Dear Axel,
Thanks for the initial clarifications. I am providing a bit more details of the simulated system for further reference:
The IL is a 4-site model where the cation is represented by rigid 3 (type 1, 2 and 3) beads and anion is a single bead (type 4).
Total no of IL ion-pairs is 125. A single water molecule (type 5 for O and type 6 for H) is embeded in the IL box. The water is treated with rigid SPC/E model . The trajectory was generated with pair_style lj/cut/coul/long 16.0. The aim is to decompose the total electrostatic force/energy arising due to all IL atoms on the water O into cationic and anionic contributions
(Fx=Fx_cat + Fx_anion, Fy=Fy_cat + Fy_anion and Fz=Fz_cat + Fz_anion and E=E_cat+E_anion). The initial test rerun is performed on simpler pair_style coul/cut 16.0 to check the consistency of the decompositions. So, I created groups for cation (type 1, 2 and 3) , anion (type 4) and water O (type 5) as follows:
group IP type 1 2 3 4 # grouping of IL atom types
group cat type 1 2 3 # grouping of IL cation atom types
group anion type 4 # grouping of IL anion atom types
group OW type 5 #grouping of H2O O
Then the following computes are invoked:
compute 1 OW group/group IP # pairwise interaction among all IL atoms with O
compute 2 OW property/atom type fx fy fz # force arising due to pairwise interactions on O
compute 3 OW pe/atom pair #PE arising due to pairwise interactions on O
compute 4 OW reduce sum c_3 # sum to dump with thermo command (will not have any effect due single O atom in the system)
compute 5 OW group/group cat # pairwise interaction among all IL cation atoms with O
compute 6 OW group/group anion # pairwise interaction among all IL anion atoms with O
I expect that
c_1=c_5+c_6 (PE, which I found consistent)
c_1[]=c_5[]+c_6[] (Force components, which I found consistent)
c_1=c_3 (Not the case)
c_1[]=c_2[] (Not the case)
c_2[]=c_5[]+c_6[] (Not the case).
So, I am confused which is the protocol I should take to solve the problem.
Thanks
Sourav
Penn State