Open bhuvjain opened 5 years ago
I think it's not really a problem of the coordinate, more like an artifact of the way I'm plotting it (I just did a scatter plot of RA/DEC so the stuff near the pole is distorted), if you just calculate e.g. the second moments of the map using the pixel RA/DEC you should get no distortion. Alternatively, if you want to make plots that are not distorted you can do some flat-sky projection anchored on the center of the cluster.
Try e.g. plotting a scatter plot of ra_proj, dec_proj with ra_proj = (ra-ra0)cos(dec0) dec_proj = dec-dec0 where ra0, dec0 is the coordinate of the cluster
Since the coordinates used in the original scatter plot were galactic latitude and longitude I tried to implement the flat-sky projection (using the equations you gave above) with those, does that make sense to do at all since they are different coordinate systems? These are the plots I got for cluster 7 (with equally spaced axes), is this anywhere near what the projected is supposed to look like?
@tdacunha I think you caught a bug of mine :) so I should be plotting
mplot.scatter(lon_y[mask_cluster], (90. - lat_y[mask_cluster]), c=sz_map[mask_cluster], cmap='nipy_spectral')
instead of
mplot.scatter(lat_y[mask_cluster], (90. - lon_y[mask_cluster]), c=sz_map[mask_cluster], cmap='nipy_spectral')
I think the lon --> RA lat --> DEC should hold... can you try it again? also I can't tell from your code, but np.cos() takes in angles in radians, so make sure you convert it if not already.
Ok editing it now! Quick question though, why is the lat_y subtracted from 90 degrees in the scatter plot? Wasn't theta_y already subtracted from 90 to get lat_y?
you're right, that's not necessary
Ok here are some of the originals and projected plots:
Cluster 0
Cluster 1
Cluster 2
Cluster 5
Here are the stacks with all 10 clusters:
Here is what it would be if we cherrypicked the same ones as when we were originally stacking:
About half the cluster contours seem elongated. Worth checking if its a coordinate issue. If it is real, it suggests we should plot the two quadrants along and perpedincular to the major axis separately..