README.md

# deep-100m-batch-update-eval

Batch-update evaluation workload package generated from `deep-100m-static-search-eval`.

## Dataset

- Source static dataset: `deep-100m-static-search-eval`
- Vector count: `100,000,000`
- Dimension: `96`
- Dtype: `float32`
- Metric: `l2`
- Initial update index: `80,000,000` vectors with external labels equal to `A = P[0:80M]`
- Update order: `update_order.u32`, a seed-`42` permutation of source IDs `[0, 100M)`
- Insert vector source: `base_permuted.fbin`, where row `j` equals `base.fbin[P[j]]`

## Traces

- `insert-20`: starts from `A`, then inserts `P[80M:100M]` in twenty `1,000,000`-vector batches.
- `delete-20`: starts from the static `100M` state, then deletes `P[80M:100M]` in twenty `1,000,000`-vector batches.
- `mixed-replace-100`: keeps `80M` live vectors for 100 rounds; each round deletes a cyclic `1,000,000` source-ID slice and inserts `base_permuted.fbin` row ranges: batches 1-20 use rows `80M:100M`, then batches 21-100 use rows `0:80M`. Insert external IDs use existing `P[80M:100M]` labels for the first `20` rounds, then new labels in `[100,000,000, 180,000,000)`.

Batch JSON files use compact descriptors (`range`, `u32_slice`, and `u32_cyclic_slice`) instead of inline million-ID arrays. Insert `external_ids` continue to express user-visible source IDs. Insert `vector_refs` are row ranges in the reordered insert source and must be read from `base_permuted.fbin`.

## Ground Truth

Checkpoint ground truth is produced by filtering the static source ground truth in source-distance order through the checkpoint owner map. Files are exact top-10 only when every query retains at least 10 active candidates from the static source GT depth. If a checkpoint cannot provide top-10 for every query, the package writes a matching `.invalid.json` marker instead of padding.

## Files

- `workload.json`: workload contract.
- `static-workload-reference.json`: immutable static-search-eval references.
- `source_manifest.json`: generation manifest.
- `update_order.u32`: seed-42 source-ID permutation.
- `initial/index_80m_m32_efc500`: HNSW index built from `base[A]` with labels `A`.
- `groundtruth/active_80m.bin`: initial `80M` checkpoint GT, or `active_80m.invalid.json`.
- `initial/layout-sidecar/index_80m_m32_efc500.*`: optional runtime layout sidecar for the initial HNSW index.
- `initial/pq/pq_m<M>.*`: initial `80M` PQ artifacts reordered for `initial/index_80m_m32_efc500` internal IDs.
- `initial_pq_manifest.json`: source static PQ files and validation samples for the reordered initial PQ artifacts.
- `base_permuted.fbin`: reordered insert vector source, present when insert `vector_refs` are row ranges.
- `reordered_insert_manifest.json`: source, formula, size, and sample-check manifest for `base_permuted.fbin`.
- `traces/*/trace.json`: trace metadata.
- `traces/*/batches/*.json`: compact batch descriptors.
- `traces/*/groundtruth/*`: checkpoint GT or invalid markers.
- `checksums.sha256`: checksums for generated package files.

Static PQ codebooks and metadata are reused. `initial/pq/pq_m<M>.pqcodes` contains only `80,000,000` rows and is ordered by the initial HNSW internal ID, so it can be used directly with `initial/index_80m_m32_efc500`.

## PQ row-order compatibility

The initial 80M PQ payloads were already generated in HNSW internal-ID order;
they do not inherit the former base-row ordering of the static dataset and were
not rewritten during the static PQ correction. For internal row `i`,
`initial/pq/pq_m<M>.pqcodes[i]` encodes the external label stored in
`initial/index_80m_m32_efc500` row `i`.

`initial_pq_manifest.json` records `code_order: hnsw_internal_id` and validation
samples. `tools/reorder_pq_codes_for_index.py` is included to reproduce or
validate the same transformation when a new initial HNSW index is built.