STIPS 2.3.1 bug fixes with the documentation updates by eunkyuh · Pull Request #204 · spacetelescope/STScI-STIPS

eunkyuh · 2025-12-17T21:51:43Z

Several bug fixes include:

Proper unit conversion for the Pandeia background
Logic to replace SCAs with WFIs (resolves STIPS 2.3 fast_galaxy=True very slow #201)
Distinguished in the output header for the input X & Y positions and the ePSF X & Y positions (resolves Source positions in HISTORY and in screen output are off by 21 in both X and Y #200)

Documentation updates include

Remaining mention of WebbPSF to replace with STPSF
Removal of 'oversample' key in the example that no longer works in the newer versions of STIPS
Removal of JBT mentions since STIPS no longer uses the tool
psf_grid_size fixed to 3x3
Updated help desk contact instructions (resolves Update the help (for real this time) #202)
Alternate package managers to consider in installation instructions (resolves Update to installation instructions #185)

Installation yml updates include pandeia.engine requirements to be 2025.9 or higher and allowing newer versions of poppy than 1.0.3. The utilities infrastructure has also been updated so it can work with newer versions of Pandeia.

Implemented dynamic calculation of zeropoints based on WFI throughput files from the user's local Pandeia reference data. This should help mitigate output flux mismatches between STIPS and Pandeia if a user has installed an unexpected version of the latter. The calculated values from the current Pandeia reference data match those found in the Roman Technical Repo.

ojustino · 2026-04-22T21:39:10Z

-                ins.detector = self.detector.replace('WFI', 'SCA')
+                ins.detector = self.detector.replace('SCA', 'WFI')


This should be left as it was – before v2.1.0, STPSF used SCANN to refer to detectors instead of WFINN. The ins object here comes from STPSF. STIPS supports STPSF v2.0.0, so when that version is installed, we have to revert to calling detectors with "SCA" as this line did before this PR.

The real bug happens with ins.psf_grid(). STPSF appends its internal detector name to the files generated with that function, so v2.0.0 will append "SCANN.fits". STIPS' AstroImage.make_epsf_array() looks for file names ending in "WFINN.fits". When it doesn't find them, it regenerates the PSF grid, even when psf_cache_enable is on. That causes the extra computation time experienced in #201.

I've added a commit to make the file name check in AstroImage.make_epsf_array() more backward compatible as well since we can't change the behavior of ins.psf_grid().

ojustino · 2026-04-22T22:04:49Z

-            ot['x'] = Column(data=xfs+self.out_origin, unit='pixel')
-            ot['y'] = Column(data=yfs+self.out_origin, unit='pixel')
+            ot['x_epsf'] = Column(data=xfs+self.out_origin, unit='pixel')
+            ot['y_epsf'] = Column(data=yfs+self.out_origin, unit='pixel')
+            ot['x_input'] = Column(data=self.out_origin, unit='pixel')
+            ot['y_input'] = Column(data=self.out_origin, unit='pixel')


One of these should remain as 'x' and 'y' instead of being moved to a different name in order to maintain backward compatibility in the headers. I will change 'x_epsf' and 'y_epsf' back and also edit the updated text in bugs.rst.

Looks good to me!

ojustino · 2026-04-23T21:34:40Z

We would still like this release to start dynamically updating WFI.ZEROPOINTS_AB (based on ~~synphot reference files~~ Pandeia throughput files) instead of hard-coding it. Since STIPS no longer pins Pandeia to a specific version, the possibility exists that a new Pandeia release might incorporate zero points calculated from different ~~synphot reference files~~ Pandeia throughput files than were used with the latest version of STIPS.

ojustino · 2026-04-27T19:49:33Z

+            self.photfnu = self.PHOTFNU[self.filter]
+            self.photplam = self.PHOTPLAM[self.filter]
+            self.custom_background *= self.convertToCounts('p')


This function is triggered when Instrument.reset() attempts to set self.background = self.pixel_background. (Users would encounter it when ObservationModule.nextObservation() callsInstrument.reset().)

In Instrument.reset(), self.photfnu and self.photplam are defined just after self.background, but this highlighted snippet needs them to exist beforehand because Instrument.convertToCounts() references them.

I edited Instrument.reset() to define self.photfnu and self.photplam before self.background, removing the need to set them again here in Instrument.pixel_background_unit().

I also changed the log message to show the converted background value in counts/s.

If quantum efficiency is always assumed to be 1 across the package, I do still wonder whether Instrument.convertToCounts() should have separate paths for photons and electrons. If not, should we make 'p' and 'e' equivalent in that function (with both being read as 'p'?) and note the change both in the docstring and in docs/using_stips/catalogue_formats.html (especially in the Mixed catalogs section)?

I agree with you that we should make 'p' and 'e' equivalent in that function and note that STIPS assumes QE=1. But all together, I think we are also assuming counts=p=e as well with gain = 1 because in the validation code, we don't do the gain correction to compare the STIPS output to Pandeia (e- and e-/s).

