The ABINIT Users Mailing List ( CLOSED )

["Raty Jean-Yves" <jyraty@ulg.ac.be>]

Sven,

If you use one of the two thermostat (Langevin/Nose), symmetry is of no
use (velocities are not constrained).

> Dear ABINIT users,
>      I am trying to do a finite temp. MD calculation in a Si surface
> system. From the ouput of a small run, I have the following questions,
> Any help would be very useful.
>
> 1. Should I use nsym=1 for finite T MD calculations? The default seems
> to give errors because the symmetry finder cannot find symmetry related
> atoms after 1 or 2 MD steps.
>
> 2. Is the effective temp. of the system printed out somewhere after
> every MD step? It does say "converged velocities for T=... Ha" after
> every step, but this value of T remains constant and is *exactly* equal
> to the input temp. It seems unlikely that this is the actual temp. of
> the
> system. One does expect some fluctuations for a finite system. It prints
> an "Effective temperature" only at the beginning of the calculation.

What is printed is the temperature that is asked by the driver (eventually
with a ramp), and what is called TKA in the output file. It is the average
kinetic energy DIVIDED by the thermostat temperature. I used this quantity
because it represents directly the fluctuations around the temperature
required by the thermostat (so, the actual temperature is the product of
this TKA and the temperature of the thermostat, at each step.)

> 3. In *a few* MOLDYN steps, the SCF cycle does not converge within the
> allowed number of steps. In fact, in these cases, the total energy
> oscillates between two values, and there does not seem any tendency of
> convergence of the SCF cycle. Has anybody else come across such behavior
> for any other system?

Yes, many times when a semiconductor system gets disordered by heating. So
I usually use a metallic occupation, together with a value of tsmear that
roughly corresponds to the temperature of the thermostat.

> 4. How do I realise if I am using the right number for the Nose' mass
> (the parameter noseinsert)?


Only try and error (by monitoring the behavior of the fluctuations -> TKA.)

I will add to other considerations that it is perfectly possible to do
mol. dynamics with systems of about 100 atoms with abinit. The main
difference in speed with other programs (besides their use of ultrasoft)
is found in the absence of parallelism for simulations at the gamma-point
only. (and I used both CPMD and abinit)

Regards,

Jean-Yves Raty


>      Thanks in advance for any help.
>
>    Sincerely,
>    Prasenjit Sen
>
> P.S. I am enclosing my input file below
>
>  =======================================================================
>   Harish-Chandra Research Institute     Phone: +91-532-2667509 Extn.
> 2007 Chhatnag Road, Jhunsi                  Or    +91-532-2667511*2007
> Allahabad, UP 211019
>        INDIA                            Fax:   +91-532-2568036
>  =======================================================================
> ==== basic input variables ===
> acell   3.818400000   3.818400000   15.269900000   angstrom
> chkprim  0
> diecut   2.3
> diemac   12.0
> diemix    0.5
> ecut   450.0 eV
> iprcel   45
> iscf  3
> ixc   1
> kptopt   1
> natom   28
> nband   55
> nline   5
> kptrlatt
>         2  0  0
>         0  2  0
>         0  0  1
> nstep   60
> shiftk
>        0.500  0.500  0.000
> ntypat   2
> occopt   1
> rprim
>        2.0000000000000000    0.0000000000000000    0.0000000000000000
> 0.0000000000000000    2.0000000000000000    0.0000000000000000
> 0.0000000000000000    0.0000000000000000    1.0000000000000000
> toldff  5.0d-6
> typat   20*1  8*2
> xred
>               0.0000000000E+00  1.5996995304E-01  3.7126247555E-01
> 0.0000000000E+00  4.4379593651E-01  4.1979580693E-01
> 5.0000000000E-01  1.5996995304E-01  3.7126247555E-01
> 5.0000000000E-01  4.4379593651E-01  4.1979580693E-01
> 2.5000000000E-01  1.3922297530E-02  3.2293877643E-01
> 2.5000000000E-01  4.8813582203E-01  3.2805145313E-01
> 7.5000000000E-01  1.3922297530E-02  3.2293877643E-01
> 7.5000000000E-01  4.8813582203E-01  3.2805145313E-01
> 2.5000000000E-01  2.6105290841E-01  2.2538172919E-01
> 2.5000000000E-01  7.4766515713E-01  2.4170413180E-01
> 7.5000000000E-01  2.6105290841E-01  2.2538172919E-01
> 7.5000000000E-01  7.4766515713E-01  2.4170413180E-01
> 0.0000000000E+00  2.4979029122E-01  1.3886883731E-01
> 0.0000000000E+00  7.5069430588E-01  1.4950521475E-01
> 5.0000000000E-01  2.4979029122E-01  1.3886883731E-01
> 5.0000000000E-01  7.5069430588E-01  1.4950521475E-01
> 0.0000000000E+00  0.0000000000E+00  5.6968284010E-02
> 0.0000000000E+00  5.0000000000E-01  5.6968284010E-02
> 5.0000000000E-01  0.0000000000E+00  5.6968284010E-02
> 5.0000000000E-01  5.0000000000E-01  5.6968284010E-02
>              -1.5055385507E-01 -1.1738360888E-02 -3.5839766227E-03
>               1.5055385507E-01 -1.1738360888E-02 -3.5839766227E-03
>              -1.5043461137E-01  5.1252747451E-01 -3.6714917881E-03
>               1.5043461137E-01  5.1252747451E-01 -3.6714917881E-03
> 3.4944614493E-01 -1.1738360888E-02 -3.5839766227E-03
> 6.5055385507E-01 -1.1738360888E-02 -3.5839766227E-03
> 3.4956538863E-01  5.1252747451E-01 -3.6714917881E-03
> 6.5043461137E-01  5.1252747451E-01 -3.6714917881E-03
> zatnum   14.0   1.0
>
> === geometry optimization variables ===
> #amu use defaults
> dtion    100
> ionmov   8
> mditemp  700
> noseinsert  1.0d5
> nsym     1
> ntime    50