zxzhai / emulator Goto Github PK

I was wondering whether the errors are simply the variance over the test boxes or something else. There is also a strange 2.303 value used while training on the monopole (https://github.com/zxzhai/emulator/blob/master/gp_mono_cos_hod_9bins.py#L66), could you please explain to me where does it come from?

If I plot the mean of the monopole and its variance, I get this,

This is the code I use to read the variance, in case it helps,

variance = np.loadtxt(DATA_DIR+"data/training/RSD_results/Cosmo_err.dat")
variance = abs(variance[:,0])

I can't find where the file : HOD_test_np8_n1000.dat is used, but this file contains more parameters than there are in the multipoles test dataset. Am I right to assume that we should only select the first 700? (100 per cosmology), as in:

hod_parameters = np.loadtxt(DATA_DIR + hod_filename)
hod_parameters = hod_parameters[0:700]

Thanks in advance!

Would it be possible to know what is the ordering in the p0 files stored with optimal parameters? Newer versions of George use the WhiteNoise kernel differently and it is difficult to tell whether the parameters are read in the correct order