Re: NPT then NVT simulation in a single config file for water box?

From: Tanooj Shah (shaht2_at_rpi.edu)
Date: Sun Nov 01 2020 - 09:36:15 CST

Just as a follow up to this question in case anyone read it - I did
triage the modified simulation box size after the first NPT run and
after using that in the NVT simulation did get a consistent reading for
the density of TIP3P water (0.98 g/cm3). I am guessing this was because
the void left after deleting the sodium ion needed to be filled. I
suspect there must be a way to write a pgn script to replace the sodium
ion with a water molecule without all three water atoms overlapping.
However I would still like to ask if it possible to add a solvation box
with "no" molecules (i.e. an empty pdb and psf input file).

Best,
Tanooj

On 2020-10-30 22:52, Tanooj Shah wrote:
> Hello,
>
> I wanted to ask if it is possible to run an NPT simulation for a
> certain number of frames and then subsequently an NVT simulation
> starting from the coordinates at the end of the NPT part. I believe it
> is possible to do this in separate config files but I am getting some
> inconsistent results which is why I wanted to ask if it possible to
> modify the config file to perform both (similar to how in LAMMPS you
> would write a fix, unfix and then another fix command).
>
> Specifically, I am trying to run a simulation of just a water box. It
> didn't seem clear how to create a box of water purely from scratch so
> I did it in a roundabout way - I wrote a pgn script to create a sodium
> ion, added a solvation box around it, and then wrote another pgn
> script to delete the sodium ion. Not sure if there is a cleaner way to
> fill a simulation box with water and nothing else.
>
> I initially ran an NPT simulation at T=300K,
> LangevinPistonTarget=1.01325 bar (the default) and averaged the
> density of the water in the box over all frames and got a result of
> around 0.94 g/cm3, which is lower than the expected water density for
> the TIP3P model (0.98 g/cm3). I increased the LangevinPistonTarget
> until I was able to get a stable density of 0.98 g/cm3 (this was at a
> LangevinPistonTarget of around 875 atm, I had to try multiple
> different values to get the "correct" density).
>
> I used the last frame of the .coor file as an input pdb file (and
> generated a corresponding psf file) to an input to an NVT simulation
> with a slightly reduced cell size (to match the compression from the
> NPT run). However at the end of the NVT run I am still getting a
> density too low - around 0.92 g/cm3. However I suspect this might be
> because the new cell size is not exact.
>
> I have two questions mainly - one, is is it possible to run an NPT and
> an NVT simulation in the same config file? And the second is more
> general and not sure if it is appropriate to post here - why am I
> getting a simulation lower than 0.98 g/cm3 for water at STP? Is this
> happening because when I deleted the sodium ion that created a void?
> Is there a way to create a solvation box for "no" atoms so the entire
> simulation box is uniformly filled with water?
>
> Thank you if you took the time to read this entire message.
>
> Best,

-- 
Tanooj Shah
Materials Science Department
Rensselaer Polytechnic Institute

This archive was generated by hypermail 2.1.6 : Thu Dec 31 2020 - 23:17:14 CST