I applied a bug fix suggested by John Zweizig in the demodulation routine in the GDS pipeline that reduces error due to finite machine precision. After this, it appears that the kappas as computed by GDS, especially the cavity pole, are significantly less noisy, but still not in agreement with the SLM tool (See this aLOG for reference: https://alog.ligo-wa.caltech.edu/aLOG/index.php?callRep=30888). Below is a table of mean and standard deviation values for the data taken from GDS, SLM, and the ratio GDS / SLM: SLM mean SLM std GDS mean GDS std ratio mean ratio std Re(kappa_tst) 0.8920 0.0068 0.8916 0.0056 0.9995 0.0043 Im(kappa_tst) -0.0158 0.0039 -0.0145 0.0008882 1.0013 0.0041 Re(kappa_pu) 0.8961 0.0080 0.8958 0.0057 0.9997 0.0065 Im(kappa_pu) -0.0050 0.0056 -0.0035 0.0013 1.0015 0.0059 kappa_c 1.1115 0.0094 1.1154 0.0072 1.0035 0.0060 f_c 354.2338 2.9305 345.6435 0.7686 0.9758 0.0084 Here are covariance matrices and correlation coefficient matrices between SLM and GDS: Covariance Correlation Re(kappa_tst) 1.0e-04 * 0.4615 0.3157 1.0000 0.8238 0.3157 0.3181 0.8238 1.0000 Im(kappa_tst) 1.0e-04 * 0.1506 0.0007 1.0000 -0.0216 0.0007 0.0079 -0.0216 1.0000 Re(kappa_pu) 1.0e-04 * 0.6387 0.3113 1.0000 0.6866 0.3113 0.3219 0.6866 1.0000 Im(kappa_pu) 1.0e-04 * 0.3139 -0.0036 1.0000 -0.0490 -0.0036 0.0174 -0.0490 1.0000 kappa_c 1.0e-04 * 0.8895 0.4815 1.0000 0.7118 0.4815 0.5144 0.7118 1.0000 f_c 8.5876 -0.0023 1.0000 -0.0010 -0.0023 0.5908 -0.0010 1.0000 Plots and histograms are attached.