This is a modified version of the program STARK.F written by H.M.Pickett. Like its predecessor, SZK is a postprocessing program working on output from the SPFIT/SPCAT package.  The program calculates the same quantities as STARK, but it takes its data from an .STR file produced by a prior run of SPCAT.  The .STR file contains in the second column the reduced transition dipole matrix element which is equal to the square root of the linestrength. This is the basis for evaluating the Golden&Wilson type Stark coefficients as defined in J.Chem.Phys. 16, 669 (1948). The .STR file is produced by setting STRFLAG in the .INT file to 1 (i.e. the tens digit in the first number in the top line has to be 1). The output from SZK is written to an .STK file.

        Modifications to Pickett's original are as stated at the top of the listing and have gone mainly into producing what is hopefully self-explanatory output. Only the coefficients for the energy levels are calculated so that those for the observed transitions have to be set up by hand. Note that with this version the coefficients should be calculated by setting only one of the three possible dipole moment components to unity - if several components contribute to the Stark shift then results from two or three such separate runs of SPCAT and SZK should be combined.

        The incentive for writing this program came from the necessity to deal with Stark shifts measured in FTMW work on quadrupolar molecules.   The available field strength is quite low so such shifts fall into the inconvenient intermediate field regime. The only robust solution is through matrix diagonalization, which has to be carried out for each combination of field and M_F (the quantisation used is J, K₊₁, K_-1 ,F, M_F).  QSTARK has been developed from Q2FIT and irreducible tensor matrix elements for quadrupolar coupling are from that program.  The matrix elements of H_E are from:

        H.P.Benz, A.Bauder, Hs.H.Gunthard, J.Mol.Spectrosc. 21, 156 (1966). 

• Calculation for linear, symmetric, and asymmetric rotors, with zero or one quadrupolar nuclei
• All of the observed Stark shifts can be included in one data set (if the field calibration is good enough) - for example for asymmetric tops without quadrupolar nuclei it is possible to fit second order and mixed order shifts simultaneously
• The fit can be made either directly to frequencies or to frequency differences
• It is possible to fit either the effective electrode separation (calibration) or the dipole moment components and, if desired, any of the remaining constants in the Hamiltonian - the latter allows enhanced determination of spectroscopic constants from Stark shift perturbations
• All types of ΔM transitions can be fitted: 0 and ±1
• Fit can be weighted according to estimated measurement errors
• It is possible to calculate and plot the behaviour of selected Stark components with the electric field. The program can produce simple diagnostic ASCII plots in the standard output file, as well as appropriate files for the gle program, and thus to obtain higher quality PDF etc. output.
• Experimental measurements can be plotted on top of predictions (these are placed in a simple two column file of voltages and frequencies during a fitting run, and this file is reused during a subsequent predictive run)
  
• It is possible to plot predicted Stark lobe behaviour as a linear or quadratic function of applied voltage or electric field
• Separating blank lines and comments can be embedded between the measured frequencies and will be echoed to the output if required, and the number of transitions declared in the data can also be automatically counted by the program

       The recommended paper for citing the the use of QSTARK is:

• Z.Kisiel, J.Kosarzewski, B.A.Pietrewicz, L.Pszczolkowski, Chem. Phys. Lett. 325, 523-530 (2000).

       Known bugs:

       The program calculates correct energies but is known to run into labelling problems when the off-diagonals in the H matrix become sufficiently large. Thus indices may be incorrectly assigned to the eigenvalues. Known instances of such behaviour are:

• first order Stark effects in symmetric tops
• highly perturbing states in asymmetric tops
• high field calculations when μ_c is non-zero and QSTARK switches to the complex H matrix formulation

       Extensive dump output, as controlled by the IDUMP parameter, allows checking of the internal workings in order to obtain a more detailed insight into such behaviour.