Why is the structure shifted after a restart GCMC?

Dear all,

I ran GCMC calculations (water adsorption in a mesoporous material).
When I read the dump file with ovito, everything is fine. the pore is centered.

Then, as the adsorption was not converged I restarted it.
However, this time if I read the second dump file, the whole structure is shifted (since the 1st step of the trajectories)

Regarding the structure, since the system is periodic in 3 dimensions it is not an issue. However, I shall perform MD after that and for post-processing the trajectories it is cumbersome. I guess I can “recenter” the system before the MD run, but I guess I missed something in my GCMC restart.

variable	T equal 298 # initiale temperature, Kelvin
variable	dt equal 1.0 # fs unit 
variable    Tdamp equal 100
variable	seed equal 6354
variable	mu index -8.1 # not used since pressure keywords in GCMC is used
variable	P index 1.0e-04 # atm
variable	disp index 0.5 # not used
variable	tfac equal 5.0/3.0 # (3 trans + 2 rot)/(3 trans), from simon gravelle. Try without it

timer timeout 156:00:00 every 10000

read_restart      MCM41.${T}K.${P}atm.restart.2000000

variable	         zlo equal zlo
variable	         zhi equal zhi

kspace_style	pppm/tip4p 1.0e-5
region 	        tube cylinder z 0.0 0.0 10.0 ${zlo} ${zhi} 

# settings
group        wat type 1 2
molecule    h2omol H2O_TIP4P_2005.txt 

compute                 twat wat temp
compute_modify    twat dynamic/dof yes

fix                 mynvt wat nvt temp ${T} ${T} ${Tdamp}                      
fix_modify    mynvt temp twat
fix                 myshk wat shake 1.0e-4 200 0 b 1 a 1 mol h2omol

fix        mygcmc wat gcmc 100 100 0 0 54341 ${T} ${mu} ${disp} &
             mol h2omol tfac_insert ${tfac} group wat pressure ${P} region tube shake myshk

# dynamically update default temperature ndof
compute_modify    thermo_temp dynamic/dof yes

# for visualisation
dump         mydmp all atom 100 dump.MCM41.${T}K.${P}atm.GCMC.lammpstrj

timestep           ${dt}
thermo             100
thermo_style    custom step temp press pe ke density atoms v_iacc v_dacc v_tacc v_racc v_nO

restart 100000   MCM41.${T}K.${P}atm.restart
run     3000000

Unfortunately, it is impossible to give specific advice without seeing more details and being able to reproduce the issue. What you describe here is not consistent with how LAMMPS works, so we have to narrow down where things go into the wrong direction.

It is unlikely that this is related to GCMC and one cannot “see” this from just looking at your “production” input, so please try to produce a very minimal pair of input decks (e.g. based on the “melt” or “peptide” example) that show how you a) generate the restart after a very short run and b) how you continue and then c) how the visualization is “shifted”.

Please also describe in detail how exactly you visualize your trajectory data.

I will try to do that soon, thanks for your answer.

Actually, I could’nt reproduce the problem. I guess I made a typo somewhere for this run. Everything is fine for others.