Spectra can have POISSON ERRORS and no error column #11
Conversation
ogip/spec.py
Outdated
|
|
||
| if 'STAT_ERR' in f['SPECTRUM'].data.names: | ||
| stat_err = f['SPECTRUM'].data['STAT_ERR'] | ||
| elif 'POISSERR' in f['SPECTRUM'].header and f['SPECTRUM'].header['POISSERR']: |
There was a problem hiding this comment.
| elif 'POISSERR' in f['SPECTRUM'].header and f['SPECTRUM'].header['POISSERR']: | |
| elif f['SPECTRUM'].header.get('POISSERR'): |
ogip/spec.py
Outdated
| elif 'POISSERR' in f['SPECTRUM'].header and f['SPECTRUM'].header['POISSERR']: | ||
| if 'COUNTS' in f['SPECTRUM'].data.names: | ||
| stat_err = np.sqrt(f['SPECTRUM'].data['COUNTS']) / f['SPECTRUM'].header['EXPOSURE'] | ||
| elif 'RATE' in f['SPECTRUM'].data.names: |
There was a problem hiding this comment.
is that anticipated in the standard? POISSERR with RATE columns? EXPOSURE might be somehow corrected (e.g. deadtime) and RATE*EXPOSURE can be non-int. Interpretation might be complex.
If it is anticipated, could you cite it in a comment for clarity?
There was a problem hiding this comment.
If the POISSERR keywork is set, it is meant that uncertainty is the square root of counts. As the output is only using rate, a correct assumption is that RATE=COUNTS/EXPOSURE and STAT_ERR = sqrt(COUNTS)/EXPOSURE
ogip/spec.py
Outdated
| if 'RATE' in f['SPECTRUM'].data.names: | ||
| rate = f['SPECTRUM'].data['RATE'] | ||
| elif 'COUNTS' in f['SPECTRUM'].data.names: | ||
| rate = f['SPECTRUM'].data['COUNTS'] / f['SPECTRUM'].header['EXPOSURE'] |
There was a problem hiding this comment.
Header is asked for EXPOSURE many times making the lines longer than needed.
Maybe:
| rate = f['SPECTRUM'].data['COUNTS'] / f['SPECTRUM'].header['EXPOSURE'] | |
| rate = f['SPECTRUM'].data['COUNTS'] / exposure |
and above
exposure = f['SPECTRUM'].header['EXPOSURE']
?
|
Did you manage to make progress on this, @ferrigno ? Do you need any help? |
|
I think it is ready now. |
This is just an extension in case of Poissonian errors