| 10 May 2021 |
 Matt Craig | In reply to@slack_astropy_U01NUNNVB0D:openastronomy.org
Ah, I haven't digged into the code that far. So that means that uncertainties on the raw images are completely ignored in the combination? Unfortunately, yes, that is the case 😬😢 | 15:04:33 |
 Tim-Oliver Husser | In reply toundefined
Okay, thanks. When I find some time, maybe I'll have a look. | 15:05:11 |
| Matt Craig | In reply to@slack_astropy_U01NUNNVB0D:openastronomy.org
Okay, thanks. When I find sometime, maybe I'll have a look. On the upside, I will have some time later this week and next to finish up some modest improvements in the speed of the image combination code and do a new release. I can try to get an option for properly propagating uncertainties included. | 15:06:29 |
| Tim-Oliver Husser | In reply to@slack_astropy_U765NS6SG:openastronomy.org
On the upside, I will have some time later this week and next to finish up some modest improvements in the speed of the image combination code and do a new release. I can try to get an option for properly propagating uncertainties included. Even better, thanks! | 15:06:53 |
| Tim-Oliver Husser | In reply toundefined
(edited) ... find sometime, maybe ... => ... find some time, maybe ... | 15:07:02 |
| Matt Craig | In reply to@slack_astropy_U01NUNNVB0D:openastronomy.org
Okay, thanks. When I find some time, maybe I'll have a look. Rieke Bohemann -- this answers part of your question about uncertainty in the flats. | 15:07:02 |
| Matt Craig | In reply to@slack_astropy_U765NS6SG:openastronomy.org
Rieke Bohemann -- this answers part of your question about uncertainty in the flats. > Classic case of RTFM
tbh, tfm really needs some work. If you or your students have suggestions for improving it they would be appreciated! | 15:08:03 |
| Tim-Oliver Husser | In reply toundefined
Funny thing: I played around with Rieke's code and, as I mentioned, normalizing the flats (and their uncertainties) before running combine really helped. So the actual uncertainties were used for the final uncertainty? Or are the final uncertainties calculated on the unscaled data? | 15:08:53 |
| Tim-Oliver Husser | In reply toundefined
(edited) ... code and as ... => ... code and, as ... | 15:09:06 |
| Tim-Oliver Husser | In reply to@slack_astropy_U01NUNNVB0D:openastronomy.org
Funny thing: I played around with Rieke's code and, as I mentioned, normalizing the flats (and their uncertainties) before running combine really helped. So the actual uncertainties were used for the final uncertainty? Or are the final uncertainties calculated on the unscaled data? Sorry, I'd look at the code myself, but I'm in a meeting right now... 😇 | 15:10:37 |
| Matt Craig | In reply to@slack_astropy_U01NUNNVB0D:openastronomy.org
Sorry, I'd look at the code myself, but I'm in a meeting right now... 😇 That is surprising 🤔 | 15:14:22 |
| Tim-Oliver Husser | In reply to@slack_astropy_U765NS6SG:openastronomy.org
That is surprising 🤔 Looks to me that the uncertainties are calculated on the unscaled data:
# set the uncertainty
uncertainty = uncertainty_func(self.data_arr, axis=0)
# Divide uncertainty by the number of pixel (#309)
uncertainty /= np.sqrt(len(self.data_arr) - masked_values) | 15:14:44 |
| Tim-Oliver Husser | In reply to@slack_astropy_U01NUNNVB0D:openastronomy.org
Looks to me that the uncertainties are calculated on the unscaled data:
# set the uncertainty
uncertainty = uncertainty_func(self.data_arr, axis=0)
# Divide uncertainty by the number of pixel (#309)
uncertainty /= np.sqrt(len(self.data_arr) - masked_values) self.data_arr is unscaled, correct? _get_scaled_data is only used for creating a mask | 15:15:24 |
| Tim-Oliver Husser | In reply to@slack_astropy_U01NUNNVB0D:openastronomy.org
self.data_arr is unscaled, correct? _get_scaled_data is only used for creating a mask taken from Combiner.median_combine | 15:17:17 |
| Matt Craig | In reply to@slack_astropy_U01NUNNVB0D:openastronomy.org
taken from Combiner.median_combine Ah, yes, you are correct. To be consistent it should be calculated on the scaled data. On the upside, that is an easy fix. | 15:29:49 |
| Tim-Oliver Husser | In reply to@slack_astropy_U765NS6SG:openastronomy.org
Ah, yes, you are correct. To be consistent it should be calculated on the scaled data. On the upside, that is an easy fix. Thanks. That would be good, at least with a good statistics with many images. In that case I guess we don't really need a proper propagation of uncertainties. | 15:37:12 |
| Tim-Oliver Husser | In reply to@slack_astropy_U01NUNNVB0D:openastronomy.org
Thanks. That would be good, at least with a good statistics with many images. In that case I guess we don't really need a proper propagation of uncertainties. But... closing time for me for today. Thanks for your help! | 15:39:04 |
|  Andrew Liounis joined the room. | 17:14:06 |
| Andrew Liounis changed their display name from slack_astropy_U0224V75Q0Y to Andrew Liounis. | 17:14:38 |
| Andrew Liounis set a profile picture. | 17:14:39 |
| Andrew Liounis changed their profile picture. | 17:14:41 |
|  nick-matrix set a profile picture. | 18:02:57 |
| 11 May 2021 |
|  Server Stats Discoverer (traveler bot) set a profile picture. | 12:42:53 |
| 13 May 2021 |
|  Mfon Eti-mfon changed their profile picture. | 16:14:55 |
|  rencugamob1981 joined the room. | 18:41:40 |
| 14 May 2021 |
|  Bob Auburger joined the room. | 03:40:06 |
| Bob Auburger changed their display name from slack_astropy_U0222Q7JSQ4 to Bob Auburger. | 03:44:07 |
| Bob Auburger set a profile picture. | 03:44:08 |
| Tim-Oliver Husser | In reply to@slack_astropy_U01NUNNVB0D:openastronomy.org
But... closing time for me for today. Thanks for your help! btw, FYI, Matt Craig, I saw that there already is an issue for this on github:
https://github.com/astropy/ccdproc/issues/719 | 12:28:58 |
| 15 May 2021 |
 Sid | Hi, I need to use world_to_pixel for a personal project of mine, but am unable to figure out how to give the inputs. I want to take user input as its parameters. Unlike pixel_to_world, giving two comma-separated floating point values as coordinates doesn't work in this case. I checked for online documentation and help but in vain. Could someone please help me out? I apologise if this is a very simple doubt as I am a complete beginner in this field. | 08:15:00 |
