Post-processing of Driven Swept Sine Transfer Functions?
I'm interested in resurrection the discussion started a few years ago on the CSWG mailing list on how to process swept-sine excitations from the past (see thread archive here):
But today we had a new idea that we wonder if it's possible:
- Create a swept-sine template that drives an excitation into the system (say, PCAL excitation into the IFO), and grabs a few response channels that are relavent and accessible via online NDS (say, DARM_ERR and -- even though the DTT template uses it implicitly -- also store the PCAL EXC for completeness. Maybe even make it an "A" channel.).
- Save the template.
- Over the next ~10-30 seconds, the GDS-CALIB_STRAIN channel is produced by gstlal-calibration on the DMT machines, and then "written to frames." (I don't even know what I mean when I say that, so don't be mad I've used the wrong terminology; I'm just parroting what I've heard happens)
- 5 minutes later, re-open the template, switch to the session "user nds" mode, where it's now pointing to NDS2, which has all of those original "online NDS" channels, PCAL EXC, DARM ERR, etc. but now critically also has GDS-CALIB_STRAIN (or any of the other calibrated / cleaned data products from the low-latency pipeline infrastructure).
Can I now, 5 minutes later (a) keep my original PCAL_EXC, DARM_ERR but now add in GDS-CALIB_STRAIN, then (b) un-check the box to actually drive the excitation, and (c) run the template and it'll then use all of the original measurement settings to "re-drive the swept sine" (even though it's not actually driving), and therefore reprocess the swept-sine for me?
Please presume that all of the requirements that @lee.mcculler defined about storing the excitation channel in the frame is met, so don't get stuck on that. The template should obviously have all the timing information about when it was driving what frequency, so I would hope those requirements are met as well with this "hack."
Can DTT do this? If not, is it straight-foward to make it be able to do this?
I figure this should be testable on a test stand somewhere, and doesn't need the IFO. BUT, maybe it does need to be the IFO so that you can replicate the whole "a new channel is available that wasn't there before" situation.