{"key": "numba-1", "gates_match": true, "gates": {"10000": 443.66352, "10001": 571.977293, "10002": 526.411631}, "holdout_identity": {"n": 64, "differ": 0, "max_abs_diff": 2.739284354902338e-10}, "seconds_per_run": {"pure": 14.4, "numba": 0.1445}, "speedup": 99.5, "code_zb64": "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", "code_sha256": "cedaab193c345ae445d0a656a00e0fc8bcf01d7233c44f47bc06459add19bebe", "notes": "Install: pip install numpy numba (tested numba 0.68.1, numpy 2.5.4, py3.12); put fast_replica.py beside run.py. Call: python3 -I run.py fast_replica:Lattice 10000 3 out.json qm=qm_k1e-5.json (Gauss: fast_replica:Gauss soft=1 qm=..); python3 -I fast_replica.py perfect|gauss|lattice seed0 n out.json for the perfect-info column and run.py-format rows. Importing it replaces replica2.run by a jitted version only for Blind(Lattice|Gauss) and bare DPPolicy; other policies, options fill/oraclearb and FAST=0 or no numba use the original pure Python (verified: 1 seed on system python3 w/o numpy). Verified to 6 decimals on seeds 10000-10002 for all four Gates columns; 64 holdout seeds vs ref/lattice_qm.json: 0 differ, max 2.7e-10; 32 seeds 60000-60031 pure vs numba identical (max 0.0). Timing single process: pure 460s/32 runs on this slow box (14.4s/run), numba 4.62s/32 excl. 2s import. Stayed Python: random streams are bit-identical (numpy MT19937 loaded from random.Random state, CPython gauss/normalvariate re-implemented, self-checked at first use, else falls back to random.Random). Pitfalls: first call compiles ~75-100s (cache=True, so warm it with PROCS=1 before PROCS=8, else workers compile at once); x**2 became x*x with no digit change; float op order kept as in the Python."}