Change request #2976
Improve procedure to deal with exclusion regions in reflected background generation in csphagen
| Status: | Closed | Start date: | 07/25/2019 | |
|---|---|---|---|---|
| Priority: | Normal | Due date: | ||
| Assigned To: |  Tibaldo Luigi | % Done: | 100% | |
| Category: | - | |||
| Target version: | 1.7.0 | |||
| Duration: |
Description
Currently the reflected regions are placed (independently of the exclusion regions), and afterward Off regions overlapping with exclusion regions are rejected. This may end up with cases in which many (all) Off regions are discarded because they overlap partially with exclusion regions, while placing the Off regions with prior knowledge of where the exclusion regions are would lead to higher background statistics.
off_new1.png (118 KB)
Recurrence
No recurrence.
History
#1
Updated by Tibaldo Luigi over 4 years ago
- Status changed from New to In Progress
- Assigned To set to Tibaldo Luigi
Duplicates #2886
#2
Updated by Tibaldo Luigi over 4 years ago
- File off_new1.png added
- % Done changed from 0 to 20
I am testing a new procedure in which rather than 1) pre-define the Off regions, 2) reject those overlapping with exclusion regions I place the Off regions one by one and check immediately whether the latest one overlaps with the exclusion region, and if not a new tentative exclusion region is placed with a shift in position angle of 10% of the size of an Off region.
For the test I’m using the HESS public observations of MSH 15-52 and an exclusion region provided by Andreas Specovius. I’m using a circular On region centered at RA = 228.547, Dec = -59.174, and with a radius of 0.2 deg. The current reflected-region method finds at least 2 Off regions only for 9 observations out of the 20 available. The remaining 11 observations (pointed North of the source) are therefore rejected (they have one Off region not overlapping with the exclusion region).
For the new algorithm I’m testing I find >= 2 Off regions for all 20 observations. The image below shows how the Off regions are distributed in this case. Left panel: count map. Right panel: exclusion region. In both panels overlaid: cyan cross - the pointing direction to the North of the source, black circle - On region, green circles - Off regions found by the new algorithm.
#3
Updated by Tibaldo Luigi over 4 years ago
- Status changed from In Progress to Feedback
- % Done changed from 20 to 40
I verified that for the simpler test cases in test_csphagen.py the modifications leave the number of Off regions unchanged:
- the Off regions are exactly the same for the cases with no exclusion regions;
- the number of Off regions is the same with a slight shift in position for the case with a simple exclusion region.
Awaiting for feedback from Andreas to see if the improvement can be considered sufficient or there are other test cases to consider.
#4
Updated by Specovius Andreas over 4 years ago
Nice to hear that your approach works so well!
I am not sure how to understand “one by one”, do you scan the position angle parameter starting from the ON-region and place the corresponding regions as soon as they are free of overlap?
#5
Updated by Tibaldo Luigi over 4 years ago
Exactly.
Are there any other test cases you would recommend to check?
#6
Updated by Tibaldo Luigi about 4 years ago
- Status changed from Feedback to Pull request
- % Done changed from 40 to 100
I think the current changes ready to be merged into devel.
Branch 2976-reflreg
#7
Updated by Knödlseder Jürgen about 4 years ago
- Status changed from Pull request to Closed
Merged into devel.