I'm trying to use ASE and exciting to calculate gamma surfaces, which involves calculating the energy of a number of different atomic configurations. In some of these configurations, the default muffin tin sizes lead to overlaps. The exciting documentation implies that I can simply resize the muffin tins manually or set autormt = true. However, when I repeat the whole calculation with autormt = true, the energies I get are much larger. More importantly for my purposes, the differences in energy between the different configurations are much larger (and approximately 150 times literature values).
Can anyone point me to any literature explaining why the size of muffin tins selected should influence the energy calculated? What is the correct way to deal with this issue?
Thanks in advance,