Function for estimating uncertainty on intensities - [merged]

[nabobalis]

In GitLab by @wtbarnes on Jun 11, 2021, 14:21

Merges calc-errors -> master

Fixes #60

A few things left to do:

• [x] Naming? estimate_uncertainty or maybe just uncertainty?
• [x] Clean up docstrings
• [x] Clarify "magic" numbers
• [x] Clarify units in shot noise calculation
• [x] changelog

[nabobalis]

In GitLab by @wtbarnes on Sep 23, 2021, 07:17

added 4 commits

• 1026387d...e79a06f3 - 2 commits from branch LMSAL_HUB/aia_hub:master
• c728c1b2 - add uncertainty estimation function
• 7c0ad1ff - add test data for error estimation

Compare with previous version

[nabobalis]

In GitLab by @wtbarnes on Sep 29, 2021, 10:56

marked the checklist item Clean up docstrings as completed

[nabobalis]

In GitLab by @wtbarnes on Sep 29, 2021, 10:56

marked the checklist item Naming? estimate_uncertainty or maybe just uncertainty? as completed

[nabobalis]

In GitLab by @wtbarnes on Sep 29, 2021, 12:29

added 2 commits

• f15358c4 - add a function that returns the telescope number for a given channel
• 268ceb3d - move get_error_table to calibrate.util; fixup docstrings and clarify magic numbers

Compare with previous version

[nabobalis]

In GitLab by @wtbarnes on Sep 29, 2021, 12:31

I've decided to defer the example to another merge request. Additionally, I'm not sure we need to handle the single dispatch stuff now. There are probably a number of places in the repo where it would be nice to have the ability to operate both on scalars/arrays as well as maps.

[nabobalis]

In GitLab by @wtbarnes on Sep 29, 2021, 12:33

added 6 commits

• 268ceb3d...9c08605f - 2 commits from branch LMSAL_HUB/aia_hub:master
• 31d93206 - add uncertainty estimation function
• 40b9124a - add test data for error estimation
• 0fe26509 - add a function that returns the telescope number for a given channel
• 743197c0 - move get_error_table to calibrate.util; fixup docstrings and clarify magic numbers

Compare with previous version

[nabobalis]

In GitLab by @wtbarnes on Sep 29, 2021, 12:46

added 1 commit

• f8063f11 - changlelog entries

Compare with previous version

[nabobalis]

In GitLab by @wtbarnes on Sep 29, 2021, 12:46

marked the checklist item changelog as completed

[nabobalis]

In GitLab by @wtbarnes on Sep 29, 2021, 12:46

marked this merge request as ready

[nabobalis]

In GitLab by @wtbarnes on Sep 29, 2021, 12:47

Regarding the units on the shot noise issue, @pboerner gave the following response a while back:

The units on the shot noise are confusing…it looks like you’ve cancelled all the units to get the number of photons, taken the sqrt, and put the units back on, which is basically what the IDL function does (it doesn’t have units of course). I guess you could be really rigorous and calculate the standard deviation of the Poisson distribution; it only really differs from sqrt(n) for small n, but maybe that’s more trouble than it’s worth…?

The current implementation is still a bit ambiguous, but I don't think it needs to block this merge request.

[nabobalis]

        The corresponding channel for the observed counts.

[nabobalis]

[nabobalis]

Even if it's kind of redundant, a description here would be nice.

[nabobalis]

        Use the preflight photometric calibration. If True, ``include_eve`` must be False.

[nabobalis]

        Use the EVE photometric calibration. If True, ``include_preflight`` must be False.

[nabobalis]

        If True, include the atomic data errors from CHIANTI in the uncertainty.

[nabobalis]

    `~astropy.units.Quantity`

[nabobalis]

What is going on?

[nabobalis]

Is a paper or a source for this value?

