[roytsmart]

Adds a 26-slide reveal.js deck that walks collaborators through the "Invert a Synthetic Scene with MART" tutorial step by step. The audience already knows CTIS inversion and is evaluating whether to adopt ctis. It is served next to the docs at talks/mart/ (https://ctis.readthedocs.io/en/latest/talks/mart/ once merged).

Contents

• Steps 1–15 follow the tutorial cell by cell. Each shows the tutorial's code and what it prints (shapes, units, results), plus the resulting figure.
• Conventions: named axes (ScalarArray), units, grids as cell edges, and FunctionArray.
• Under the hood: the IdealInstrument forward model (geometry, conservative-regridding weights, the image() pipeline, backproject()), backprojection as the conservative transpose (the rotated-cell example from regridding.transpose_weights_conservative), and an abridged MartInverter.__call__.
• Adopting ctis: wrapping a plain NumPy array as input images; OptikaInstrument vs. subclassing AbstractInstrument; timings at tutorial scale; status and roadmap.

Links from the docs

• A new Talks section on the landing page links to the deck.
• A note at the top of the MART tutorial links to it too. That link is raw HTML because nbsphinx warns about relative links to files outside the source tree.

How it works

• docs/_extra/talks/mart/index.html is the deck: plain HTML that loads reveal.js 6.0.2, KaTeX and highlight.js from jsDelivr. html_extra_path = ['_extra'] copies it into the build unchanged.
• docs/_extra/talks/mart/make_figures.py repeats the tutorial's calculation and saves slide-sized figures, an MP4 of the iterations, and numbers.js. That file holds every printed shape, unit and number the slides quote; the HTML fills them in at load time, so the text can't drift from the code.
• docs/conf.py runs every _extra/talks/*/make_figures.py in a subprocess on each build. It's a subprocess because the agg backend the script needs would drop the jupyter-sphinx figures if it were set in the Sphinx process.
• The shot noise uses a fixed seed, applied after the wavelength sum. That has the same Poisson distribution as IdealInstrument.image(), which has no seed argument.
• .readthedocs.yml installs ffmpeg for the movie. The generated figures are gitignored.

Speaker notes are on every slide (press S), and ?print-pdf gives a PDF.

Test plan

• Extracted every runnable code block from the slides and executed them in order against main (16 blocks, including the own-images snippet)
• make_figures.py gives identical numbers across runs and between the local and Read the Docs builds
• Full local sphinx-build -b html docs docs/_build/html succeeded with no warnings (first revision)
• Screenshotted all 25 slides at 1920×1080 and fixed layout issues
• ruff check and black --check pass on the new script
• Read the Docs preview of this revision

🤖 Generated with Claude Code

https://claude.ai/code/session_01GqceVSp8s2msxj1Dr9MrGU

[codecov]

Codecov Report

✅ All modified and coverable lines are covered by tests.
✅ Project coverage is 100.00%. Comparing base (fc49ae6) to head (e499dd5).
⚠️ Report is 1 commits behind head on main.

Additional details and impacted files

@@            Coverage Diff            @@
##              main       #27   +/-   ##
=========================================
  Coverage   100.00%   100.00%           
=========================================
  Files           24        24           
  Lines          883       883           
=========================================
  Hits           883       883           

Flag	Coverage Δ
unittests	100.00% <ø> (ø)

Flags with carried forward coverage won't be shown. Click here to find out more.

☔ View full report in Codecov by Harness.
📢 Have feedback on the report? Share it here.

🚀 New features to boost your workflow:

• ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.