Question about lj/cut/coul/long/soft

Dear all,

I want to perform a free energy of hydration of a simple organic molecule on water using the FEP method. To do so, I read the fep/examples for the hydration of methane and try to do my own input file to solve my problem.

Unfortunately, the simulation got a nan value for the potential energy when I used the lj/cut/coul/long/soft. I check the topology and is fine, in fact if I run it with lj/cut/coul/long instead of lj/cut/coul/long/soft everything is ok.

As it was a “complex” system, with two kind of molecules, I test this pair_style for two other inputs: 1) a organic fluid and 2) water. For the 1) case the pair_style lj/cut/coul/long/soft went ok…but for the second case I got the nan value for potential energy.

I tried two water models and I got the nan results on both of them. (SPC -with shake- and the SPCFw -without shake)

I understood that if I use the lj/cut/coul/long/soft with a lambda parameter of 1, it should give me the same that the lj/cut/coul/long pair_style…but it did not work on my water molecules tests.

In order to continue with the FEP calculation I need to solve the pair_style problem…I would appreciate any suggestion on this matter.

I copy the input file:

#---------Defines system---------------------------------------------------#
units real
atom_style full
boundary p p p

pair_style lj/cut/coul/long/soft 2 0.5 9.0 9.0 9.0
kspace_style pppm 1.0e-4

bond_style harmonic
angle_style harmonic
read_data final.topo

pair_coeff 1 1 0.1554 3.166 1.0
pair_coeff 2 2 0.0 0.0 1.0
pair_coeff 1 2 0.0 0.0 1.0

group water type 1 2

neighbor 1.0 bin
neigh_modify delay 5

Dear all,

I want to perform a free energy of hydration of a simple organic molecule
on water using the FEP method. To do so, I read the fep/examples for the
hydration of methane and try to do my own input file to solve my problem.

Unfortunately, the simulation got a nan value for the potential energy
when I used the lj/cut/coul/long/soft. I check the topology and is fine,
in fact if I run it with lj/cut/coul/long instead of
lj/cut/coul/long/soft everything is ok.

As it was a "complex" system, with two kind of molecules, I test this
pair_style for two other inputs: 1) a organic fluid and 2) water. For the
1) case the pair_style lj/cut/coul/long/soft went ok...but for the second
case I got the nan value for potential energy.

I tried two water models and I got the nan results on both of them. (SPC
-with shake- and the SPCFw -without shake)

I understood that if I use the lj/cut/coul/long/soft with a lambda
parameter of 1, it should give me the same that the lj/cut/coul/long
pair_style...but it did not work on my water molecules tests.

In order to continue with the FEP calculation I need to solve the
pair_style problem...I would appreciate any suggestion on this matter.

​the problem is that sigma enters the computation of the soft potential
forces in the denominator, and thus you must not set it to zero. there is
no need to do this (in fact, there is not much sense to it in the first
place), since setting epsilon to zero will already sufficiently turn off
the LJ interactions between hydrogen atoms and everything else in SPC.​

​axel.​

Sorry, I didn’t realize that. Thank you!!! Now it is working fine.