This paper describes Fortran program dParFit, which performs least-squares fits of diatomic molecule spectroscopic data involving one or more electronic states and one or more isotopologues, to parameterized expressions for the level energies. The data may consist of any combination of microwave, infrared or electronic vibrotational bands, fluorescence series or binding energies (from photo-association spectroscopy). The level energies for each electronic state may be described by one of: (i) band constants {Gv,Bv,Dv,…} for each vibrational level, (ii) generalized Dunham expansions, (iii) pure near-dissociation expansions (NDEs), (iv) mixed Dunham/NDE expressions, or (v) individual term values for each distinct level of each isotopologue. Different representations may be used for different electronic states and/or for different types of constants in a given fit (e.g., Gv and Bv may be represented one way and centrifugal distortion constants another). The effect of Λ-doubling or 2Σ splittings may be represented either by band constants (qvB or γvB, qvD or γvD, etc.) for each vibrational level of each isotopologue, or by using power series expansions in (v+12) to represent those constants. Fits to Dunham or NDE expressions automatically incorporate normal first-order semiclassical mass scaling to allow combined analyses of multi-isotopologue data. In addition, dParFit may fit to determine atomic-mass-dependent terms required to account for breakdown of the Born–Oppenheimer and first-order semiclassical approximations. In any of these types of fits, one or more subsets of these parameters for one or more of the electronic states may be held fixed, while a limited parameter set is varied. The program can also use a set of read-in constants to make predictions and calculate deviations [ycalc−yobs] for any chosen input data set, or to generate predictions of arbitrary data sets.
Read full abstract