping=0.0
$forceinit on
diag=default
$energy file=energy
$grad file=gradient
$forceapprox file=forceapprox
$lock off
$closed shells
a1g 1-3 ( 2 )
e1g 1 ( 2 )
e2g 1-3 ( 2 )
a2u 1 ( 2 )
b1u 1 ( 2 )
b2u 1-2 ( 2 )
e1u 1-3 ( 2 )
$dft
functional b-p
gridsize m3
$scfconv 6
$scfdamp start=0.700 step=0.050 min=0.050
$ricore 200
$rij
$jbas file=auxbasis
$actual step statpt
$end
Output:
OPTIMIZATION CYCLE 1
Sat Dec 22 10:00:58 CST 2012
Parallel program rdgrad_mpi will be taken out of the TURBODIR directory.
STARTING rdgrad ON 6 PROCESSORS!
RUNNING PROGRAM //usr/local/theochem/turbomole/v64/bin/em64t-unknown-linux-gnu_ga/rdgrad_mpi.
PLEASE WAIT UNTIL rdgrad HAS FINISHED.
Look for the output in slave1.output.
No file slave1.output found?
fine, there is no data group "$actual step"
script actual: unknown actual step define
next step = unknown
operating system is UNIX !
hostname is m1
statpt (m1) : TURBOMOLE V6.4 3 Apr 2012 at 16:44:13
Copyright (C) 2012 TURBOMOLE GmbH, Karlsruhe
2012-12-22 10:00:58.367
this is S T A T P T
hessian and coordinate update for
stationary point search
by barbara unterreiner, marek sierka,
and reinhart ahlrichs
quantum chemistry group
universitaet karlsruhe
germany
Keyword $statpt not found - using default options
*************** Stationary point options ******************
************************************************************
Maximum allowed trust radius: 3.000000E-01
Minimum allowed trust radius: 1.000000E-03
Initial trust radius: 1.500000E-01
GDIIS used if gradient norm < 1.000000E-02
Number of previous steps for GDIIS: 5
Hessian update method: BFGS
*** Convergence criteria ***
Threshold for energy change: 1.000000E-06
Threshold for max displacement element: 1.000000E-03
Threshold for max gradient element : 1.000000E-03
Threshold for RMS of displacement: 5.000000E-04
Threshold for RMS of gradient: 5.000000E-04
************************************************************
<getgrd> : data group $grad is missing
MODTRACE: no modules on stack
error reading energy and gradient in rdcor
statpt ended abnormally
statpt step ended abnormally
next step = statpt
operating system is UNIX !
hostname is m1
data group $actual step is not empty
due to the abend of statpt
statpt (m1) : TURBOMOLE V6.4 3 Apr 2012 at 16:44:13
Copyright (C) 2012 TURBOMOLE GmbH, Karlsruhe
2012-12-22 10:00:58.399
this is S T A T P T
hessian and coordinate update for
stationary point search
by barbara unterreiner, marek sierka,
and reinhart ahlrichs
quantum chemistry group
universitaet karlsruhe
germany
Keyword $statpt not found - using default options
*************** Stationary point options ******************
************************************************************
Maximum allowed trust radius: 3.000000E-01
Minimum allowed trust radius: 1.000000E-03
Initial trust radius: 1.500000E-01
GDIIS used if gradient norm < 1.000000E-02
Number of previous steps for GDIIS: 5
Hessian update method: BFGS
*** Convergence criteria ***
Threshold for energy change: 1.000000E-06
Threshold for max displacement element: 1.000000E-03
Threshold for max gradient element : 1.000000E-03
Threshold for RMS of displacement: 5.000000E-04
Threshold for RMS of gradient: 5.000000E-04
************************************************************
<getgrd> : data group $grad is missing
MODTRACE: no modules on stack
error reading energy and gradient in rdcor
statpt ended abnormally
statpt step ended abnormally
next step = statpt
operating system is UNIX !
