quantum-learning

Not explicitly credited in this document · 2026-06-30; README also records 2026-07-06 update

Quantum & quantum-inspired retrieval / long-term-memory research — code, experiments,
benchmarks, literature, and cross-project analysis, consolidated into one repo
(2026-06-30, from sandbox/learning).

Framing: the native quantum parts (Grover, QKNN, QRAM, hardware) are a research

sandbox — they need a QPU and don't beat classical on a simulator. The parts with

near-term value are quantum-inspired classical algorithms (holographic/VSA memory,

complex/phase embeddings, hypergraph relations, MI-fusion). "Speedup" claims throughout

refer to operation-count scaling, not wall-clock.

Layout

Path	What it is
`quantum_rag_demo/`	Main system (~46k LOC, 581 tests). 7 retrieval methods, graph/HDC modules, hardware runners, benchmarks, `research_agent/` (arXiv→validate→JIRA), `ai_extensions/` (AI→QC).
`quantum_computing/`	Learning sandbox — fundamentals notebooks, Grover-for-vector-search, Maestro adapter.
`ai_quantum_symbiosis/`	Research markdown — AI-for-quantum error correction, control, circuit optimization, etc.
`quantum_rag_demo/docs/research/expansion_2026-06-30/`	Literature corpus (66 CSVs, ~7,195 ranked papers) + 2 synthesis reports. Agent-generated — spot-check citations.

Analysis documents (root)

Doc	Purpose
`QUANTUM_BASELINE_2026-06-30.md`	Frozen capability + measured-test baseline.
`QUANTUM_LEARNING_PATHWAY.md`	Evidence-tiered learning path (build Tier-1, read Tier-3).
`QUANTUM_x_CONDUIR_MEMORY_ROADMAP.md`	Pass-1: how this maps onto conduir-evolution's long-term memory.
`QUANTUM_x_CONDUIR_DEEP_ANALYSIS_2026-06-30.md`	Pass-2 (read this): corrected deep analysis + roadmap; capability mapping, integration seams, gaps.

A condensed version of the conduir mapping is also merged into
conduir-evolution/delivery/reviews/2026-06-30-quantum-inspired-memory/.

Setup

cd quantum_rag_demo
python3 -m venv quantum_env && source quantum_env/bin/activate
pip install -r requirements.txt qiskit-algorithms pytest pytest-asyncio
python -m pytest tests/ -q                 # core tests
cd research_agent && python -m pytest . -q  # research_agent tests (run from its own dir)

For credentials, copy quantum_rag_demo/.env.example → quantum_rag_demo/.env and fill in.

Status

2026-07-06 (Stage 0.1): rebuilt quantum_env; suite green — 552 pass / 4 skip / 0 fail

(556 collected across tests/ + research_agent/; FAISS tests excluded — faiss uninstalled,
segfaults on Apple Silicon). Fixed the 2 token_optimizer failures (added standard research
abbreviations: HRR, SOTA, RL, CNN, RNN, NLU, QA) and deflaked the wall-clock-based
test_ham_complexity_empirical (warmup + median + min-time floor). Stack: qiskit 2.5.0, numpy 2.5.0.

Baseline 2026-06-30: 581 pass / 2 fail (token_optimizer) / 4 skip; qiskit 2.4.2, qiskit-aer 0.17.2.
Venvs and .env are gitignored — recreate locally.

See QUANTUM_BASELINE_2026-06-30.md for the full inventory and known issues.

quantum-learning