Digitized Observatory Resources for Automated Data Operations (DORADO) is a python package that aims to expand Astropy(and affiliated packages) and to be a simple and common framework for data storage, reduction, and analysis, tailored for life at the Allan I. Carswell Observatory at York university, Toronto, Ontario, Canada
1
stars
0
forks
source link
Output file size inflation due to AUX HDUs and Bit-bloat #6
Auxiliary HDUs (MASK, UNCERT) are both un-needed/unwanted at the current time; but are also affected by datatype bit precision inflation/bloat that is a broader issue that has been omnipresent since the DRACO days.
It is most likely that the bit bloat issue is born out of interactions between astropy and numpy where the astropy data is taking on the default numpy array float precision of np.float64.
The aux HDUs are something that definitely requires more investigation to nail down.
Auxiliary HDUs (MASK, UNCERT) are both un-needed/unwanted at the current time; but are also affected by datatype bit precision inflation/bloat that is a broader issue that has been omnipresent since the DRACO days.
Example
It is most likely that the bit bloat issue is born out of interactions between astropy and numpy where the astropy data is taking on the default numpy array float precision of
np.float64
.The aux HDUs are something that definitely requires more investigation to nail down.