hostname is m1
data group $actual step is not empty
due to the abend of statpt
statpt (m1) : TURBOMOLE V6.4 3 Apr 2012 at 16:44:13
Copyright (C) 2012 TURBOMOLE GmbH, Karlsruhe
2012-12-22 10:00:58.445
this is S T A T P T
hessian and coordinate update for
stationary point search
by barbara unterreiner, marek sierka,
and reinhart ahlrichs
quantum chemistry group
universitaet karlsruhe
germany
Keyword $statpt not found - using default options
*************** Stationary point options ******************
************************************************************
Maximum allowed trust radius: 3.000000E-01
Minimum allowed trust radius: 1.000000E-03
Initial trust radius: 1.500000E-01
GDIIS used if gradient norm < 1.000000E-02
Number of previous steps for GDIIS: 5
Hessian update method: BFGS
*** Convergence criteria ***
Threshold for energy change: 1.000000E-06
Threshold for max displacement element: 1.000000E-03
Threshold for max gradient element : 1.000000E-03
Threshold for RMS of displacement: 5.000000E-04
Threshold for RMS of gradient: 5.000000E-04
************************************************************
<getgrd> : data group $grad is missing
MODTRACE: no modules on stack
error reading energy and gradient in rdcor
statpt ended abnormally
statpt step ended abnormally
next step = statpt
-
@ Alisa Yang.
I have the same problem with Turbomole 6.4 (not 6.3). We use SLURM for the queue. Sometimes I have to restart the calculations ten times until it works. Qur adminitrator suggest thats may be a problem with mpi.
-
Hello,
Turbomole runs in parallel across different nodes. If it does not (or not always), please check:
- is your input located on a network disk, so that it is visible on all nodes used for the calculation?
- is the number of nodes and machines correct? Make sure that PARNODES is set to the number of CPUs provided by the queuing system.
- in most cases insufficient memory limits kill the jobs. If a calculation uses several nodes, the default stack size limit must be set in the system settings. It is not sufficient to add ulimit or limit to the queuing system start script or in your .bashrc or .profile file. See: http://www.turbo-forum.com/index.php?topic=23.0 (http://www.turbo-forum.com/index.php?topic=23.0)
- if you do not use one of the supported queuing systems (PBS, Platform LSF, SGE, IBM LoadLeveler), find out in which file the queuing system writes the list of hosts that shall be used and set $HOSTS_FILE to this file.
- if SLURM or any other system forces you to use srun or prun, you will have to change the scripts in $TURBODIR/mpirun_scripts. Under Linux search for the
case $MPIVER in
HP-MPI )
lines and adapt the options given to mpirun (or mpirun.mpich, depending on the Turbomole version you have).
Turbomole uses IBM Platform MPI http://www-03.ibm.com/systems/technicalcomputing/platformcomputing/products/mpi/index.html (http://www-03.ibm.com/systems/technicalcomputing/platformcomputing/products/mpi/index.html) which also supports srun and prun. The IBM Platform MPI documentation contains in a very decent form everything you need to use almost any queuing system or fast network interconnect.
- the MPI parallelization of dscf, grad, ridft and rdgrad needs an additional process which is responsible for the task distribution only. Some queuing systems deny MPI to start N+1 processes if only N are provided. In that case please reduce $PARNODES by one. This has been discussed several times here on the Forum already.
If all that does not help, just contact the support by sending an email to turbomole at cosmologic.de!
There you will get help and more detailed information.
To figure out what happens when Turbomole does not run in parallel, please do not run jobex as test case, because here different parallel binaries are called one after each other which can cause them to overwrite the original error message.
Regards,
Uwe
-
Maybe I'm wrong, but I suspect that this error crops up when at the same time on the node the program "GROMACS" is executed.
-
@ Alisa Yang: I think one can fix this problem with export MPI_IC_ORDER="TCP"