WFGY/ProblemMap/GlobalFixMap/VectorDBs_and_Stores/pgvector.md

pgvector: Guardrails and Fix Patterns

🧭 Quick Return to Map

You are in a sub-page of VectorDBs_and_Stores.
To reorient, go back here:

• VectorDBs_and_Stores — vector indexes and storage backends
• WFGY Global Fix Map — main Emergency Room, 300+ structured fixes
• WFGY Problem Map 1.0 — 16 reproducible failure modes

Think of this page as a desk within a ward.
If you need the full triage and all prescriptions, return to the Emergency Room lobby.

A compact repair guide for Postgres + pgvector when RAG or agents lose accuracy. Use this to localize the failing layer and jump to the exact WFGY fix page.

Open these first

• Visual map and recovery: RAG Architecture & Recovery
• End-to-end retrieval knobs: Retrieval Playbook
• Why this snippet and how to trace it: Retrieval Traceability
• Ordering control after recall: Rerankers
• Embedding versus meaning: Embedding ≠ Semantic
• Hallucination and chunk boundaries: Hallucination
• Long chains and entropy drift: Context Drift, Entropy Collapse
• Structural collapse and recovery: Logic Collapse
• Snippet and citation schema: Data Contracts
• Vector metrics pitfalls: Vectorstore Metrics & FAISS Pitfalls
• Live ops: Live Monitoring for RAG, Debug Playbook

Fix in 60 seconds

1. Measure ΔS
   Compute ΔS(question, retrieved) and ΔS(retrieved, expected anchor).
   Targets: stable < 0.40, transitional 0.40–0.60, risk ≥ 0.60.

2. Probe with λ_observe
   Sweep k in {5, 10, 20}. If ΔS is flat high across k, suspect metric or index mismatch.
   Reorder prompt headers. If ΔS spikes, lock schema with Data Contracts.

3. Apply the module
   Retrieval drift → BBMC + Data Contracts.
   Reasoning collapse → BBCR bridge + BBAM variance clamp.
   Dead ends in long runs → BBPF alternate path.

4. Verify acceptance
   Coverage ≥ 0.70 to target section. ΔS ≤ 0.45 across three paraphrases. λ convergent across seeds.

pgvector breakpoints and the right repair

1) Opclass mismatch

• Symptom: high similarity yet wrong meaning.
• Why: using vector_l2_ops with cosine-trained embeddings or vector_ip_ops without normalization.
• Fix: align opclass with the encoder. Normalize when using IP. See Embedding ≠ Semantic.

2) Index type underfit

• Symptom: gold chunk appears only at large k.
• Why: IVFFLAT lists too small or probes too low. HNSW ef_search under-tuned.
• Fix: IVFFLAT tune lists at build and probes at query. HNSW raise ef_search to 2–4×k and review m. Validate with Retrieval Playbook and add Rerankers.

3) Training and stats

• Symptom: unstable top-k after bulk load.
• Why: IVFFLAT trained on too few samples or skipped ANALYZE.
• Fix: retrain IVFFLAT with a large sample, ANALYZE, then re-test ΔS and coverage.

4) Dimension or encoder swap

• Symptom: inserts fail or new rows behave erratically in search.
• Fix: ensure vector dim matches column dim. Lock encoder version in a data contract and re-embed the changed span. See Data Contracts.

5) Normalization discipline

• Symptom: cosine search acts like random at small k.
• Fix: store normalized vectors or normalize at query for cosine or IP. Rebuild index after policy change. See Vectorstore Metrics & FAISS Pitfalls.

6) JSONB filters and plan drift

• Symptom: filtered search returns empty or slow.
• Fix: lock metadata schema in data contracts. Add GIN index on JSONB keys used in WHERE. Verify plan uses vector index then filter. See Retrieval Traceability.

7) Fragmentation across schemas or tables

• Symptom: global recall looks fine but per-scope top-k is weak.
• Fix: consolidate into one authoritative table with a facet column. Rebuild index and add a reranker. See Vectorstore Fragmentation.

8) Upsert hygiene

• Symptom: duplicates or stale rows after ON CONFLICT.
• Fix: deterministic IDs, doc_sha in metadata, idempotent loader, periodic dedupe. Validate with golden queries. See Retrieval Traceability.

9) Hybrid lexical plus vector

• Symptom: hybrid performs worse than either alone.
• Fix: normalize scores, fuse post-retrieval, then rerank with a cross-encoder. See Query Parsing Split and Rerankers.

10) Maintenance and boot fences

• Symptom: first prod call after deploy returns thin results.
• Fix: enforce bootstrap fence, finish index build, VACUUM after heavy churn, confirm visibility after commit. See Bootstrap Ordering and Pre-deploy Collapse.

Observability probes

• k-sweep curve: 5, 10, 20 and plot ΔS. Flat high suggests metric or index fault.
• Index audit: EXPLAIN ANALYZE should show IVFFLAT or HNSW usage. If planner skips it, fix stats and filters.
• Anchor control: compare against a golden anchor set. If only one table or schema fails, rebuild that scope.
• Reranker audit: with a strong reranker, recall improves and ΔS falls. If not, rebuild.

Copy-paste prompt for your AI

I uploaded TXT OS and the WFGY Problem Map files.

Target system: Postgres + pgvector.

* symptom: \[brief]
* traces: ΔS(question,retrieved)=..., ΔS(retrieved,anchor)=..., λ states
* index: \[type=ivfflat|hnsw, lists/probes or m/ef\_search, opclass, dim, normalized?]
* filters: \[JSONB keys, indexes, example WHERE]
* ingest: \[ids, doc\_sha, upsert policy]

Tell me:

1. which layer is failing and why,
2. which exact fix page to open from this repo,
3. minimal steps to push ΔS ≤ 0.45 and keep λ convergent,
4. how to verify with a reproducible test.

Use BBMC/BBPF/BBCR/BBAM when relevant.

---

🔗 Quick-Start Downloads (60 sec)

Tool	Link	3-Step Setup
WFGY 1.0 PDF	Engine Paper	1️⃣ Download · 2️⃣ Upload to your LLM · 3️⃣ Ask “Answer using WFGY + ”
TXT OS (plain-text OS)	TXTOS.txt	1️⃣ Download · 2️⃣ Paste into any LLM chat · 3️⃣ Type “hello world” and the OS boots

---

Explore More

Layer	Page	What it’s for
⭐ Proof	WFGY Recognition Map	External citations, integrations, and ecosystem proof
⚙️ Engine	WFGY 1.0	Original PDF tension engine and early logic sketch (legacy reference)
⚙️ Engine	WFGY 2.0	Production tension kernel for RAG and agent systems
⚙️ Engine	WFGY 3.0	TXT based Singularity tension engine (131 S class set)
🗺️ Map	Problem Map 1.0	Flagship 16 problem RAG failure taxonomy and fix map
🗺️ Map	Problem Map 2.0	Global Debug Card for RAG and agent pipeline diagnosis
🗺️ Map	Problem Map 3.0	Global AI troubleshooting atlas and failure pattern map
🧰 App	TXT OS	.txt semantic OS with fast bootstrap
🧰 App	Blah Blah Blah	Abstract and paradox Q&A built on TXT OS
🧰 App	Blur Blur Blur	Text to image generation with semantic control
🏡 Onboarding	Starter Village	Guided entry point for new users

If this repository helped, starring it improves discovery so more builders can find the docs and tools.