logaritmisk
  • Joined on 2023-10-06
logaritmisk pushed to main at logaritmisk/trueskill-tt 2026-09-08 04:04:47 +00:00
f345e7690e fix!: make the joint span slices, not just the latest one
logaritmisk commented on issue logaritmisk/trueskill-tt#46 2026-09-08 00:24:46 +00:00
No joint posterior: anything derived from two or more competitors has to add marginals and overstate its width

posterior_of in 0.5.0 is exactly the shape asked for here — a linear functional, correlations intact — so the API half of this issue is delivered. Thank you.

**It does not yet reach this…

logaritmisk commented on issue logaritmisk/trueskill-tt#47 2026-09-08 00:24:34 +00:00
A node's reported sigma is ~10x wider than the estimate's measured reproducibility

Measured against 0.5.0, as asked. posterior_of cannot answer either question on this model yet, and the reason is structural: it sees the latest time slice, and this is Through-Time.

##…

logaritmisk closed issue logaritmisk/trueskill-tt#6 2026-09-08 00:18:14 +00:00
Opt-in synchronous-EP Schedule with snapshot + barrier merge
logaritmisk commented on issue logaritmisk/trueskill-tt#6 2026-09-08 00:18:10 +00:00
Opt-in synchronous-EP Schedule with snapshot + barrier merge

Closing, per the reasoning above.

To leave a clean record of what was decided rather than just that it was:

  • The Schedule trait this issue would have implemented is gone as of 633a503
logaritmisk pushed to main at logaritmisk/trueskill-tt 2026-09-08 00:17:03 +00:00
d9e85cda1d chore: Release trueskill-tt version 0.5.0
logaritmisk pushed tag v0.5.0 to logaritmisk/trueskill-tt 2026-09-08 00:17:03 +00:00
logaritmisk commented on issue logaritmisk/trueskill-tt#6 2026-09-08 00:15:01 +00:00
Opt-in synchronous-EP Schedule with snapshot + barrier merge

Heads-up: this issue's premise no longer holds after 633a503.

It proposes a synchronous-EP Schedule with snapshot and barrier merge. The Schedule trait has been removed, along with…

logaritmisk pushed to main at logaritmisk/trueskill-tt 2026-09-08 00:14:48 +00:00
633a503900 refactor!: remove the factor-graph surface nothing used, add try_winner
logaritmisk closed issue logaritmisk/trueskill-tt#20 2026-09-08 00:14:48 +00:00
Error API: two variants are never constructed, converge() never returns Err, and MismatchedShape is misused for type errors
logaritmisk closed issue logaritmisk/trueskill-tt#42 2026-09-08 00:14:48 +00:00
The Schedule trait is public API that the engine never uses — wire it into run_chain or make it private
logaritmisk pushed to main at logaritmisk/trueskill-tt 2026-09-08 00:08:36 +00:00
7cf45db5cf feat: add expected_variance_reduction for scored active learning
logaritmisk closed issue logaritmisk/trueskill-tt#49 2026-09-08 00:08:36 +00:00
expected_information_gain is discrete-outcome only, so scored consumers cannot ask "which matchup next?"
logaritmisk pushed to main at logaritmisk/trueskill-tt 2026-09-08 00:02:21 +00:00
c866210c65 feat: add History::predict_margin for scored matchups
logaritmisk closed issue logaritmisk/trueskill-tt#48 2026-09-08 00:02:21 +00:00
Every predict_* is a ranked outcome — there is no predictive distribution for a SCORE
logaritmisk commented on issue logaritmisk/trueskill-tt#47 2026-09-07 23:53:28 +00:00
A node's reported sigma is ~10x wider than the estimate's measured reproducibility

Investigated. Your hypothesis (1) is right, hypothesis (2) is wrong, and the direction is the opposite of what the issue assumes. There is also now a call that gives you the quantity you were…

logaritmisk pushed to main at logaritmisk/trueskill-tt 2026-09-07 23:52:43 +00:00
1f791bcddd test: pin what an additive model does to combined uncertainty
logaritmisk pushed to main at logaritmisk/trueskill-tt 2026-09-07 23:46:40 +00:00
c52e2550af feat: add History::posterior_of for a linear combination of competitors
logaritmisk pushed to main at logaritmisk/trueskill-tt 2026-09-07 23:39:24 +00:00
4924bc8b57 style: use arrays rather than vec! in the calibration fixture
logaritmisk pushed to main at logaritmisk/trueskill-tt 2026-09-07 23:38:52 +00:00
36eacf5f67 test: calibrate the marginals against the exact posterior