From edges-analysis: total power filter

[steven-murray]

In edges-analysis there is the possibility of doing a total power filter when averaging calibration spectra. Here's the code:

	# total power filter
	# ------------------
	tp = np.sum(ta_sel1, axis=1)
	W  = np.ones(len(tp))
	IN = np.arange(len(tp))
	
	Nsigma_tp = 3
	bad_old   = -1
	bad       =  0		
	iteration =  0
	
	

	while (bad > bad_old):

		iteration = iteration + 1
		print('Iteration: ' + str(iteration))
			
		std = np.std(tp[W>0])
		ave = np.mean(tp[W>0])
		res = tp-ave	

		IN_bad    = IN[(np.abs(res) > Nsigma_tp*std)]
		W[IN_bad] = 0
		
		bad_old = np.copy(bad)
		bad     = len(IN_bad)

		print('STD: ' + str(np.round(std,3)) + ' K')
		print('Number of bad points excised: ' + str(bad))
		print(IN_bad)

And in its original context: https://github.com/edges-collab/edges-analysis/blob/ac17ed386f4dfdccc8872d6fdf7d0926d899fecd/src/edges_analysis/basic.py#L1006

@gitul : is this useful to keep in this code?

[gitul]

Yes Steven, I would like to keep this filter if possible. Thank you Raul ps: I don't know if a simple email reply is the right way to reply to git comments/questions.

[steven-murray]

@gitul excellent, thanks. I'll work it into edges-cal.

PS: yes, you can reply via email, though usually clicking the link to go to the issue on Github gives you more options and better formatting :-)

[steven-murray]

This seems to be more pertinent to field data than lab data, so I will leave it in edges-analysis.