[lammps-users] different results by changing the shape of model

Dear LAMMPS developers

I am Jing Guo from UK. I have been working on the MD simulation for high entropy alloys recently. I carried out the simulation of tensile process for NbMoTaW HEA. I built cylinder and rectangular models with similar size of Φ60×240Å and 60×60×240Å considering different orientations, the code for both of these two models is exactly the same. However, the results for these two models are different from each other. Please see attached file. In addition, there is no necking and fracture happened in the simulation even the strain value is 0.5. The lammps code has been listed below. Could you please tell me is there something wrong with my code? What caused the results different from each other after I changed the shape of model?



units metal

atom_style atomic

boundary p p p

neighbor 0.3 bin

#create geometry#

read_data NbMoTaW110.lmp

#define interatomic potentail#

pair_style snap

pair_coeff * * MoNbTaW.snapcoeff MoNbTaW.snapparam Ta Nb Mo W

#Output thermodynamics information#

thermo 1000

thermo_style custom step lx ly lz press pxx pyy pzz pe temp

compute 1 all stress/atom NULL

compute 2 all ke/atom

compute 3 all displace/atom


dump 1 all custom 500 tensile.dump id type x y z

#Relax structure#

velocity all create 300 12345 mom yes rot no

fix 1 all npt temp 300 300 1 iso 0 0 1 drag 1


timestep 0.001

run 20000

#Tensile process (Z direction)#

unfix 1

fix 1 all npt temp 300 300 1 x 0 0 1 y 0 0 1 drag 1

fix 2 all deform 1 z erate 0.01 units box remap x

variable Ltmp equal lz

variable Lini equal ${Ltmp}

variable strain equal (lz-v_Lini)/v_Lini

variable stress equal -pzz/10000

fix 3 all print 100 “{strain} {stress}” file strain_stress.txt screen no


run 50000

Best regards,


Screenshot 2021-09-16 110527.png

This is really a topic for a discussion that you should have with your adviser.
The first question you have to address is whether these simulations should produce the exact same results or not.
have you checked whether you get identical results for the same system when varying the random number generator seeds?
in other words, are your results really representative? or just some random selections of possible results?

Also, whether your input is suitable or not is often difficult to tell from just looking at it. in many cases the log file or screen output will contain warnings if something is not quite proper and LAMMPS can detect it. sometimes LAMMPS cannot detect issues because what is suitable in one case may be not acceptable in another. in simulations there is often no yes/no situation but a sliding scale when conscious choices need to be made. however, those choices require knowledge of the specific research topic and are not an issue of the simulation software itself but about how to conduct research in that area in general.

TLDR; this doesn’t seem to be something that is specific to LAMMPS but about your research in general.