Modifying dump_custom syntax for VDAOS computing

Dear all,

I am wondering if I can modify the syntax applied used in the custom dump file. The purpose is for some vibrationnal density of states calculations : I need to extract the velocity per atom every N timestep, to compute the autocorrelation function and FT it, using an external C code with fftw.The syntax between the dump file and the readable type of file for the external code are a bit different (I need to write the elapsed time at each timestep, to erase the “ITEM : etc” at every timestep, and some time to convert some component of the velocity from int to float - eg 5 to 5.0).

For the example, this is my current dump_custom line :

variable f file tmp.times
variable s equal next(f)
dump veloc all custom 5 crystal_v.vel vx vy vz
dump_modify veloc sort id
dump_modify veloc every v_s

For the moment I am preparing the simulations, so in the first place I figured I could just use a small matlab routine. But I wondered if it was possible to do it directly from lammps? Without modifying the dump_custom c++ file for controlling directly the printings, because I work on a cluster.

I have another general question, concerning VDAOS simulations : as a first guess, I would use a fix NVT to obtain the veolcity field, and pay attention to the frequency at witch it is dumped. Could somebody share his experience?


Arthur France-Lanord

erratum : VDOS instead of VDAOS of course.

And I didn’t mentionned the systems I study : amorphous silicon, and amorphous/crystalline si superlattices. But first I would like to obtain realistic VDOS of cristalline silicon in order to validate the procedure.


Arthur France-Lanord

There are no options in dump modify that change the syntax
of a custom dump file. E.g. get rid of the ITEM tags.
I think that is best done in post-processing.

There are some provided Python scripts in tools/python
that convert dump files to other formats. If you know
Python it would be pretty simple to change one of
those scripts to convert it to another format. The
tools/python/pizza/ is what reads the custom
dump file. You could then write a few Python commands
to extract columns from each snapshot and spit
them out however you want.