I am doing simulation on bicrystal crack propagation with uniaxial tensile load. But, same input file when run again and again gives a slightly different stress-strain curve and also crack path. I am using parallel computing facility provided by my university.
It will give you advices on how to formate your question better and investigate your problem in more depth. Here are some comments I can make to help you format your question so that you get useful answers:
What do you mean by “same input file”? Some command can use random values which can result in differences in some outputs. It is also important to provide the LAMMPS version and compare with the previous version used to produce the initial results.
“same input file when run again and again gives a slightly different stress-strain” this is vague. Do you run the same input file and get a different output? How different? “slightly” is not precise as some statistical noise is to be expected. How different?