feat(perturbation-sim): inertia §0 promotion gate + CAKES/CHAODA + witness standing-wave + H ingest#513
Merged
Conversation
…ed column
Flips the §0 anti-invention guardrail review for the one additive SoA member
(operator sign-off 2026-06-16), and makes it real:
- GuardrailVerdict {Proposed, RatifiedReuse} + INERTIA_PROMOTION record. Verdict
RatifiedReuse: inertia_buffer takes ResidueEdge slot INERTIA_SLOT (=5, after the
five contingency factors), reuses an EXISTING value tenant, and invents no new
axis — so it passes §0 by reuse, not waiver. Topology stays the HHTL-OGAR GUID
key; the buffer is one more value, orthogonal by the key/value split (the study's
Spearman(λ₂,buffer)≈0 confirms, does not establish).
- study_slot_assignments(): the 6 members → ResidueEdge slots 0..6, deterministic
and collision-free (the read-carrier assignment).
- buffer::inertia_buffer_column(per_bus_h, df_band): the computed producer —
per-bus impulse_buffer min-max normalized to [0,1] (the form the calibrated spec
stores; degenerate/empty input → zeros, never NaN).
Tests: +3 (ratified-via-reuse, unique-slots-fit-ResidueEdge, column normalized +
monotone-in-H + degenerate-safe). fmt + clippy -D warnings clean; 78 lib tests.
Disjoint from PR #512's files (calibrate.rs / hhtl.rs / TECH_DEBT.md) by design.
The CLAM-family similarity (attraction) / anomaly (repulsion) pair, applied to
grid resilience exactly as the CAKES/CHAODA picture frames it (papillary muscles,
terrain tiles, invoices — domain-agnostic; only the feature adapter is grid-specific):
- HHTL = the family basin ("where in the tree", the GUID-key cascade).
- CAKES (cakes_neighbors): attraction — a basin's k nearest relatives by
resilience-feature distance ("who looks similar / who are my relatives").
- CHAODA (chaoda_scores / anomaly_ranking): repulsion — per-basin kNN-distance
anomaly score, normalized [0,1]; the top-anomaly basin is the fail-first
compartment ("why am I different"). CHAODA_FLAG=0.75 mirrors ndarray CLAM's flag.
- resilience_basin_features: per-basin [λ₂, size, inertia] rows (topology / scale /
buffer — the three orthogonal axes), min-max normalized per axis.
Self-contained CHAODA-lite (one kNN scorer), zero-dep; ndarray::clam's full
ClamTree ensemble is the gated production path (no local ndarray sibling). Example
`chaoda` lights up the planted brittle block (synthetic 8×8 → basin 1.1.0 flagged
score 1.000) and prints its CAKES relatives; runs on real PyPSA core via args.
Tests: +4 (outlier-flagged, CAKES-pulls-similar, degenerate-safe, adapter
normalized+deterministic). fmt + clippy --all-targets -D warnings clean; 82 tests.
…HODS §11) Proves the particle == wave identity §11 asserts: a witness arc (a Markov #1 reference chain over the SoA) read by walking it (particle, O(hops) pointer-chase) equals its standing-wave evaluation via Parseval on the Walsh-Hadamard pyramid (Hᵀ H = N·I, the fwht involution-up-to-N already in sketch.rs): ⟨field, arc⟩ = (1/N)·⟨Ĥfield, Ĥarc⟩. - witness_particle: the pointer-chase walk (∑ field·arc). - field_spectrum: the standing wave, computed ONCE (O(N log N)). - witness_from_spectrum: read any arc off the spectrum (O(N)) — the amortization win (one transform, q arcs) vs q independent particle walks. - witness_wave / particle_equals_wave: convenience + the probe's pass predicate. Self-contained in perturbation-sim; the contract witness_table evaluator is the separate gated step (SoA spine = additive-only behind the iron rules). Example `witness` runs it on the inertia-buffer field (ties §11's "inertia field on power grids" to the promoted member) — particle vs wave agree to 0.00e0. Tests: +4 (Parseval identity, one-transform-many-arcs amortization, non-pow2 + ragged-arc padding, degenerate-safe). fmt + clippy --all-targets -D warnings clean; 86 tests.
