GCMC

[Matt McGrath]

CP2K has a pretty consistent naming scheme.  Usually, routines related
to a particular topic begin with TOPIC_.  In this case, the Monte
Carlo ensemble routines are in mc_ensembles.F.  From there you should
be able to figure out how the move types are selected (the definition
of an ensemble is just the selection of different move types).

                          Cheers, Matt

On Oct 1, 3:02 am, Isaak Daniels <isaakdan... at gmail.com> wrote:
> I think I will try to implement GCMC into the source code. What files
> should one modify if one wants to introduce a new ensamble?
>
> Thank you
>
> On Sep 3, 12:56 am, Matt McGrath <obfis... at gmail.com> wrote:
>
>
>
>
>
>
>
> > Hi.
>
> > That's not possible.  The more molecules you have, the less the
> > density will change, but it will always change by a little.
>
> >                                            Cheers, Matt
>
> > On Sep 3, 4:24 am, Isaak Daniels <isaakdan... at gmail.com> wrote:
>
> > > How does one make sure the density isn't changed in the vapor box when
> > > one moves a molecule out of it?
>
> > > Thank you
>
> > > On Aug 30, 11:21 pm, Matt McGrath <obfis... at gmail.com> wrote:
>
> > > > The chemical potential of an ideal gas is a function of the
> > > > temperature and number density (N/V).  So, if you have a chemical
> > > > potential you want at a certain temperature, you can solve that
> > > > equation for the number density.  Once you have the number density,
> > > > you decide how many vapor molecules you want (400, 500 maybe) and
> > > > calculate the vapor box volume from that.  Then you run the simulation
> > > > and make sure the vapor density does not change over the course of the
> > > > simulation, because if it does, your chemical potential changes as
> > > > well.
>
> > > >                              Cheers, Matt
>
> > > > On Aug 31, 12:14 am, Isaak Daniels <isaakdan... at gmail.com>
> > > > wrote:
>
> > > > > What variables should one vary to affect the chemical potential?
>
> > > > > Thank you
>
> > > > > On Aug 30, 2:41 am, Matt McGrath <obfis... at gmail.com> wrote:
>
> > > > > > Nope, not possible.  I never saw a reason for having such a feature
> > > > > > (zero particles in the Gibbs ensemble vapor box is still a valid
> > > > > > configuration).
>
> > > > > >                                 Cheers, Matt
>
> > > > > > On Aug 30, 12:51 am, Isaak Daniels <isaakdan... at gmail.com>
> > > > > > wrote:
>
> > > > > > > Thank you. Is there a way to stop the program from aborting when there
> > > > > > > are no molecules left in one box?
>
> > > > > > > Thank you
>
> > > > > > > On Aug 29, 12:44 am, Matt McGrath <obfis... at gmail.com> wrote:
>
> > > > > > > > Hi Issak.  Neither of those is possible at the moment.  The only
> > > > > > > > option is to write a script to generate the coordinates yourself (for
> > > > > > > > example, creating a grid inside a box the size of your desired
> > > > > > > > simulation box, and putting a molecule on each gridpoint, and then
> > > > > > > > using those coordinates as the input to one of the boxes in CP2K).  If
> > > > > > > > you are using an ideal gas box (i.e. all the FIST interactions are set
> > > > > > > > to zero), you don't have to equilibrate the coordinates at all, which
> > > > > > > > is a small help.  Otherwise you should run some short equilibration
> > > > > > > > (not letting the molecules swap) before doing a real run.
>
> > > > > > > >                                 Cheers, Matt
>
> > > > > > > > On Aug 29, 2:36 am, Isaak Daniels <isaakdan... at gmail.com> wrote:
>
> > > > > > > > > Also, how do you do it so that one can specify a number of molecules
> > > > > > > > > without needing coordinates?
>
> > > > > > > > > Thank you for helping me.
>
> > > > > > > > > On Aug 28, 11:06 am, Isaak Daniels <isaakdan... at gmail.com>
> > > > > > > > > wrote:
>
> > > > > > > > > > Just to confirm, there is no way to directly have an initial number of
> > > > > > > > > > molecules set that can be changed, for example doing 400 gas molecules
> > > > > > > > > > that can go in and out of boxes?
>
> > > > > > > > > > Thank you
>
> > > > > > > > > > On Aug 28, 12:46 am, Matt McGrath <obfis... at gmail.com> wrote:
>
> > > > > > > > > > > Hi Issak.  You can find the average number of molecules in each box by
> > > > > > > > > > > writing a script to average the mc_molecules file that is printed
> > > > > > > > > > > out.  The format is
>
> > > > > > > > > > > Step number     # of molecules of type 1        # of molecules of type
> > > > > > > > > > > 2 ..... etc, for box 1
> > > > > > > > > > > Step number     # of molecules of type 1        # of molecules of type
> > > > > > > > > > > 2 ..... etc, for box 2 (if there is a second box in the system)
>
> > > > > > > > > > > At one time this was printed out at the end of a run, but for some
> > > > > > > > > > > reason I removed it.  There is no way to set this number via the
> > > > > > > > > > > pressure...you just need to change the box volume, run the simulation,
> > > > > > > > > > > see what the average density is (assuming the simulation has
> > > > > > > > > > > equilibrated), change the box volume so that average density will give
