[Pw_forum] Error while running metadynamics.

carlo sbraccia sbraccia at sissa.it
Tue Aug 8 22:38:38 CEST 2006


What you posted is not enough to understand what is wrong in your 
calculation: you should send (to the forum or to me directly) the entire 
input&output files.

 From the Doc/INPUT_PW file:

delta_t        REAL ( default = 1.D0 )
               delta_t = 1                 : every 'nraise' step the actual
                                             temperature is rescaled to 
tempw
               delta_t /= 1 && delta_T > 0 : at each step the temperature is
                                             multiplied by delta_t; this is
                                             done rescaling all the 
velocities.
               delta_t < 0                 : every 'nraise' steps 
temperature
                                             is reduced by -delta_T

Unless you want to change dynamically the temperature (not to be done in 
a meta-dynamics run) you shouldn't modify the default value of this 
keyword. Also, to run a finite temperature meta-dynamics simulation I 
suggest you to set nraise=1.

I add few more comments about meta-dynamics in Quantum-ESPRESSO:

1) The study of a 6-dimensional free-energy surface (FES) is an 
EXTREMELY expensive task (in general the computational cost of exploring 
an N-dim landscape grows exponentially with N). Maybe that, by using 
good collective variables (such as coordination numbers) instead of 
simple distances, you can reduce the dimensionality of the FES. In any 
case you should consider reducing the number of collective variables.

2) So far meta-dynamics has been used at T=0 only. Finite temperature 
simulations are possible, but not yet tested. Moreover, neither the 
mean-force corrections nor the dynamical corrections that arise at 
finite temperature are implemented (Ciccotti et al.; ChemPhysChem, 6, 
1809, 2005). This makes the finite-temperature meta-dynamics which is 
implemented in Quantum-ESPRESSO a method only suited for qualitative 
studies of the FES.

3) PWscf still lacks a  good thermostat like a Nose-Hoover chain. A good 
alternative can be the over-damped Langevin dynamics (see keyword 
ion_temperature in Doc/INPUT_PW) which, however, still requires some 
tuning. The Car-Parrinello code (cp.x) represents an excellent 
alternative for finite-temperature simulations.

Regards,

carlo

K.Hari Krishna Reddy wrote:
> Hi
> I'm doing metadynamics on dimethyl-ether(as a test case) using the pwscf
> code(-3.1.1 version). The following are some details from the input file
>
> COLLECTIVE_VARS
> 6
> "distance"  1   2
> "distance"  2   3
> "distance"  3   1
> "distance"  1   5
> "distance"  1   4
> "distance"  1   6
> .....................
>
> ATOMIC_POSITIONS { angstrom }
> C       -1.171165000     -0.196607000      0.000000000
> O        0.000000000      0.587847000      0.000000000
> C        1.174595000     -0.195947000      0.000000000
> H       -1.242668000     -0.832333000      0.897697000
> H       -1.242668000     -0.832333000     -0.897697000
> H       -2.020165000      0.493833000      0.000000000
> H        1.231895000     -0.836253000     -0.896096000
> H        1.231895000     -0.836253000      0.896096000
> H        2.021127000      0.495894000      0.000000000
>
> While running the calculation is stopped showing the following error message
>
> " from check_constraint : error #         1
>      on some constraint g = 0 is not satisfied"
>
> am a beginner in running dynamics..could anybody help me regarding this
> case. And I also want some explanation regarding the keyword "delta_t" in
> the name list &IONS.
> Thank You
>
>   




More information about the Pw_forum mailing list