Real per-bus inertia H is not in PyPSA/OSM topology (buffer.rs flags this); it comes from operator data (ENTSO-E inventory / ESIOS dispatch / TSO estimate). This is the ingest path, honest about provenance and bundling nothing: - parse_bus_inertia(&str): minimal `bus,H` CSV (`,`/`;`, column-alias bus + h), skipping non-positive/unparseable rows; extra columns ignored (ENTSO-E exports parse as-is). - inertia_for_buses(bus_ids, measured, fallback) -> (Vec<f64>, InertiaProvenance): align measured H to the grid bus order, fill misses with a fallback, and DISCLOSE measured-vs-proxy counts (mirrors PypsaImport's n_estimated_* honesty). - proxy_inertia(n, base, span, seed): deterministic SplitMix64 no-data stand-in, decoupled from wiring (buffer ⊥ topology preserved); never bundled. Feeds straight into inertia_buffer_column (probe-0 promotion gate). Example `inertia_ingest` parses an inline fixture, aligns to 5 buses (1 proxy fallback), prints the buffer column + provenance, and shows the no-data proxy field. Tests: +4 (parse skips bad rows, align discloses proxy fill, proxy deterministic + bounded, ingested H feeds the buffer column). fmt + clippy --all-targets -D warnings clean; 90 tests. No external data committed (network/data policy).
|
Warning Review limit reached
More reviews will be available in 4 minutes and 30 seconds. Learn how PR review limits work. Your organization has used up its prepaid credits, and credit purchases are no longer available. Enable the review add-on in the billing tab to keep reviews running — you're only billed for reviews past your plan's rate limits ($0.25/file). ⌛ How to resolve this issue?After more reviews become available, a review can be triggered using the We recommend that you space out your commits to avoid hitting the rate limit. 🚦 How do rate limits work?CodeRabbit enforces hourly rate limits for each developer per organization. Our paid plans include higher PR review limits than trial, open-source, and free plans. In all cases, reviews become available again over time. During sustained high-volume PR review activity, CodeRabbit may temporarily slow when the next review becomes available. Please see our Fair Usage Limits Policy for further information. ℹ️ Review info⚙️ Run configurationConfiguration used: Organization UI Review profile: CHILL Plan: Pro Plus Run ID: 📒 Files selected for processing (10)
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 |
![chatgpt-codex-connector[bot]](https://avatars.githubusercontent.com/in/1144995?s=60&v=4){kind=small}
There was a problem hiding this comment.
💡 Codex Review
Here are some automated review suggestions for this pull request.
Reviewed commit: 723472a955
ℹ️ About Codex in GitHub
Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you
- Open a pull request for review
- Mark a draft as ready
- Comment "@codex review".
If Codex has suggestions, it will comment; otherwise it will react with 👍.
Codex can also answer questions or update the PR. Try commenting "@codex address that feedback".
| ',' | ||
| }; | ||
| let cols: Vec<String> = header | ||
| .split(delim) |
There was a problem hiding this comment.
When the inertia table comes from a CSV writer that quotes fields, this raw split leaves quotes in the header/value strings and does not respect delimiters inside quoted fields. For an export like "bus_id","inertia_h" with rows such as "ES_NUC_1","6.5", the header lookup fails (or the numeric parse fails on quoted values), so parse_bus_inertia returns no measured entries and the alignment path silently falls back to proxies for every bus. Reuse the quote-aware CSV splitting already used by the PyPSA ingest path or unquote fields before matching/parsing.
Useful? React with 👍 / 👎.
