Up until now we have assumed that the antenna gains vary smoothly with time, and so it is appropriate to interpolate between them. However this is not always the case. If there is a major instrumental glitch between calibrator scans, then it may be more appropriate to attribute all the antenna gain difference to this glitch. In this case, interpolating the gain solutions will probably give poorer results than assuming constant gains before and after the glitch.
The task gpbreak can be used to insert a `break-point' into the antenna gains table, which indicates to the software which applies the calibration, that interpolation across this point should not be performed. Rather the gain immediately before and after the break-point should be the gain from the previous or next calibrator observation.
Inserting a break-point modifies the antenna gains table, and so it is only meaningfully performed after the antenna gains table has been determined. To insert a break-point, you will need to know the exact time of the glitch. You may well know this from the observing log, or you may be able to deduce it from the program source data if you are observing a strong source.
Task gpbreak takes as its parameters the name of the visibility dataset ( vis parameter), the times at which to insert break-points ( break parameter), and the antennas and feeds for which to insert the break-points (parameters ants and feeds -- the default is to insert break-points for all antennas and feeds). The break parameter gives the time in normal MIRIAD format -- either hh : mm : ss or yymmmdd : hh: mm : ss. The second form (giving year, month and day) will generally not be required for a single day's observing. Several times can be given, separated by commas. For example:
Unfortunately break-points do not survive recalibration. If you redetermine the antenna gains ( e.g. re-run gpcal or the like), then you will overwrite the old antenna gains table, and will need to re-insert the break-points.