[nabobalis]

In GitLab by @wtbarnes on Sep 30, 2021, 06:05

Commented on aiapy/calibrate/uncertainty.py line 36

changed this line in version 6 of the diff

[nabobalis]

In GitLab by @wtbarnes on Sep 30, 2021, 06:05

Commented on aiapy/calibrate/tests/test_uncertainty.py line 3

changed this line in version 6 of the diff

[nabobalis]

In GitLab by @wtbarnes on Sep 30, 2021, 06:05

Commented on aiapy/calibrate/uncertainty.py line 41

changed this line in version 6 of the diff

[nabobalis]

In GitLab by @wtbarnes on Sep 30, 2021, 06:05

Commented on aiapy/calibrate/uncertainty.py line 43

changed this line in version 6 of the diff

[nabobalis]

In GitLab by @wtbarnes on Sep 30, 2021, 06:05

Commented on aiapy/calibrate/uncertainty.py line 45

changed this line in version 6 of the diff

[nabobalis]

In GitLab by @wtbarnes on Sep 30, 2021, 06:05

Commented on aiapy/calibrate/uncertainty.py line 54

changed this line in version 6 of the diff

[nabobalis]

In GitLab by @wtbarnes on Sep 30, 2021, 06:05

added 1 commit

• e8f5645e - apply suggestions from code review

Compare with previous version

[nabobalis]

In GitLab by @wtbarnes on Sep 30, 2021, 06:17

added 1 commit

• 971229da - more docstring

Compare with previous version

[nabobalis]

In GitLab by @wtbarnes on Sep 30, 2021, 06:18

Commented on aiapy/calibrate/uncertainty.py line 73

This text comes from @pboerner. I'm not sure if there is a reference anywhere?

[nabobalis]

In GitLab by @wtbarnes on Sep 30, 2021, 06:30

Commented on aiapy/calibrate/uncertainty.py line 65

:shrug:

This is the IDL code,

numphot = counts / dnperpht
stdevphot = SQRT(numphot)
shotnoise = (stdevphot * dnperpht)
shotnoise = nullcount + shotnoise / SQRT(n_sample)      ;   DN uncertainty due to shot noise

shotnoise should have the same units as counts, DN/pixel. dnperpht has units of DN / photon so stddevphot has units of sqrt(photon / pixel) such that shotnoise has units of DN / sqrt(photon pixel).

Maybe the trick here is that photon and pixel are really "unitless" so as long as we wind up with DN in the numerator, it doesn't really matter?

[nabobalis]

In GitLab by @wtbarnes on Oct 19, 2021, 12:05

Commented on aiapy/calibrate/uncertainty.py line 65

changed this line in version 8 of the diff

[nabobalis]

In GitLab by @wtbarnes on Oct 19, 2021, 12:05

added 1 commit

• 70375a71 - clarifying comment on shot noise units

Compare with previous version

[nabobalis]

In GitLab by @wtbarnes on Oct 19, 2021, 12:06

marked the checklist item Clarify units in shot noise calculation as completed

[nabobalis]

In GitLab by @codecov on Oct 19, 2021, 12:36

Codecov Report

Merging #116 (9508db3) into master (9c08605) will increase coverage by 0.14%. The diff coverage is 98.64%.

@@            Coverage Diff             @@
##           master     #116      +/-   ##
==========================================
+ Coverage   97.26%   97.41%   +0.14%     
==========================================
  Files          15       16       +1     
  Lines         475      541      +66     
==========================================
+ Hits          462      527      +65     
- Misses         13       14       +1     

Impacted Files	Coverage Δ
aiapy/calibrate/uncertainty.py	97.36% <97.36%> (ø)
aiapy/calibrate/__init__.py	100.00% <100.00%> (ø)
aiapy/calibrate/prep.py	98.18% <100.00%> (ø)
aiapy/calibrate/util.py	100.00% <100.00%> (ø)
aiapy/response/channel.py	100.00% <100.00%> (ø)
aiapy/util/util.py	100.00% <100.00%> (ø)

---

Continue to review full report at Codecov.

Legend - Click here to learn more Δ = absolute <relative> (impact), ø = not affected, ? = missing data Powered by Codecov. Last update 9c08605...9508db3. Read the comment docs.

[nabobalis]

In GitLab by @wtbarnes on Oct 19, 2021, 17:23

added 1 commit

• b212455f - add test for invalid error table input

Compare with previous version

[nabobalis]

In GitLab by @wtbarnes on Oct 19, 2021, 17:38

added 1 commit

• 7b481896 - add test for invalid flag combination

Compare with previous version

[nabobalis]

In GitLab by @wtbarnes on Oct 23, 2021, 19:11

added 1 commit

• 9508db3b - add test for scalar values

Compare with previous version

[nabobalis]

approved this merge request

[nabobalis]

In GitLab by @wtbarnes on Oct 26, 2021, 12:30

resolved all threads

[nabobalis]

In GitLab by @wtbarnes on Oct 26, 2021, 12:32

mentioned in commit 9d7d4a1e2007626e13f61a3a02904ef772d2819c