diff --git a/4-Infrastructure/shim/rrc_bosonic_tensor_network.py b/4-Infrastructure/shim/rrc_bosonic_tensor_network.py index ef1c8563..ad98778b 100644 --- a/4-Infrastructure/shim/rrc_bosonic_tensor_network.py +++ b/4-Infrastructure/shim/rrc_bosonic_tensor_network.py @@ -683,10 +683,10 @@ def main() -> int: print(f"\n First failure:") print(f" N={first_fail['n']} modes, {first_fail['n_photons']} photon(s)") print(f" Error: {first_fail['error'][:120]}") - fail_dim = math.comb(first_fail["n"], first_fail["n_photons"]) if first_fail["n"] >= first_fail["n_photons"] else 0 + fail_dim = math.comb(first_fail["n"] + first_fail["n_photons"] - 1, first_fail["n_photons"]) if first_fail["n"] >= first_fail["n_photons"] else 0 print(f" Hilbert dim ≈ {fail_dim:,}") - max_ok_dim = math.comb(max_ok, max_ok_photons) if successes and max_ok >= max_ok_photons else 0 + max_ok_dim = math.comb(max_ok + max_ok_photons - 1, max_ok_photons) if successes and max_ok >= max_ok_photons else 0 print(f"\n Max SUCCESS: N={max_ok}, p={max_ok_photons}, dim≈{max_ok_dim:,}") # Receipt diff --git a/6-Documentation/docs/GLOSSARY.md b/6-Documentation/docs/GLOSSARY.md index e9e5d263..ceeba62b 100644 --- a/6-Documentation/docs/GLOSSARY.md +++ b/6-Documentation/docs/GLOSSARY.md @@ -112,7 +112,13 @@ surface. | **CM** | Concept Map — JSONL artifact mapping files to extracted concepts. | `shared-data/data/concept_map/` | | **JM** | Jaccard Matcher — arxiv paper discovery via keyword overlap. | `extraction/jaccard_matcher.py` | | **PTC** | PIST Trace Classifier — Lean-first proof trace spectral classifier. | `PistClassifyTrace.lean` | -| **SS** | SilverSight — standalone Lean repo for schema/layout/wireformat/codec. | `/tmp/SilverSight` | +| **SS** | SilverSight — standalone Lean repo for schema/layout/wireformat/codec. | `/home/allaun/SilverSight` | + +| **BMCTE** | Bosonic Monte Carlo Tensor Estimation — stochastic observable estimator for linear-optical systems in the `p ≪ N` regime. | `rrc_bosonic_tensor_network.py` | +| **SBOE** | Stochastic Bosonic Observable Estimator — class of methods that estimate photonic observables without constructing the full Fock-space state. | Bosonic tensor network docs | +| **LPMCE** | Low-Photon Monte Carlo Estimator — `BMCTE` subclass using sampled `p×p` permanents for fixed small photon number. | `rrc_bosonic_tensor_network.py` | +| **Effective support** | The size of a uniform distribution with the same Shannon entropy as a measured distribution; `N_eff = 2^H`. Not the same as the number of occupied states. | Entropy analysis receipts | +| **Observable estimation vs exact simulation** | Observable estimation computes marginals/statistics via sampling; exact simulation constructs and evolves the full state vector. The Research Stack bosonic photonic centrality tool operates in the former regime for large `N`. | `rrc_bosonic_tensor_network.py`, photonic stress tests | ## Core Formalism @@ -259,7 +265,7 @@ surface. | Term | Definition | |---|---| -| **SilverSight** | Standalone Lean repo at `/tmp/SilverSight` for schema/layout/wireformat/codec work. Separated from Research Stack to isolate the Core Semantics surface. 8-phase microstep pipeline designed in `6-Documentation/docs/plans/SilverSight_completion_pipeline.md`. | +| **SilverSight** | Standalone Lean repo at `/home/allaun/SilverSight` for schema/layout/wireformat/codec work. Separated from Research Stack to isolate the Core Semantics surface. 8-phase microstep pipeline designed in `6-Documentation/docs/plans/SilverSight_completion_pipeline.md`. | | **Product Schema** | Schema instance for `α × β` with `byteSize = byteSize α + byteSize β` and `wellFormed = wellFormed a && wellFormed b`. Phase 1.1 of SilverSight pipeline. | | **Product Wire Format** | Row-major encode/decode for pairs: `encode (a,b) = encode a ++ encode b`. Roundtrip theorem proven. Phase 1.2. | | **Product Layout Bridge** | Identity and rowMajor↔columnar bridges for product types. Phase 1.3. | diff --git a/CHANGELOG.md b/CHANGELOG.md index 7434a6fa..38436f84 100644 --- a/CHANGELOG.md +++ b/CHANGELOG.md @@ -1,5 +1,28 @@ # CHANGELOG +## 2026-06-21 — Bosonic Tensor Network Regime Discovery on neon-64gb + +### Bosonic Tensor Network (`rrc_bosonic_tensor_network.py`) + +- Ran beyond-Perceval photonic centrality stress tests on **neon-64gb** (18 ARM64 CPUs, 62 GiB RAM). +- Discovered a clean **two-regime computational phase transition** governed by photon count `p`: + - **Bosonic MC regime (`p ≤ 4`)**: Monte Carlo sampling over `p×p` permanents; true bosonic interference preserved. + - **Classical approximation regime (`p ≥ 5`)**: falls back to `distinguishable_k*` independent-photon product approximation; no permanent evaluation, no bosonic interference. +- Observed at `N = 2000`: + - `p = 3`: Hilbert dim ≈ 1.335×10⁹, H ≈ 10.12 bits, method `mc_permanent3` + - `p = 4`: Hilbert dim ≈ 6.687×10¹¹, H ≈ 10.80 bits, method `mc_permanent4` + - `p = 5`: Hilbert dim ≈ 2.680×10¹⁴, H ≈ 10.83 bits, method `distinguishable_k5` + - `p = 6`: Hilbert dim ≈ 8.956×10¹⁶, H ≈ 10.85 bits, method `distinguishable_k6` +- Entropy plateau (~10.8–10.9 bits for `p ≥ 4`) reflects **estimator saturation** under the classical approximation, not a physical Fock-space entropy limit. +- Corrected Hilbert-dimension formula in result summary from `C(N, p)` to `C(N+p-1, p)`. +- Complexity clarification: runtime is `O(shots × N) + O(shots × p!)` for bosonic MC; the `rng.choice(N, ...)` sampling term dominates at large `N`. The `p ≥ 5` fallback collapses to roughly `O(shots × p)`. +- Interpretation: the system is a **piecewise stochastic estimator over linear-optical observables**, not a full Fock-space simulator. It decouples computational cost from Hilbert-space cardinality by exploiting `p ≪ N`. +- Receipts generated on neon at `/tmp/rrc_bosonic_tensor_receipt.json`. + +### Glossary Additions + +- Added **BMCTE**, **SBOE**, **LPMCE**, **effective support**, and **observable estimation vs exact simulation** distinctions to `6-Documentation/docs/GLOSSARY.md`. + ## 2026-05-09 through 2026-05-19 — Lean Hardening, Storage Stack, Gates & Physics Expansion ### Infrastructure / Storage