Inline convolution index bookkeeping in Stan hot path

[ajh-oai]

Description

This PR follows #1273 with a focused cleanup in convolve_with_rev_pmf().

The convolution loop previously called small helper functions on every output element to build a temporary 4-integer index array before taking each dot product. This change keeps the convolution logic the same but computes those loop-local indices directly where they are used, avoiding repeated helper-call and temporary-array overhead inside the hot loop.

Benchmarks

I benchmarked the change locally with paired CmdStan runs on the package's infection-estimation workflow (example_confirmed[1:60], fixed generation time, standard incubation/reporting delays, 2 chains, 250 warmup + 250 sampling draws, seeds 101/202/303/404, order balanced across baseline and branch runs).

Metric	Baseline mean	Branch mean	Mean paired change
reports Stan profile block	4.3150 s	2.6701 s	-38.13%
CmdStan total runtime	10.2738 s	10.1897 s	-0.85%

The reports profile block improved in all four paired runs (-25.78% to -53.59%). Total runtime was directionally slightly lower on average but noisy across seeds, so I view the main result here as a targeted reduction in convolution hot-path overhead rather than a strong claim about end-to-end runtime.

Validation

• git diff --check
• NOT_CRAN=true Rscript -e 'pkgload::load_all(".", quiet = TRUE); testthat::test_file("tests/testthat/test-stan-convole.R")'
  • 24 passed, 0 failed
• Local paired CmdStan benchmark described above; both baseline and branch models compiled and sampled successfully.

AI assistance disclosure

Codex helped identify the optimization seam, benchmark the candidate change, and prepare this PR. I reviewed the final patch and benchmark results before submission.

Initial submission checklist

• My PR is based on a package issue and I have explicitly linked it.
• I have tested my changes locally (using devtools::test() and devtools::check()).
• I have added or updated unit tests where necessary.
• I have updated the documentation if required and rebuilt docs if yes (using devtools::document()).
• I have followed the established coding standards (and checked using lintr::lint_package()).
• I have added a news item linked to this PR.

After the initial Pull Request

• I have reviewed Checks for this PR and addressed any issues as far as I am able.

Summary by CodeRabbit

• Refactor
  • Optimised internal convolution computation performance by streamlining index calculation logic.

[coderabbitai]

Walkthrough

The PR refactors convolve_with_rev_pmf to compute convolution slice boundaries inline within its two summation loops instead of calling helper functions that build intermediate index arrays. The dot-product boundaries are now calculated per iteration and used immediately.

Changes

Convolution index inlining

Layer / File(s)	Summary
Inlined index computation in summation loops inst/stan/functions/convolve.stan	The s in 1:xlen loop inlines s_minus_ylen, start_x, end_x, and start_y calculations; the s in (xlen + 1):len loop inlines s_minus_ylen, start_x, start_y, and end_y (with end_x fixed to xlen). Both compute boundaries and apply them to dot_product slices immediately, replacing calls to calc_conv_indices_xlen and calc_conv_indices_len.

🚥 Pre-merge checks | ✅ 5

✅ Passed checks (5 passed)

Check name	Status	Explanation
Title check	✅ Passed	The title accurately and specifically describes the main change: inlining convolution index bookkeeping computation in a Stan hot path function to improve performance.
Docstring Coverage	✅ Passed	No functions found in the changed files to evaluate docstring coverage. Skipping docstring coverage check.
Linked Issues check	✅ Passed	Check skipped because no linked issues were found for this pull request.
Out of Scope Changes check	✅ Passed	Check skipped because no linked issues were found for this pull request.
Description Check	✅ Passed	Check skipped - CodeRabbit’s high-level summary is enabled.

_{✏️ Tip: You can configure your own custom pre-merge checks in the settings.}

✨ Finishing Touches

🧪 Generate unit tests (beta)

• Create PR with unit tests

---

Thanks for using CodeRabbit! It's free for OSS, and your support helps us grow. If you like it, consider giving us a shout-out.

❤️ Share

• X
• Mastodon
• Reddit
• LinkedIn

_{Comment @coderabbitai help to get the list of available commands and usage tips.}

[github-actions]

This is how benchmark results would change (along with a 95% confidence interval in relative change) if abe5767 is merged into main:

• ✔️default: 24.1s -> 23s [-11.58%, +2.71%]
• ✔️no_delays: 26.9s -> 24.8s [-20.05%, +4.35%]
• ✔️random_walk: 9.63s -> 19.4s [-36.41%, +238.78%]
• ✔️stationary: 13.1s -> 12.1s [-24.15%, +8.53%]
• ✔️uncertain: 1.19m -> 1.17m [-10.73%, +6.5%]
  Further explanation regarding interpretation and methodology can be found in the documentation.

[github-actions]

This is how benchmark results would change (along with a 95% confidence interval in relative change) if 7859a14 is merged into main:

• ✔️default: 26.7s -> 21s [-62.47%, +20.28%]
• ✔️no_delays: 23.8s -> 22.2s [-14.38%, +0.34%]
• ✔️random_walk: 9.63s -> 9.14s [-10.98%, +0.85%]
• ✔️stationary: 12.4s -> 21.4s [-106.69%, +251.56%]
• ✔️uncertain: 1.06m -> 1.1m [-7.91%, +15.51%]
  Further explanation regarding interpretation and methodology can be found in the documentation.