mattyowl
commented
3 years ago Do you get any error message when using the .pickle files? If not, then there is no problem. Earlier versions might not have been saved in binary format, ones I made later were. The only time I had an issue with the included .pickle files was when switching from python2 -> python3. Since no one should be using python2 any more, I don't think there should be a problem.
I'd also like to clarify that '1x2' refers to a binning of '1' along the x axis (dispersion direction) and '2' along the y axis (spatial direction along the slit)? I assume so, given that our binning of '4 rows x 2 columns' gets written as "2x4" by the rss_mos_create_arc_model script.
Yes, that's right - it's XxY.
The other thing I'd like to check is related to the README saying that each reference is for a particular combination of grating + lamp + detector binning. For the grating, is the camera station/grating angle just the default setting? e.g. for pg0900 the camera station is 31.75 deg and the grating angle is 15.875 deg?
Grating angle shouldn't matter, most of the time - the code shifts + stretches the reference model to match your arc file in order to figure out the wavelength calibration. But the code will actually search multiple reference models (with suffixes __1.pickle, _2.pickle etc.) if they exist. Hence there are some lamp/grating/binning combinations with multiple models (I've been adding reference models as needed).
svw26
Hey,
Something I noticed whilst checking what a .pickle file is supposed to look like is that these ones appear like a normal text file when I view them using
more filename.pickle:more RefModel_PG1300_Cu\ Ar_2x2.pickle]...whereas these (like the new RefModel_PG0900_Xe_2x4.pickle file) are interpreted as a binary file(?), with lots of funny characters appearing, making it look a mess:
I thought maybe it's to do with the latter having non-standard binning. Or do you find that they're all fine? If so, then perhaps something 'extra' is happening for me because I'm going through a container...
The reason I'm asking is that photometric-redshift estimates were available for the SALT programme that Neo is working on, and so I made use of these to predict where we'd expect the emission lines to lie. [It would be annoying to observe such optically-faint sources for a long time only to find that the crucial line(s) coincided with a chip gap.] The downside is that means that Neo and I would need to make reference models for different combinations of pg0900 + lamp + 2x4 + grating angle, so I'm wondering whether the rss_mos_create_arc_model script could take that extra parameter (grating angle) into account?
Many thanks, Sarah x