search

prisms-center / CASMcode

First-principles statistical mechanical software for the study of multi-component crystalline solids
Other
105 stars 70 forks source link

Regarding Energy trends in canonical MC #368

Open darjaved opened 2 months ago

darjaved commented 2 months ago

Below is the output of the MC at different temperatures starting from 1200 K til 0 K. Is it the formation enrgy per unit cell? My clex formation energy for the ground state is around -0.-0.267 ev/unitcell and here the MC predicts the formationn energy of around -0.315 eV/unitcell so the MC should give me the clex_hull_dist as (-0.315+0.267 = 0.04 ev/unitcell around. While taking measurements in MC does in take measurements per unit cell i.e. the same cell we have in the casm project i.e. the primitive cell?
clex_hull_dist at 0 K is 20 meV/atom as i have 2 atoms in the cell. is this value large or ok?

Also when i increase the supercell size, the formation per unit cell decreases especially for the left size of convex hull at some compositions.

"<clex_hull_dist(casm_learn_input,comp)>" : [ 0.001633153948, -0.004065673728, -0.010922273895, -0.018230587406, -0.028260098399, -0.033675103797, -0.039560439515, -0.043592750323, -0.045632750804, -0.046741408207, -0.047406238932, -0.047415785869, -0.047415785869 ], "" : [ -0.265846084677, -0.271544912353, -0.278401512520, -0.285709826031, -0.295739337024, -0.301154342422, -0.307039678140, -0.311071988948, -0.313111989429, -0.314220646832, -0.314885477557, -0.314895024494, -0.314895024494 ], "" : [ 0.000000000000, 0.000000000000, 0.000000000000, 0.000000000000, 0.000000000000, 0.000000000000, 0.000000000000, 0.000000000000, 0.000000000000, 0.000000000000, 0.000000000000, 0.000000000000, 0.000000000000 ], "" : [ -0.265846084677, -0.271544912353, -0.278401512520, -0.285709826031, -0.295739337024, -0.301154342422, -0.307039678140, -0.311071988948, -0.313111989429, -0.314220646832, -0.314885477557, -0.314895024494, -0.314895024494 ],

xivh commented 2 months ago

Formation energy is per primitive unit cell. All cluster expansion related things should be per primitive cell. Have you selected a supercell which can fit your ground state? You shouldn't be getting a formation energy which breaks your clex convex hull during monte carlo.

Do you have vacancies? I am worried that your formation energy per atom is 0 but the formation energy is not 0.

darjaved commented 1 month ago

I don't have vacancies, i wonder why the formation energy per atom is 0? Any suggestions? i think may there is no option to get the formation energy per atom in MC simulations. because when i do the casm format --monte i get these as the possible options as measurable quantity.

Possible options for "quantity" are: "comp": composition, relative the composition axes "comp_n": composition, number of atoms per unit cell "site_frac": composition, normalized per basis site "atom_frac": composition, normalized per total number of atoms "formation_energy": formation energy (per unit cell) "potential_energy": potential energy (per unit cell), (= formation_energy - sum_i(mu_i*comp_i)) "non_zero_eci_correlations": correlations (per unit cell) which have non-zero eci values. "<anything else>": is interpreted as a 'casm query' query

And while having supercell of higher size, there is decrease in formation energy, so can we say there is some ordering taking place?