AdaWorldAPI
pushed a commit
that referenced
this pull request
Jun 16, 2026
Two lance-graph PRs merged this evening, both touching the standalone perturbation-sim crate: #513 (merged 20:27:59 UTC, 8a3e335) — +1009/-2: - §0 promotion gate: GuardrailVerdict::RatifiedReuse for inertia_buffer (takes ResidueEdge INERTIA_SLOT=5, reuses existing tenant, no new axis → passes by reuse not waiver) - Probe 1: CAKES + CHAODA-lite over HHTL basins - Probe 2: witness arc as standing wave via Parseval (particle == wave to 0.00e0; O(N)-per-arc read-many amortization) - Probe 3: per-bus inertia (H) ingest path #512 (merged 20:33:00 UTC, 1e23c41) — +591/-5: CODE FIXES (review of #511): - calibrate.rs: divide-by-zero guard on degenerate grid (CodeRabbit Major) - basin_lambda2: assert_eq!(keys.len(), grid.n, …) precondition, promoting silent corruption to loud panic (CodeRabbit Major) - TECH_DEBT.md MD018 reflow (CodeRabbit Minor) DOCTRINE (the structural delivery, NEW to this session): - .claude/knowledge/core-first-transcode-doctrine.md (218 LOC, mandatory-read) - 3 specialist agent cards: core-first-architect, core-gap-auditor, adapter-shaper - BOOT.md + README.md wires - EPIPHANIES entry (24 LOC) - CLAUDE.md doctrine wire-up (+21 LOC, content unread by this thread) The core-first doctrine is likely directly aligned with the ontology-first stance the operator locked in on odoo-rs this session (Odoo's ontology runs natively in SurrealDB, codegen is the deferred "cut tail"). Cross-doctrine reconciliation between core-first-transcode-doctrine.md and the existing lab-vs-canonical spine not yet inspected here — flagged in PR_ARC entry confidence line. Retroactive batch hygiene — both merges landed before the same-commit rule could fire. PR_ARC prepend (newest #512 above #513 above #511) + LATEST_STATE dated bullets + recent-shipped rows. https://claude.ai/code/session_01Xzyc27Nx3f8WC5KzwfWfjx
AdaWorldAPI
pushed a commit
that referenced
this pull request
Jun 16, 2026
Two lance-graph PRs merged this evening, both touching the standalone perturbation-sim crate: #513 (merged 20:27:59 UTC, 8a3e335) — +1009/-2: - §0 promotion gate: GuardrailVerdict::RatifiedReuse for inertia_buffer (takes ResidueEdge INERTIA_SLOT=5, reuses existing tenant, no new axis → passes by reuse not waiver) - Probe 1: CAKES + CHAODA-lite over HHTL basins - Probe 2: witness arc as standing wave via Parseval (particle == wave to 0.00e0; O(N)-per-arc read-many amortization) - Probe 3: per-bus inertia (H) ingest path #512 (merged 20:33:00 UTC, 1e23c41) — +591/-5: CODE FIXES (review of #511): - calibrate.rs: divide-by-zero guard on degenerate grid (CodeRabbit Major) - basin_lambda2: assert_eq!(keys.len(), grid.n, …) precondition, promoting silent corruption to loud panic (CodeRabbit Major) - TECH_DEBT.md MD018 reflow (CodeRabbit Minor) DOCTRINE (the structural delivery, NEW to this session): - .claude/knowledge/core-first-transcode-doctrine.md (218 LOC, mandatory-read) - 3 specialist agent cards: core-first-architect, core-gap-auditor, adapter-shaper - BOOT.md + README.md wires - EPIPHANIES entry (24 LOC) - CLAUDE.md doctrine wire-up (+21 LOC, content unread by this thread) The core-first doctrine is likely directly aligned with the ontology-first stance the operator locked in on odoo-rs this session (Odoo's ontology runs natively in SurrealDB, codegen is the deferred "cut tail"). Cross-doctrine reconciliation between core-first-transcode-doctrine.md and the existing lab-vs-canonical spine not yet inspected here — flagged in PR_ARC entry confidence line. Retroactive batch hygiene — both merges landed before the same-commit rule could fire. PR_ARC prepend (newest #512 above #513 above #511) + LATEST_STATE dated bullets + recent-shipped rows. https://claude.ai/code/session_01Xzyc27Nx3f8WC5KzwfWfjx
AdaWorldAPI
pushed a commit
that referenced
this pull request
Jun 17, 2026
Two lance-graph PRs merged this evening, both touching the standalone perturbation-sim crate: #513 (merged 20:27:59 UTC, 8a3e335) — +1009/-2: - §0 promotion gate: GuardrailVerdict::RatifiedReuse for inertia_buffer (takes ResidueEdge INERTIA_SLOT=5, reuses existing tenant, no new axis → passes by reuse not waiver) - Probe 1: CAKES + CHAODA-lite over HHTL basins - Probe 2: witness arc as standing wave via Parseval (particle == wave to 0.00e0; O(N)-per-arc read-many amortization) - Probe 3: per-bus inertia (H) ingest path #512 (merged 20:33:00 UTC, 1e23c41) — +591/-5: CODE FIXES (review of #511): - calibrate.rs: divide-by-zero guard on degenerate grid (CodeRabbit Major) - basin_lambda2: assert_eq!(keys.len(), grid.n, …) precondition, promoting silent corruption to loud panic (CodeRabbit Major) - TECH_DEBT.md MD018 reflow (CodeRabbit Minor) DOCTRINE (the structural delivery, NEW to this session): - .claude/knowledge/core-first-transcode-doctrine.md (218 LOC, mandatory-read) - 3 specialist agent cards: core-first-architect, core-gap-auditor, adapter-shaper - BOOT.md + README.md wires - EPIPHANIES entry (24 LOC) - CLAUDE.md doctrine wire-up (+21 LOC, content unread by this thread) The core-first doctrine is likely directly aligned with the ontology-first stance the operator locked in on odoo-rs this session (Odoo's ontology runs natively in SurrealDB, codegen is the deferred "cut tail"). Cross-doctrine reconciliation between core-first-transcode-doctrine.md and the existing lab-vs-canonical spine not yet inspected here — flagged in PR_ARC entry confidence line. Retroactive batch hygiene — both merges landed before the same-commit rule could fire. PR_ARC prepend (newest #512 above #513 above #511) + LATEST_STATE dated bullets + recent-shipped rows. https://claude.ai/code/session_01Xzyc27Nx3f8WC5KzwfWfjx
5 tasks
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
2 participants
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
What
Four additive deliverables on the standalone (workspace-
excluded)perturbation-simcrate, continuing the resilience arc after #511. Disjoint files from #512 (calibrate.rs/basin_lambda2body /TECH_DEBT.md) by design.1.
