I am running MD simulations using the ELC correction to the P3M method and am trying to decide on an appropriate error value. I was wondering if the error specified in the tuning is the error on the force between any two particles or the error on the total electrostatic force for any one particle. I assume it is the latter and thus as long as it is below the thermal noise in the system things should be okay. My plan was to use 0.01 with a kT of 1 so that it would be 100 times less than the thermal noise. I think I may have already asked this question and if I have I apologize.