ERROR: Too many MSM grid levels

For simulating of nitrogen and oxygen molecules I use TraPPE-Small model. Based on this model atoms of these molecules have electric charges. Also my simulation box has fix boundary condition at z direction. So, with respect to lammps manual I used coul/msm for coulombic term. But when I run my simulation I face with this error: ERROR: Too many MSM grid levels . I don’t know if this error relates to initial position of atoms or something else. I searched over net but I couldn’t find any solution.
You can also read the force field detail below
I would appreciate anyone help me to solve this problem

Vahid Ahmadi

variable        epsilon_O equal 0.097373008691
variable        sigma_O   equal 3.02000

variable        epsilon_N equal 0.07153934924399673
variable        sigma_N   equal 3.31000

variable        epsilon_NO equal (${epsilon_O}*${epsilon_N})^0.5
variable        sigma_NO   equal (${sigma_O}+${sigma_N})/2

pair_style	    lj/cut/coul/msm 14.0 14.0     
pair_modify     tail yes			     		 
bond_style      harmonic
angle_style     harmonic                
kspace_style	msm 1.0e-4
kspace_modify   cutoff/adjust no

pair_coeff      * 2 0.0000 0.0000                
pair_coeff      * 4 0.0000 0.0000  
pair_coeff		1 1 ${epsilon_O}  ${sigma_O}  
pair_coeff		3 3 ${epsilon_N}  ${sigma_N}	 
pair_coeff		1 3 ${epsilon_NO} ${sigma_NO}	 

bond_style	    harmonic
angle_style	    harmonic
dihedral_style	none
improper_style	none

bond_coeff	    1 5000.00 0.60500
bond_coeff	    2 5000.00 0.55000

angle_coeff	    1 500.0 180.00
angle_coeff	    2 500.0 180.00

This is likely due to the combination of the requested MSM accuracy and the kspace_modify request to not adjust the cutoffs. There is a built-in limit of 10 MSM levels.

Using MSM correctly and accurately is a rather tricky business and it is not recommend for people with limited experience. The implementation is rather old, too, and is missing a lot of what is state-of-the art in multi-level summation these days which would lead to improved performance and accuracy.

Since you say that you have a system that is non-periodic in z direction, you may want to consider using a suitable PPPM or Ewald based kspace style and then kspace_modify slab. That feature has its own challenges, but there is much more residual knowledge floating around about how to use that correctly.

I will try kspace_modify slab.
Thanks for your reply Axel. :heart: