Thank you for fixing this issue.

As hw_rho_total component of cdft_group_type derived type is used exclusively within hirshfeld_constraint_low() subroutine, would it be better to simplify the code a little further by removing hw_rho_total from cdft_group_type and declaring it as a local variable?

It should get rid of the magic index in the expression cdft_control%group(1)%hw_rho_total at least.

Thanks for the helpful suggestion Sergey.

I've now moved hw_rho_total from cdft_group_type to cdft_control_type, as the promolecular density is the same for all CDFT groups so it should not be stored per CDFT group.

hw_rho_total is also used in file qs_cdft_utils.F so keeping it in the CDFT control structure seems appropriate. If you would prefer it defined as a local variable and then passed to qs_cdft_utils.F I'm happy to do so