I just completed a simulation on a long chain alkane and from the thermo output I noticed that the outputted E_pair value does not equal the sum of E_coul, E_vdwl and E_long.
No, it has (obviously) nothing to do with it, but it is a warning that should be taken seriously.
Any lattice sum will diverge for non-neutral cells and such an object would explode instantly in reality. The Ewald/PPPM formulation thus ignores divergent terms from the net charge which is physically equivalent to embedding the system into a distributed counter charge. While this is mathematically “clean” it can be problematic. The issue is to understand where the net charge originates from. Usually this will not happen unless you have either incorrect computed partial charges on atoms or have incorrectly rounded/truncated them. For an alkane chain you should have a net zero charge. In fact, each repeat unit should be neutral. So it is very likely that there is an error or an inaccuracy in your system preparation.
Thank you, my post processing script was causing the issue with the pairwise interaction sum.
Regarding the warning, would a system of, for example, phenol have the same problem? Would the same styles and settings be applicable?
(I built the system with moltemplate using the OPLSAA ff)
If you have neutral molecules then the system should be neutral. If moltemplate does not produce a neutral system (and I can only guess it is because of rounding/truncation of assigned partial charges) then this is a bug in the software or the parameters/inputs used.
MD follows the GI-GO principle and if your input is already tainted by rounding errors, then this doesn’t bode well for the reliability of any results, especially if those may be looking at subtle differences.