5/23/18: This repo contains the covariance code I wrote from scratch that determines approximate errors for CMB data, and then calculates Fisher matrix entries, F_ij. The code determines one entry at a time, so multiple function calls are required to build a full matrix. The Fisher ellipse code turns matrices into ellipse plots, allowing us to visualize any degeneracies between CMB parameters.
This code should work for any kind of CMB parameter, although the data files provided here are for the modified gravity parameter, D_0. The named directories contain outputs from our modified New Higgs Inflation version of ISITGR. ‘log_s_N’ refers to the fixed s value, which corresponds to the transition rate from modified gravity back to textbook GR. ‘log_s_8’ means that log_10 (s) = 8.0 for all data contained within the directory (that is, s = 100,000,000). Within each directory there are ‘HIGH’, ‘LOW’, ‘SHORT’, and ‘LONG’ descriptors. ‘HIGH’ and ‘LOW’ refer to whether or not D_0 started above or below the nominal GR value of D_0 = 1.0. ‘SHORT’ and ‘LONG’ refer to the step size used to calculate a numerical derivative. For changes in D_0, long is +/- 10^{-6} around D_0 = 1, and short is +/- 10^{-7} around D_0 = 1. For changes in standard Lambda-CDM parameters (eg: H_0 or Omega_b), long is +/- 10^{-2} and short is +/- 10^{-3} around the Planck best constraint value. To make sure everything is clear, the data inputs for every run are contained within each run’s directory, as ‘values.txt’ for the Lambda-CDM values, and ‘valuespfp.txt’ for the modified gravity parameters. The ‘pfp’ suffix refers to Post Friedmann Parameterization.