|Date:||Wed, 20 Dec 2017 14:24:18 +0800 (GMT+08:00)|
Thanks for your time and I am writing to enquire about some problems in my lammps program. I’m a student and I’m studying the nano-scratch characteristics of gallium arsenide (GaAs) with a diamond tool.
These days I was writing a program about scratching on a block gallium arsenide workpiece with a spherical diamond tool, and I used the potential file GaAs.bop.table . The problem was that the program faulted in the process and had errors like “rank XX in job XX computer_name caused collective abort of all ranks exit status of rank XX: killed by signal XX ”( XX means different numbers). The size of my workpiece was 12nm*5nm*6nm, and the best result we got was the diamond tool moved along the x axis and reached about half of the workpiece, and then it stopped with this error. However, it had a good phenomenon of interaction between the tool and workpiece. If we made a bigger model, it would error earlier when the tool was just touching the workpiece. When we made a two times bigger model, it would error at the beginning without running. By the way, the relaxation part of the workpiece was very stable and didn’t have errors, but the calculation was slow for such a small model.
I am not sure if it was the lack of computing resource, and I also wonder the suitable situation (like the size of the workpiece) when using the GaAs.bop.table. Could we used it in big models which have great extrusion phenomenon? The computer has been used to compute models with millions atoms in other projects, but this time it didn’t work even though there were only less than one hundred thousand atoms. Last mouth I have tried lots of ways to avoid the problems but failed. Could you, please, give me some advice about this problem or some points which you think are helpful when using bond-order potential of GaAs? I’d be very grateful if you could help me to solve this!
Best wishes, and happy holiday!