Consider the example lat.in system described in the mmaps -h output:
Armed with a suitable cluster expansion, perform a canonical simulation using memc2 (with an appropriate initial structure and conccons.in file) to generate a representative structure at T=1000 K with x=1.0 (i.e fully lithiated) and some desired y value.
Starting with the fully lithiated 1000 K structure, manually replace half the Li with Vac(ancies), and use this as the initial structure for a new canonical simlation at some relatively low temperature, e.g. 300 K.
Question: Is it possible to swap only Li and Vac, leaving Co and Al atoms unperturbed?
To freeze one sublattice, you can create a file called frozensite.in that contains a list of the sites you want to freeze (one number per site, indicating its line number, starting at 0). One caution: the sites are internally re-ordered for efficiency - to get the right order type: