feat(analysis): bind irregular event-time log-rate to an analysis-run profile - #389
feat(analysis): bind irregular event-time log-rate to an analysis-run profile#389seonghobae wants to merge 1 commit into
Conversation
… profile GAP-006 remaining operator-visible slice: cutoff-safe irregular_event_time_v1 maps discrete lags through a and refuses pooled unequal-interval coefficients. Not DSEM, not a Driver p.16 std-family restore, and not implemented-main.
|
Warning Review limit reachedNext included review available in 39 minutes. View limit detailsLimit details: You’ve used the included review currently available. You've used all free OSS reviews for now. Wait for the free limit to reset to keep reviewing this public repository. Review configuration: ⚙️ Run configurationConfiguration used: Organization UI Review profile: CHILL Plan: Pro Plus Run ID: ⛔ Files ignored due to path filters (1)
📒 Files selected for processing (9)
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. Comment |
|
Hour-22 exact-head review request. New unique GAP-006 operator-visible slice (no prior live implementation PR). Current head @opencode-agent review |
![devin-ai-integration[bot]](https://avatars.githubusercontent.com/in/811515?s=60&v=4){kind=small}
| reference_delta, | ||
| lag_clock, | ||
| )?; | ||
| let _ = claim_causal_effect(CausalHeuristic::TemporalPrecedence); |
There was a problem hiding this comment.
🟡 Causal refusal is silently discarded
Every run discards claim_causal_effect's guaranteed refusal and emits a successful artifact. The causal-claim gate never blocks promotion.
Prompt for agents
The irregular-event-time executor calls psychometric_core::claim_causal_effect(CausalHeuristic::TemporalPrecedence), which always returns PsychometricError::CausalUnderidentified, but discards the result. This contradicts ADR 0040's claim that invoking the gate prevents causal promotion. Decide whether this profile is actually attempting a causal claim. If it is, propagate the refusal and return no successful artifact. If it is not, remove the meaningless call and enforce the non-causal boundary through a validated artifact/status contract that consumers cannot reinterpret as causal.
Was this helpful? React with 👍 or 👎 to provide feedback.
| || self.occasion_count < 2 | ||
| || self.interval_count == 0 | ||
| || self.interval_count != self.occasion_count.saturating_sub(1) |
There was a problem hiding this comment.
🟡 Impossible evidence counts pass validation
validate accepts counts exceeding the 100,000-score execution limit, including u64::MAX. Fabricated artifacts can represent runs the executor cannot produce.
Prompt for agents
IrregularEventTimeArtifact::validate checks only the relationship between occasion_count and interval_count. It does not enforce MAX_EVIDENCE_UNITS or ensure occasion_count + excluded_after_cutoff_count stays within that input limit. Add overflow-safe count validation matching execute_irregular_event_time_run's bounded input contract, and add round-trip rejection tests for oversized eligible counts, oversized excluded counts, and overflowing totals.
Was this helpful? React with 👍 or 👎 to provide feedback.
| let mean_log_rate = recover_event_series_mean_log_rate(&eligible.occasions, lag_clock)?; | ||
| let (source_lag, source_delta) = first_source_lag(&eligible.occasions)?; | ||
| let mapped_reference_lag = map_discrete_lag_across_event_intervals( | ||
| source_lag, | ||
| source_delta, | ||
| reference_delta, | ||
| lag_clock, | ||
| )?; |
There was a problem hiding this comment.
🔍 Reference lag represents one pair
mapped_reference_lag uses only the earliest pair, while mean_log_rate summarizes every interval. Clarify this estimand prominently for artifact consumers.
Was this helpful? React with 👍 or 👎 to provide feedback.
| if request.snapshot_id != snapshot_id { | ||
| return Err(AnalysisEngineError::SnapshotMismatch); | ||
| } | ||
| if request.knowledge_cutoff != knowledge_cutoff.to_rfc3339() |
There was a problem hiding this comment.
🔍 Equivalent cutoffs require canonical spelling
Valid RFC 3339 cutoffs denoting the same instant can fail the exact string comparison. Document canonical formatting or compare parsed instants.
Was this helpful? React with 👍 or 👎 to provide feedback.
| ## [Unreleased] | ||
|
|
||
| ### Added | ||
|
|
||
| - **Irregular event-time analysis-run profile**: cutoff-safe `irregular_event_time_v1` binds `recover_event_series_mean_log_rate`, maps pooled discrete lags across unequal intervals, and maps a reference-interval lag through `a` (`analysis_engine`). Not DSEM and not implemented-main. |
There was a problem hiding this comment.
🔍 Unreleased entries remain split
A second Unreleased heading leaves pending changes in two sections. Consolidate them before release tooling consumes the changelog.
Was this helpful? React with 👍 or 👎 to provide feedback.
|
|
||
| ### Added | ||
|
|
||
| - **Irregular event-time analysis-run profile**: cutoff-safe `irregular_event_time_v1` binds `recover_event_series_mean_log_rate`, maps pooled discrete lags across unequal intervals, and maps a reference-interval lag through `a` (`analysis_engine`). Not DSEM and not implemented-main. |
There was a problem hiding this comment.
🔍 Changelog reverses pooling behavior
The entry says unequal-interval lags are pooled and mapped, but the profile rejects pooling. Correct the operator-facing capability description.
Was this helpful? React with 👍 or 👎 to provide feedback.
|
Hour-23 exact-head review request. Current head @opencode-agent review |
1 participant
Summary
GAP-006 remaining operator-visible slice with no live implementation PR: bind Voelkle et al. (2012) irregular event-time local log-rate recovery to a cutoff-safe analysis-run profile.
irregular_event_time_v1output profile onanalysis_enginerecover_event_series_mean_log_rateandmap_discrete_lag_across_event_intervalswithout reimplementing the exponential maptepp.irregular_event_time.v1with inference statuscomposed_interval_mapped_lags_not_dsemNot implemented-main. Exact-head Checks and two independent APPROVE reviews are required. Author COMMENTED is not independent APPROVE. Do not self-approve. Do not --admin merge.
Known-truth
Scores
1.0,0.5,0.25at event times0,1,2recover mean log-rateln(0.5)and mapped reference lag0.25atΔt_ref = 2.Verification