I've changed Instrument.convertToCounts() to treat 'e' like 'p' and edited the docs as well.

eunkyuh · 2026-05-07T14:18:01Z

+	  conversion assumes a quantum yield of 1, where photons are equivalent to
+	  electrons.
+
+	* the string value 'jbt', which will use the JBT background tool to


I'm not sure if the current STIPS is configured to call JBT. If yes, we want to change that to RBT but if not, I think we need to remove any mention of JBT.

OK, I see the v2.0 changelog includes the removal of JBT. It's still mentioned in stips/data/stips_config.yaml, so I'll remove that as well as the mention in the documentation. observation_jbt_location should also be removed from this docs page as well.

eunkyuh · 2026-05-07T15:22:29Z

For Pandeia version later than 2025.9, we need this extra environment for Pandeia:

export PSF_DIR="<absolute_path_to_this_folder>/ref_data/pandeia_psfs"

ojustino · 2026-05-08T21:01:11Z

Merging after successful test run and reaching consensus with @eunkyuh.

eunkyuh requested a review from ojustino December 17, 2025 21:51

ojustino reviewed Apr 22, 2026

View reviewed changes

STIPS 2.3.1 bug fixes with the documentation updates

972a2ba

ojustino force-pushed the release-2.3.1 branch 5 times, most recently from 24fad5f to 88688a8 Compare April 23, 2026 15:52

ojustino added 10 commits April 23, 2026 12:02

Use PyPI stsci_rtd_theme; allow new poppy versions

f8d1160

Continued minimum requirements updates

30cce42

Clean up wfi.py

d3c8d09

Update help desk contact information

895c0d6

Undo some initial PR edits for back compatibility

0647559

Mention more package managers in installation docs

beba000

More documentation edits

20eeb4e

Fix sphinx_issues warnings on changelog

df3ac57

Fix PSF grid cache issue with STPSF 2.0.0

37ce8f1

Followed release instructions

a941d41

ojustino force-pushed the release-2.3.1 branch from 88688a8 to a941d41 Compare April 23, 2026 16:03

Make WFI zeropoint calculation dynamic

de5b606

ojustino reviewed Apr 27, 2026

View reviewed changes

ojustino added 2 commits April 27, 2026 16:37

Simplify background calculation and fill out docs

46ce8c2

Handle extra reference PSF data in pandeia 2026+

6835ab7

eunkyuh commented May 7, 2026

View reviewed changes

ojustino added 3 commits May 7, 2026 12:27

Remove JBT from docs and config file

31ec12b

Add new Pandeia environment variable to docs

5233024

Note photon/electron equality from QE=1 assumption

8d61f83

ojustino added 2 commits May 7, 2026 13:08

Followed release instructions again

fa7289c

Update Python versions in tests

52642b8

ojustino approved these changes May 8, 2026

View reviewed changes

Don't confuse electrons and ergs

4f07df1

ojustino merged commit abe1e55 into main May 8, 2026
11 checks passed

		ins.detector = self.detector.replace('WFI', 'SCA')
		ins.detector = self.detector.replace('SCA', 'WFI')

Conversation

eunkyuh commented Dec 17, 2025 • edited by ojustino Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

ojustino Apr 22, 2026

Choose a reason for hiding this comment

Uh oh!

ojustino Apr 23, 2026

Choose a reason for hiding this comment

Uh oh!

ojustino Apr 22, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Choose a reason for hiding this comment

Uh oh!

eunkyuh May 7, 2026

Choose a reason for hiding this comment

Uh oh!

ojustino commented Apr 23, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

ojustino Apr 27, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Choose a reason for hiding this comment

Uh oh!

ojustino Apr 27, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Choose a reason for hiding this comment

Uh oh!

ojustino Apr 27, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Choose a reason for hiding this comment

Uh oh!

eunkyuh May 7, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Choose a reason for hiding this comment

Uh oh!

ojustino May 7, 2026

Choose a reason for hiding this comment

Uh oh!

eunkyuh May 7, 2026

Choose a reason for hiding this comment

Uh oh!

ojustino May 7, 2026

Choose a reason for hiding this comment

Uh oh!

eunkyuh May 7, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Choose a reason for hiding this comment

Uh oh!

ojustino commented May 8, 2026

Uh oh!

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

2 participants

eunkyuh commented Dec 17, 2025 •

edited by ojustino

Loading

ojustino Apr 22, 2026 •

edited

Loading

ojustino commented Apr 23, 2026 •

edited

Loading

ojustino Apr 27, 2026 •

edited

Loading

ojustino Apr 27, 2026 •

edited

Loading

ojustino Apr 27, 2026 •

edited

Loading

eunkyuh May 7, 2026 •

edited

Loading

eunkyuh May 7, 2026 •

edited

Loading