Using the CE (ranges and size of structures)

Dear Axel,

I have got 2 questions:

1.) Is it accurate to take a CE which is not finished in the sense that there are new ground states in larger cells predicted and use it for the MC code? Example:
The internal database of structures extends at least up to 40 atoms/unit cell, see predstr.out
Among structures of known energy, true and predicted ground states agree
New ground states with at most 44 atoms/unit cell predicted , see predstr.out
Concentration range used for ground state checking: [0.25,1].
Crossvalidation score: 0.0193526

Could I just proceed with the MC code with a parametrization originating from 40 atoms instead of 44 with sufficient accuracy? Or is the predicitve error then unknown?

2.) How accurate do the concentration ranges have to be specified (-c0 -c1) with respect to the experimental range of solid solution? If experiment were to say [0;0.17] would it be ok to set -c0=0 and -c1=0.2 for the ground state fitting?

Thank you very much.

Siau

It is fine to try even if there are predicted ground states . The Monte Carlo will show you if new ground states appear in the concentration range you focus on. If no unexpected gs appear you’re ok.

The specified concentration range can be bigger than necessary - it will just take longer to get a good CE. Making it too small may give incorrect answers .