Hi
I have been running maps for a while and when I do mapsrep I get the plot attached for the fitted energies. Maps.log shows now new structures with 72 atoms can be created. I am wondering why some of the predicted ground states lie below the convex hull line.
[attachment deleted]
[attachment deleted]
[attachment deleted]
It seems that the code should be generating more structures to check those ground states. Is the last output (on std out) ""Finding best structure…"? If so it’s normal behavior.
Now, if feel that the process is too slow, you can try removing
-DSLOWENUMALGO
in the makefile (this is now the default in the most recent version). And then put some -O3 to optimize. Re-compile.