LAMMPS WWW Site - LAMMPS Documentation - LAMMPS Mailing List Archives
Re: [lammps-users] NaN when using Qeq package
[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: [lammps-users] NaN when using Qeq package


From: Ray Shan <rshan@...1795...>
Date: Tue, 29 Aug 2017 22:27:15 +0000

Your charges did not converge so atoms probably ended up with spurious charge values, causing Coulomb potential to misbehave.

 

1.      Start with more meaningful initial charges, i.e., charges that are close to equilibrium charges or just some fractional values instead of zeros would help a lot.

2.      Check your QEq parameters in “bi2o3.parm” as they might not make sense.

 

Best regards,

Ray

 

From: Xiaoyu Wang <xwang181@...5818...>
Date: Tuesday, August 29, 2017 at 3:14 PM
To: "lammps-users@lists.sourceforge.net" <lammps-users@lists.sourceforge.net>
Subject: [lammps-users] NaN when using Qeq package

 

Dear all,

 

I was trying to develop charges on a crystal by using Qeq. I found the ionization potential and electron affinity parameters to get the Mulliken electronegativity, which is used by Qeq. But it gave NaN on charges and those energies. The crystal structure is from the standard database. It seems there is something wrong blowing up the system. Can anyone know how to deal with it? I was using the same method successfully once before, but I have no idea why this time it is not working. 

 

Here is the input.

 

units metal

dimension 3

#boundary p p p

atom_style charge

read_data bi2o3_qeq.data

 

pair_style buck/coul/long 12.0

kspace_style ewald 1.0e-6

 

pair_coeff 1 1 6452.86 0.5089 0.0000

pair_coeff 1 2 6048.95 0.5162 0.0000

pair_coeff 2 2 5670.33 0.5235 8.1163

 

neighbor 2.0 bin

neigh_modify delay 0 every 1 check yes

 

group           type1 type 1

compute         charge1 type1 property/atom q

compute         q1 type1 reduce ave c_charge1

group           type2 type 2

compute         charge2 type2 property/atom q

compute         q2 type2 reduce ave c_charge2

variable        qtot equal count(type1)*c_q1+count(type2)*c_q2

 

thermo_style    custom step pe c_q1 c_q2 v_qtot spcpu

thermo          10

 

timestep        0.0001

 

velocity        all create 300.0 1281937 

fix 1 all nve

fix 2 all qeq/dynamic 1 10 1.0e-3 100 bi2o3.parm

 

run 100

 

write_data bi2o3_qeq_o.data

 

 

Here is the error:

WARNING: Charges did not converge at step 0: nan (../fix_qeq_dynamic.cpp:165)

ERROR on proc 0: Non-numeric atom coords - simulation unstable (../domain.cpp:521)

 

Best wishes

 

Shaw