inertia_buffer§0 promotion gate (33d66ca9)Flips the anti-invention guardrail review for the one additive SoA member (operator sign-off 2026-06-16) and makes it real:
GuardrailVerdict {Proposed, RatifiedReuse}+INERTIA_PROMOTION. Verdict RatifiedReuse:inertia_buffertakes ResidueEdge slotINERTIA_SLOT(=5), reuses an existing value tenant, invents no new axis → passes §0 by reuse, not waiver. Topology stays the HHTL-OGAR GUID key; the buffer is one more value, orthogonal by the key/value split.study_slot_assignments(): the 6 members → ResidueEdge slots 0..6, collision-free.buffer::inertia_buffer_column(): the computed producer (per-busimpulse_buffer, normalized to [0,1], degenerate-safe).2. Probe 1 — CAKES + CHAODA over HHTL basins (
8d87ef0e)The CLAM-family similarity (attraction) / anomaly (repulsion) pair, applied to grid resilience:
cakes_neighbors) = "who are my relatives"; CHAODA (chaoda_scores/anomaly_ranking) = per-basin kNN-distance anomaly, the fail-first compartment (CHAODA_FLAG=0.75 mirrors ndarray CLAM's flag).resilience_basin_features: per-basin[λ₂, size, inertia]rows (topology / scale / buffer — the three orthogonal axes).ndarray::clam's fullClamTreeensemble is the gated production path (no local ndarray sibling). Examplechaodaflags the planted brittle block (basin 1.1.0, score 1.000).3. Probe 2 — witness arc as a standing wave (
006d2322, METHODS §11)Proves the
particle == waveidentity: a witness arc walked hop-by-hop (O(hops)pointer-chase) equals its Walsh-pyramid evaluation via Parseval (Hᵀ H = N·I):⟨field, arc⟩ = (1/N)⟨Ĥfield, Ĥarc⟩.witness_particle(the walk) /field_spectrum(transform once) /witness_from_spectrum(read many arcs,O(N)each — the amortization win) /witness_wave/particle_equals_wave.witnessruns it on the inertia-buffer field; particle vs wave agree to 0.00e0. The contractwitness_tableevaluator remains the separate gated step (SoA spine = additive-only behind the iron rules).4. Probe 3 — per-bus inertia (H) ingest path (
723472a9)Real H is not in PyPSA/OSM topology; this is the honest ingest path:
parse_bus_inertia(bus,HCSV) /inertia_for_buses(align to grid bus order +InertiaProvenancedisclosing measured-vs-proxy, mirroringPypsaImport'sn_estimated_*) /proxy_inertia(deterministic SplitMix64 no-data stand-in, topology-blind). Feedsinertia_buffer_column.inertia_ingestparses an inline fixture (1 proxy fallback) and prints the column + provenance.Gates
fmt+clippy --all-targets -D warnings+testgreen on the standalone crate (90 lib tests, up from 75). New examples:chaoda,witness,inertia_ingest.Not in scope (honest)
ndarray::clamensemble bridge and the contractwitness_tableevaluator are deliberately deferred (gated behind features / the cognitive-spine iron rules).https://claude.ai/code/session_01CcpLeEC3XK8Eye53GKBVvi
Generated by Claude Code