How to calculate the Objective_function ?

Dear Dr. Axel,

It is very kind of you to share your excellent work from which I have learned a lot.

After carefully reading all the content of this forum, there are still several questions in my mind so far.

  1. What is the main difference between the SQS and MCSQS ?

    Should it be that the SQS uses a CE+DFT calculation with an exhaustive enumeration while the MCSQS only uses CE+MCMC without an exhaustive enumeration ?

  2. How to estimate the accuracy of the SQS ?

    Is there only comparison with the other results from experiment or calculations ?

  3. For a given structure and its corresponding cluster.out file, how to calculate the Objective_function ?
    Or in other word, is there any other way to calculate the Objective_function except for writing script codes by myself ?


That is all.

Any help regarding this would be highly appreciated.

  1. gensqs performs an exhaustive enumeration (slow but guaranteed to find the best) while mcsqs randomly samples (fast but no guarantee). The second option is usually what you want.

  2. You would typically check how the property you want converges as you generate larger and larger SQS (with more matched correlations).

  3. The objective function is explained in mcsqs -h
    at the end of the help file. Or look into the code src/mcsqs.c++