Hi George,
To add to Carlos' comments, the control settings for pair_style reax/c
are exactly the same as those in the stand-alone reaxff code, so using
the NULL keyword "cfile" command is the best starting point. More
comments below, thanks.
Ray
Hi Ray,
Thanks ! I have not built lammps with user-reaxc. I shall do that soon and
see how I can run it. Just curious if it is preferable to use pair_style
reaxc to pair_style reax ? or they are exactly the same... from my reading
I thought there is no difference between them except for the implementation.
> Hello,
>
> I am using reaxFF. I have a few questions:
Which one? reax or reax/c?
reax
> (i) I figured that if I do NTP with 0.001 fs timestep, I have small
> fluctuation in pressure (+/- 500 atmosphere) and temperature.... even
with
> 0.1 fs timestep the pressure fluctuation is enormously large (100K
> atmosphere). The average pressure is always zero. Wonder if it is normal
?
I think 0.001 fs is an unnecessary small time step for reaxff.
Pressure fluctuation is normal, and smaller the system larger the
fluctuation.
So, you believe a pressure fluctuation of +/- 100,000 atmosphere is ok. I
see some waviness in structure which I interpreted as an effect of high
pressure fluctuations.
>
> (ii) After using NPT and heating the system to 500K, I run it a million
step
> with the same above parameters (timestep = 0.001 fs). I see that
> temperature, energy, enthalpy are stable. However, fluctuations in volume
> and pressure is as before.
This adds up to only 1 ps. Hard to say if it is really equilibrated,
and by equilibrated I mean minimal fluctuation is observed.
>
> (iii) Finally, I run NVE to see if the system is equilibrated ? This
time I
> run with timestep = 0.1 fs for 100K steps (10 ps). The temperature
started
> rising and rose from 500 K to 700 K. Pressure fluctuation is now +/- 80K
> atmosphere. Being NVE, the volume and energy is almost constant.
>
"volume" is almost constant with NVE? Then you are definitely doing
something wrong otherwise volume should not fluctuate at all. If
total energy remains more or less constant, then it should be fine.
I was inaccurate in my earlier mail. The volume is constant. Energy does
fluctuate very little. But the pressure fluctuation is very large ~ +/-
80,000 atmosphere and temperature increased from 500 Kelvin to 700 Kelvin.
From (ii) I would believe, the system is equilibrated. However, from (iii)
> it looks like my system is not equilibrated. If I choose the same small
> step size, I would need to do 10 million steps to equilibrate the system
> (because I typically see 10-25 ps for the equilibration in literature).
But they don't usually use a timestep as small as 0.001 fs.
>
> (iv) I have tried "velocity all create 300 12345 mom yes rot no"
> after minimization to quickly heat the system. But it never worked for my
> system. The program stops with error message. Basically system breaks
down
> with huge movement of atoms. I guess SHAKE is not allowed for reaxFF so
that
> is not an option.
velocity create should work, and if your run stops with error message,
then something terribly wrong must have occurred. What kind of error
message?
ia(i1,2) = 21 reax_defs.h::MBONDDEF = 20 after
705 of 2232958 pairs completed.
STOP Too many bonds on atom. Increase MBONDDEF
ia(i1,2) = 21 reax_defs.h::MBONDDEF = 20 after
540 of 2232856 pairs completed.
STOP Too many bonds on atom. Increase MBONDDEF
ia(i1,2) = 21 reax_defs.h::MBONDDEF = 20 after
729 of 2307413 pairs completed.
STOP Too many bonds on atom. Increase MBONDDEF