J. Kissel

I've completed creating an updated parameter set for the matlab DARM loop model that  contains the updates from the fit of the 2016-11-12 calibration reference measurements (see LHO aLOG 31433 for original measurement set, and LHO aLOG 32989 for updated analysis). 

I've compared the model output against those measurements using
${CalSVN}/aligocalibration/trunk/Runs/ER10/H1/Scripts/DARMOLGTFs/compareDARMOLGTFs.m  (r4123, ls: r4123)
Results are attached. The agreement is excellent.

I also attached the sensing and actuation function fit parameter results, just for completeness -- including nlinfit results (and because LHO aLOG 32989 has some typos, only corrected in its comment, LHO aLOG 33090).

                                                       Former Value        MAP     (95% C.I.)  MAP       (95% C.I.)
                                                       2016-11-12          2016-11-12 (nlinfit)      2016-11-12 (MCMC)

 Optical Gain                K_C         [ct/m]        1.15e6              1.166e6 (0.0003e6)        1.164e6   (0.0002e6)
 Couple Cav. Pole Freq.      f_c         [Hz]          346.7               347.2   (5.8)             346.7     (2.7)
 Residual Sensing Delay      tau_C       [us]          2.3                 2.12    (5.5)             2.3       (2.3)
 SRC Detuning Spring Freq.   f_s         [Hz]          7.4                 7.38    (0.3)             7.38      (0.16)
 Inv. Spring Qual. Factor    1/Q_s       [ ]           0.05                0.046   (0.013)           0.046     (0.008)

 UIM/L1 Actuation Strength   K_UIM       [N/ct]        8.164e-8            n/a                       8.104e-8  (0.2%)
 PUM/L2 Actuation Strength   K_PUM       [N/ct]        6.841e-10           n/a                       6.773e-10 (0.04%)
 UIM/L3 Actuation Strength   K_TST       [N/ct]        4.389e-12           n/a                       4.346e-12 (0.05%)

Not only is the model in excellent agreement with the measurement, the nlinfit fit parameter values are in excellent agreement with the MCMC values with the sensing function. As with the updated / new 2017-01-03 model (LHO aLOG 33004), I've used the nlinfit values to update the sensing function parameter set. For the actuator coefficients, I've again used 
${CalSVN}/aligocalibration/trunk/Runs/ER10/H1/Scripts/FullIFOActuatorTFs/
    getActValuesForParamsFile.m   (r4126, lc: r4126)
and the clunky process described in LHO aLOG 33092.
Also, I've run the fitting algorithms on *only* the reference measurement data set (i.e. I did not run on multiple measurements worth, assuming no and/or compensating for time-dependence).

I've committed the following model pieces to the Cal SVN repo:
(r = svn revision, lc = last changed revision)

/ligo/svncommon/CalSVN/aligocalibration/trunk/Runs/
O2/DARMmodel/src/
    DARMmodel.m        (r4093, lc: r3511)
    computeSensing.m   (r4093, lc: r4025)
    computeActuation.m (r4093, lc: r4003)

Param Files
O2/Common/params/
    IFOindepParams.conf (r4042, lc: r4002)

ER10/H1/params/
    H1params.conf       (r4121, lc: r4121)

ER10/H1/params/2016-11-12/
    H1params_2016-11-12.conf                      (r3944, lc: 3786)
    measurements_2016-11-12_sensing.conf          (r4122, lc: r4122)
    measurements_2016-11-12_ETMY_L1_actuator.conf (r3944, lc: r3810)
    measurements_2016-11-12_ETMY_L2_actuator.conf (r3944, lc: r3810)
    measurements_2016-11-12_ETMY_L3_actuator.conf (r3944, lc: r3810)
 

More notes:
- I did not update the export of foton implementation of the inverse sensing function. This is because
    (a) DCS "offline" calibration filters DARM_ERR and DARM_CTRL directly with FIR filters. For the DARM_ERR filter, the inverse sensing function is computed directly from the matlab model and does not suffer from IIR filter warping, and does not need low- or high- frequency cutoffs that the front-end pipeline must have for causality.
    (b) If we *were* to regenerate GDS calibration, which correct the front-end for its flaws -- including this inverse sensing function filter -- then we would need to use exactly what filter was in place, and the 
${CalSVN}/aligocalibration/trunk/Runs/
    ER10/H1/Measurements/Foton/
        2016-12-07_H1CALCS_InverseSensingFunction_Foton_SRCD-2N_Gain_tf.txt (r3921, lc: r3921)
file is what was used, at least from 2016-12-07 to 2016-12-22, until it was *actually* updated on 2017-01-03.

What to do from here:
- Use the loop model to push new EPICs records that document model values at calibration line frequencies
- Use the new loop model to generate DCS FIR filters to recalibrate the all data from the pre-winter break from raw DARM_ERR and DARM_CTRL.
- Push hard for uncertainty estimations for both the first part of O2 and data post-winter break.