[lammps-users] intra-molecular pair interaction

Dear all,

I am trying to simulate a system in 2D with multiple polymer chains using 
harmonic bond potential and lj potential. I want to switch on the LJ potential between inter-molecular as well as intra-molecular atoms to prevent overlap (here I want to mention that, I am calling a polymer as a molecule). However, when I see the log file, from the value of E_pair it seems 
it does not calculate the intra-molecular LJ-interaction (As it does not match with my calculated value of E_pair which should be quite high for the bonded pair of monomers with bond-length 0.5\sigma and also this leads to highly 
overlapped intra-molecular atoms when simulated at high density). I am 
attaching the input script and the generated log-file for your reference. 
Any idea why the LJ interaction is not acting between 
intra-molecular atoms? 
Best regards,

log.lammps (6.4 KB)

in.lmp (1.34 KB)

Dear Axel,

Thank you for your prompt reply. However, I don’t get your point correctly. The cutoff is 2^{1/6}\sigma and the bondlength (distance between two beads in a polymer chain) = \sigma/2. So, the pair interaction energy for the bonded beads should be very high (~ 2^12). Does it exclude the pair interaction between bonded beads?

I actually see the “folding back” and overlap between beads of same polymer, when I simulate at high density. I have 61 beads in a single polymer.

Best regards,

for further information on exclusions, please check with your favorite text book on MD simulations and molecular force fields first and then carefully study the documentation for the special_bonds command.

Hi Andrew,

Thank you very much for your email.
In my simulations, \sigma = 2 times the bond length (the distance between consecutively bonded atoms in the chain). I am using truncated and shifted Lennard-Jones interaction and the cutoff is 2^(1/6) \sigma. \epsilon= k_BT as I use LJ units. I still see the overlaps at high density though the E_pair is not very high. Please see the values below, taken from the logfile:

Step Temp KinEng PotEng E_pair E_bond E_angle Press Fmax Volume v_y0 v_v0

100700000 0.93642775 0.93641766 5.4742082 0.86620974 0.59091579 4.0170827 19.73895 475.1082 75185.018 75.022325 1.6724001e-05

I am attaching a zoomed part of the corresponding snapshot where the large intra-filament overlaps are visible. The filaments fold back and overlap. The distance between overlapped atoms is even < the bond length. At this distance, E_pair should have been quite large and prevented the overlap. This implies that intra-filament pair interaction is not working.

What do you think? Any suggestions will be really useful.

Best regards,


could it be that you are ignoring the fact, that with reduced units LAMMPS defaults to thermo_modify norm yes (instead of no) and thus you get the energy per atom not for the entire system?


I have nothing to add to what I already wrote in my previous responses. LAMMPS will compute what you ask it to compute and that is described in the documentation.
If you believe you have found a bug, then you have to provide a more convincing argument or demonstration because your reasoning is not convincing this far.
Please also note that you are talking about functionality that is in LAMMPS for a very long time, so it is rather unlikely that there is a bug that went unnoticed all the time.


Hi Axel,

Yes, it is energy/atom and not the energy of the entire system. But for a configuration with such huge overlap, energy/atom should be huge. For an example, please see the attached configuration (zoomed part) which I generated with no-overlap at all. Below, I am copying the energy from the log-file:

Step Temp KinEng PotEng E_pair E_bond E_angle Press Fmax Volume v_phi
0 0.93829913 0.93828903 0.87707145 0.87707145 1.5167787e-28 1.4561076e-30 14.517883 0.099803979 75185.018 0.79769082

All parameters are the same as before. Please note the E_pair which is almost the same (even slightly less) as the overlapped configuration. This seems to me quite strange and it only implies that all the pair-interactions (not only 1-2 1-3 1-4 pairs) are excluded for intra-filament atoms.

Please let me know your comments.

Best regards,