> > > > > > > > > > > the desired number of molecules, and then rerun.  If the number of
> > > > > > > > > > > molecules in the box is very small, there will be large fluctuations
> > > > > > > > > > > around this number over the course of the run, but assuming the
> > > > > > > > > > > simulation is long enough and equilibrated, the average should be what
> > > > > > > > > > > you desire.
>
> > > > > > > > > > > The easiest way of keeping a large molecule rigid is to change the
> > > > > > > > > > > _MOL probabilities in the input file.  For example, if the big
> > > > > > > > > > > molecule is type 1 and the gas atoms are type 2, then a line of
>
> > > > > > > > > > > PMSWAP_MOL 0.0 1.0
>
> > > > > > > > > > > will let the gas molecules swap between boxes but not the big
> > > > > > > > > > > molecule.  Similarly,
>
> > > > > > > > > > > PMTRAION_MOL 0.0 1.0
>
> > > > > > > > > > > will attempt to change the conformation of molecule 2 but not molecule
> > > > > > > > > > > 1.  If you want 30% of molecule translations to be done on molecule 1
> > > > > > > > > > > and 70% on molecule 2, a line like
>
> > > > > > > > > > > PMTRANS_MOL 0.3 1.0
>
> > > > > > > > > > > would work.  Hope that clarifies things.  The file, tests/MC/regtest/
> > > > > > > > > > > GEMC_NVT_box1.inp uses this, since it has two molecule types.
>
> > > > > > > > > > >                                Cheers, Matt
>
> > > > > > > > > > > On Aug 27, 6:26 am, Isaak Daniels <isaakdan... at gmail.com> wrote:
>
> > > > > > > > > > > > Sorry for not being clear. By crystal, I mean a large molecule I want
> > > > > > > > > > > > rigid, and when I say atoms, I meant the gas atoms.
>
> > > > > > > > > > > > Thank you
>
> > > > > > > > > > > > On Aug 26, 4:10 pm, Isaak Daniels <isaakdan... at gmail.com> wrote:
>
> > > > > > > > > > > > > Also, how does one make sure the crystal is rigid but the atoms can go
> > > > > > > > > > > > > in and out of each box?
>
> > > > > > > > > > > > > Thank you for answering my questions
>
> > > > > > > > > > > > > On Aug 26, 3:07 pm, Isaak Daniels <isaakdan... at gmail.com> wrote:
>
> > > > > > > > > > > > > > And one would set it by using the Pressure?
>
> > > > > > > > > > > > > > Thank you
>
> > > > > > > > > > > > > > On Aug 26, 11:12 am, Isaak Daniels <isaakdan... at gmail.com>
> > > > > > > > > > > > > > wrote:
>
> > > > > > > > > > > > > > > Is there a way to find the average number of molecules in each box for
> > > > > > > > > > > > > > > a GEMC simulation?
>
> > > > > > > > > > > > > > > Thank you
>
> > > > > > > > > > > > > > > On Aug 25, 2:23 am, Matt McGrath <obfis... at gmail.com> wrote:
>
> > > > > > > > > > > > > > > > The only way to set thechemical potentialis if you use the above
> > > > > > > > > > > > > > > > tricks and they work.  You can't set it directly, but you could choose
> > > > > > > > > > > > > > > > your ideal gas to have a certainchemical potential.  GEMC would
> > > > > > > > > > > > > > > > equilibrate thechemical potentialbetween the two boxes, and provided
> > > > > > > > > > > > > > > > your ideal vapor box is big enough (i.e. a true reservoir), its
> > > > > > > > > > > > > > > >chemical potentialwould not change, so therefore thechemical> potentialin both boxes would be (approximately) the same as what you
> > > > > > > > > > > > > > > > started with for your ideal gas.  Finding what is "big enough" would
> > > > > > > > > > > > > > > > probably take some effort, though, and there would be error bars
> > > > > > > > > > > > > > > > associated with the number, so this is not a perfect solution.
>
> > > > > > > > > > > > > > > >                                    Cheers, Matt
>
> > > > > > > > > > > > > > > > On Aug 24, 9:21 pm, Isaak Daniels <isaakdan... at gmail.com> wrote:
>
> > > > > > > > > > > > > > > > > Is there a way to set achemical potential, such as -454?
>
> > > > > > > > > > > > > > > > > Thank you
>
> > > > > > > > > > > > > > > > > On Aug 23, 10:08 pm, Matt McGrath <obfis... at gmail.com> wrote:
>
> > > > > > > > > > > > > > > > > > When you say "import coordinates of a crystal", do you mean as a swap
> > > > > > > > > > > > > > > > > > move?  As in, you want to have gas molecules and crystal molecules
> > > > > > > > > > > > > > > > > > come into the box, or that you put crystal coordinates into the box at
> > > > > > > > > > > > > > > > > > the beginning, treat the whole box as QM, and then swap gas molecules
> > > > > > > > > > > > > > > > > > into the box (also treated as QM)?  The latter option should be
> > > > > > > > > > > > > > > > > > possible if the test I described above worked (since the force_env of
> > > > > > > > > > > > > > > > > > each box can be specified separately), and is what I assumed you
> > > > > > > > > > > > > > > > > > wanted to do in the first post.  If you also want to swap in crystal
> > > > > > > > > > > > > > > > > > nuclei, you could do that by making it a new molecule type (and
> > > > > > > > > > > > > > > > > > creating a psf file for it), but then it would swap into the QM box in
> > > > > > > > > > > > > > > > > > a completely random position, and would still be...
>
> read more »