Messages

I would suggest to talk to experimental people and ask why their data goes up.
I expect this is due to adsorption at the external surface. You would expect that the results for adsorption should level off to a saturation value.

There are no standard methods. By making choices you need to create a structure with no symmetry (P1) so that all atom positions are defined.

Instead of a Box use a Framework.

The ideal gas Rosenbluth weight is only depending on temperature. For each temperature you can pre-compute it (see examples).

[per]

The 'CutOffVDW' should not be different for the gas and liquid phase, otherwise you're using different force fields for them.
The 'CutOffCoulomb' should be set to something slightly smaller than half the box size. This significantly speeds up the computation for the gas-phase since the gas-phase box can be large.
Importantly, both can be set per system (just list them after the 'Box' statement).

This for the PartialReinsertionProbability move, where only a few atoms are regrown and the others are left fixed.
So, in this case there are 4 config move defined (and you see 5 lines). For each line you have the number of fixed atoms, and the the list of the fixed atom ids.
The other atoms are regrown using CBMC. This move has high efficiency, because the molecule is already present and is only partially regrown.
Note that in CBMC all branches need to be grown at the same time, so you can not keep more than one branch fixed (in a branched molecule).

An "MD path" is the path or trajectory that a particle follows in MD. if you ran too short, then the molecules have not moved enough.  Diffusion is defined as the limiting behavior of the Mean-Squared Displacement (MSD) at long-times. For diffusional processes, the slope must become linear at long times. If you did not reach that regime yet, then you have ran too short.
Slow diffusion might be outside the reach of MD, e.g. very slow diffusion can not be measured using MD. The maximum is probability up to 10^-14 m^2/s.

In general no, because diffusion coefficients are by definition the long-time behavior of molecules.
In any short MD-path you might even see opposite behavior. Of course for illustrative purposes you could just select an MD path that shows what you want to show.

These radii are only needed for flexible frameworks, and used as bond-criteria to determine bonding.
For other cases these values do not matter.

See literature on how to compute diffusion coefficients from MSD. The slops of the MSD at long times can not be negative.
You're probably looking at a situation where MD is not capable of computing diffusion because it is too slow.

[chemical element]

About the gif-file: The _atom_site_type_symbol needs to be the chemical element. The _atom_site_label is used in the forcefield definition as a label of the atom.

All atom-types need to be defined in the 'pseudo_atoms.def', so 'c', 'c3', 'hc', but also all atoms of the probe molecule.
Next define LJ parameters all of these for 'force_field_mixing_rules.def'.
The mixing rules with the be used to computed cross-interactions between all of the atoms.

You seems to have placed them in 'force_field.def' which is means to overwrite interactions for systems where mixing rules are not applicable. But these are no interactions defined then between {'c', 'c3', 'hc'} and all of the adsorbate atoms of the probe molecules.

Did you define the forcefield for your framework atoms?
(i.e. all Lennard-Jones parameters for all your atom types, and all the charges)

With
grep RestrictMovesToCylinder *.c
you can see the files where it is used. Input is in 'input.c'.

So something like:
RestrictMovesToCylinder yes
Cylinder 0
CylinderCenter 0.5 0.5 0.5
CylinderRadius 3.0
CylinderDirection X

It is used in molecule.c
(grep RestrictCylinderDirection *.c)
In the routine:
int ValidCartesianPoint(int i, POINT pos);

'Molecules per unit cell' is the the number of molecules in the 1x1x1 system.
Often experiments uses mol/kg. RASPA prints out als other units like mg/g.