TURBOMOLE Users Forum
Forum General => Miscellaneous => Topic started by: cediev87 on February 03, 2015, 12:16:45 AM
-
Hello,
I am trying to run a geometry optimization on a water molecular applying a constant external electric field on it. The SCF converges fine however, when I reach the optimization step it never states it is fully converged even though the energies between iteration no longer change within the set convergence criteria.
I setup the calculation in the GENERAL MENU of define under e : DEFINE EXTERNAL ELECTROSTATIC FIELD. I turned geofield and man from F to T and set the direction and absolute value to 0 0 1 0.1 (varying 0.1 to 0.1, 0.01 and 0.001...) The molecule is orientated in the xy-plane.
------------------------------------------------------------------------
*** specification of electrostatic field(s) ***
------------------------------------------------------------------------
<x> <y> <z> <E> : define field by direction and absolute value
<Ex> <Ey> <Ez> : specify field by components
type q(uit) or * to terminate
Am I missing any other key words to make this optimization successful?
Best,
Crystal
-
it never states it is fully converged even though the energies between iteration no longer change within the set convergence criteria.
You are aware that the energy is not the only convergence criteria?
Look in the statpt part of job.last for a table that looks like the following
Converged? Value Criterion
Energy change yes 0.0000000 0.0000010
RMS of displacement yes 0.0000000 0.0005000
RMS of gradient no 0.0314991 0.0005000
MAX displacement yes 0.0000000 0.0010000
MAX gradient no 0.0771566 0.0010000
The values for energy and gradients are for some reasons additionally written at the end of job.last (there at least with more significant digits)
energy change : actual value = -0.6500E-08 threshold = 0.1000E-05
geom. gradient : actual value = 0.7716E-01 threshold = 0.1000E-02
This should give you a hint why it does not state that it is fully converged. If it is a problem with the gradient you can check its behavior by
grep cycle gradient
In my simple test I could see that the gradient became constant but not close to 0. I'm not sure about the implication of the calculated gradient if an external electric field is implied, therefore I don't know if/how one can get it converged (in my naive picture the external electric field always creates a force on a polar molecule, maybe one can somehow subtract it or so)
-
Thank you for the replay. I want to see the changes in the molecule upon applying an external electric field to it, water has been just a test case so far.
What units is the magnitude of the electric field in? MV/cm?
-
What units is the magnitude of the electric field in? MV/cm?
I don't know (!) but how do you come to your guess of MV/cm? The SI-unit would be V/m but my guess would be atomic units (E_h/a_o, where 1 E_h/a_o = 5.14220652*10^11 V/m, see here http://physics.nist.gov/cgi-bin/cuu/Value?auefld). This is also stated in the section about the electric field here http://hincklab.uthscsa.edu/html/soft_packs/msi_docs/insight980/turbo/E_Cmds_SA.html but I don't know if this information is (still) correct.
If so, a value of 0.1 would be quite strong (and would explain why there is such a big gradient)... But as I said, I'm not 100% sure about the units.
-
Hello,
the default units in Turbomole are atomic ones, so Hauke is right. A 1.0 in the input therefore corresponds to approx. 5142 MV/cm if you prefer this unit.
Regards,
Uwe
-
I saw MV/cm as default in another program but there could be a typo (maybe they meant mV/cm) Thank you very much for the replies.
Best,
Crystal