docs: correct T2S summary statistic

[factnn]

Summary

• clarify that paired-seed T2S centers are medians
• retain standard deviation as the across-seed dispersion statistic

The benchmark table values are unchanged; this fixes only the documented aggregation method.

Summary by CodeRabbit

• Documentation
  • Updated benchmark reporting to use median timing with standard deviation across paired seeds.
  • Clarified that the torch.compile reference remains based on a single seed.

[coderabbitai]

Caution

Review failed

The pull request is closed.

ℹ️ Recent review info

⚙️ Run configuration

Configuration used: defaults

Review profile: CHILL

Plan: Pro Plus

Run ID: 3388be3a-afee-4ab6-b06f-894d802567d5

📥 Commits

Reviewing files that changed from the base of the PR and between c3a1b0f and a2cd7e3.

📒 Files selected for processing (1)

• BENCHMARK.md

---

📝 Walkthrough

Walkthrough

Updated benchmark documentation to report median plus standard deviation across paired seeds for canpinn and FlashPDE, while retaining torch.compile as a single-seed reference.

Changes

Benchmark methodology

Layer / File(s)	Summary
Timing aggregation methodology BENCHMARK.md	The “Time to solution” description now uses median plus standard deviation across seeds 42–46 for canpinn and FlashPDE; torch.compile remains single-seed.

Estimated code review effort: 1 (Trivial) | ~2 minutes

✨ Finishing Touches

🧪 Generate unit tests (beta)

• Create PR with unit tests
• Commit unit tests in branch docs/fix-t2s-statistic

---

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.}