New features

The major new features of the present version (9.2) are:

• The program is now no longer restricted to linear molecules and should run for any molecule for which the required input data is available (ie. collision coefficients and other molecular parameters).

• A new, hopefully improved, user interface that makes it easier to switch between different models.

• Several of the parameters previously contained in the input files are now calculated in the program internally - this hopefully reduces the possibilities for making mistakes in the input files.

• The convergence test^4.2 has been made more rigorous to now check all levels in all shells (rather than just all the levels in one shell). The criteria for convergence have, however, been relaxed in the outer parts of the cloud where the molecular density can be very low (which is known to cause convergence problems - these very low densities do not contribute significantly to the line strengths or shapes and so can be ignored). In an attempt to ignore the minor fluctuations in some models that occur when most shells and levels have converged, convergence is also achieved if all but 3 levels converge 5 times (not necessarily consecutively). This is not a particularly satisfactory way of dealing with the fluctuations but as they appear to have no observable impact on the output it seems to be a valid approximation.

• A new output style has been added which now uses PGPLOT to produce output on the screen in a GKS window as soon as the output has been calculated. After all calculations are completed it is possible to produce customised postscript or encapsulated postscript output or, alternatively, an A4 sheet of output can be produced which shows the predicted line shape and intensity as well as the optical depth and also includes important model parameters.

• The program now presents an option to store the data from a run as soon as it is finished. These data can then be retrieved later using the program and viewed or printed. Options to modify past models and re-run them as well as options to delete unwanted runs to save disk space are included.

• Due to the significant amount of time needed to run some of the more complex models an option has now been built in to run a series of models consecutively with variations in some model parameters. This enables part of the parameter space for a particular model to be tested overnight.

The majority of the material describing the working of the STEN program can either be found in Matthews[20] or in the sections describing the ASTRA program (eg. section 4.7). Only modifications which are unique to this version of the program will be described below. It may help to refer back to the flow chart in figure 3.3 to see where each section fits in.