I found that for dftd4_load_rational_damping function, if the third argument (mdb in C, atm in Fortran) is false, s9 will still be set to 1.0. Could you help check whether this is a bug?

Traceback

1. C-API Layer (include/dftd4.h):
   The function signature is dftd4_load_rational_damping(..., bool mdb).

2. Fortran API Layer (src/dftd4/api.f90):
   The C function calls load_rational_damping_api.

   function load_rational_damping_api(verror, charptr, atm)
      ...
      logical(c_bool), value, intent(in) :: atm
      real(wp), allocatable :: s9
      ...
      if (atm) s9 = 1.0_wp
      call get_rational_damping(method, tmp, s9)

   If atm (the third argument) is false, the variable s9 remains unallocated.

3. Parameter Retrieval (src/dftd4/param.f90):
   The get_rational_damping interface calls get_rational_damping_name, which calls get_rational_damping_id.
   Since s9 was unallocated in the caller, it is treated as not present in the optional arguments chain.

   In get_rational_damping_id:

   mbd = .true.
   if (present(s9)) mbd = abs(s9) > epsilon(s9)

   Because s9 is not present, mbd defaults to .true.. This triggers the call to get_d4eeq_bjatm_parameter.

4. Default Value Assignment:
   Inside get_d4eeq_bjatm_parameter, the helper function dftd_param is called without s9.

   pure function dftd_param(..., s9)
      ...
      s9_ = 1.0_wp
      if (present(s9)) s9_ = s9

   Since s9 is missing, it defaults to 1.0.