I am using meam pair-style
Following the manual is it possible to use pair_modify mix arithmetic.
I wrote the following commands
pair_coeff * * CuTiWZnZr.meam Cu Ti NULL Cu Ti
pair_modify mix arithmetic
which did not generate error or warning message.
Does it mean that Cu-Ti potential was created using arithmetic means for all length-like parameter and geometric means for all energy-like terms ? As I understood from reading pair_modify explanation in the manual.
thanks a lot in advance
PS I know this is a crude approximation I try this for some other element tht do not have cross interaction for starting new parametrization.
You could test this with a minimal input, outputting the energy and forces both with and without
pair_modify. But I just took a look at the source files instead; pair_meam.cpp has no calls to
mix_distance, which accounts for the settings in
pair_modify mix. For example, you can see in pair_lj_cut.cpp in the function init_one there are calls to those functions to initiate mixing.
The pair_modify mix option only applies to pair-wise additive potentials and will never result in an error, since it only changes the algorithm used, in case mixing is supported.
The situation with manybody potentials, like MEAM, is very different. If there is mixing, it has to be intrinsically implemented in the pair style or the data for such mixed terms must be present in the potential file, since you have only a mapping between atom types and element entries in the potential file.
In MEAM there are general per-element parameters in the “MEAM library” file, i.e. the first potential file. Those will also create mixed parameters for mixed element pairs of atoms. However, those will often result in rather unsatisfactory or sometimes outright incorrect results. This is where the second potential file comes into play. This usually has settings that override the general mixed terms with more specific settings that are tuned for specific element combinations. Thus MEAM offers a lot of flexibility, but you need to look up optimized parameters.
A good place is the NIST interatomic potential data base. Check out this one, for example: Interatomic Potentials Repository
Dear Axel and Michael
Thanks a lot for the help. The ultimate system I wish to study has not parameter files in the NIST repository. So I should try to parametrize by my own a new force fields. Using Gulp or MEAMfit I guess.
Thanks a lot again.