t2fit and fitting glitch parameters using data not spanning glitch epoch

[Ryan Shannon]

One strategy for making first estimates of glitch parameters is to fit the TOAs immediately following the glitch, zooming in a region after the glitch in plk. In a recent update to tempo2 this functionality has been removed. (See lines 1240 and beyond of t2fit.C). Is this a design feature?

[Michael Keith]

This was added deliberately to prevent fitting for glitches that are not within the start-finish range. I found that with par files with lots of glitches it is quite tedious to keep setting and un-setting the glitch parameters depending on the “zoom” range, and it is very easy to accidentally do a really broken fit and ruin your parameters.

My strategy for solving glitches is to solve for the pulse-numbers before and after the glitch and then just fit for the glitch parameters over the whole region, so I didn’t think of the case you described.

I guess you are leaving F0/F1 off and fitting for GLF0/GLF1 instead? Maybe we can change it so that if F0/F1 are not being fit then it will allow fitting for glitches preceding the data?

Let me know what you think and if you have any suggestion for how this feature should operate.