C. Cahillane

I have managed to use ER7 data produce preliminary carpet plots of frequency vs. strain magnitude and phase based on the Uncertainty Estimation paper T1400586.
The code that generates these plots may be found here: /ligo/svncommon/CalSVN/aligocalibration/trunk/Runs/S7/Common/MatlabTools/strainUncertainty.m

Darkhan recently did a similar study to this looking at error in magnitude and phase of Delta_L_ext when kappa_C, kappa_A, and f_c changed.  
My study currently looks at the error in magnitude and phase of strain when the magnitude and phase of kappa_tst and kappa_pu vary, as well as kappa_C and f_c.  (Recall that in general kappa_tst and kappa_pu can be complex.)

Right now I believe there is a serious error in the code, because the plot of the optical gain (plot 5) and the plot of the cavity pole (plot 6) show there is absolutely no error in strain even if these values differ greatly from the expected value.  The cavity pole varies by up to +- 100 Hz and does not vary by more than 1%.  The result is robust: I have calculated the strain using two independent methods and still I get these odd results.  These are my results right now, and this is why I call these plots preliminary.

I do believe the magnitude kappa_tst (plot 1), phase kappa_tst (plot 2), magnitude kappa_pu (plot 3), and phase kappa_pu (plot 4) plots look sensible.  Any phase in kappa_tst is generally intolerable for high frequency phase information, while phase in kappa_pu yields high error in low frequency phase information.  Since we do not expect any phase component at all in kappa_tst and kappa_pu, this makes sense.
Also, the magnitude of kappa_tst and kappa_pu must be tracked carefully at high and low frequency respectively.  10% errors in these magnitudes are enough to give more than 9% errors in strain magnitude.

Note that these results use ER7 data (GPS_start = 1116990382).  I'll soon be able to get ER8 data when it is all available (go calibration week!)