Documentation

LeanMlir.Proofs.Certificates.SmoothingDecChunk4

Decimal-radius scorecard, scan chunk 4/6 (panels 1650…2200) #

GENERATED by scripts/smooth_dec_scorecard_gen.py — see SmoothingDecScorecard.lean for the corpus story. Each chunk lives in its own module because (a) one whole-grid kernel evaluation retains ~15 GB of kernel-cache bignums (OOM on 16 GB CI runners) and (b) memory is NOT reclaimed between declarations within one lean process — per-module processes cap the worst chunk at ~5 GB.

Grid values 2200…1650 (descending) as NUMERATORS over the common denominator 10¹² (every grid value is 1/2 + Σ (1/1000)·(k/10⁹)). ℕ literals elaborate in ~1 s where flat division literals price out in pending-mvar instance synthesis (>10 min). Untrusted input — verified by the kernel checks below.

Equations
  • One or more equations did not get rendered due to their size.
Instances For
    Equations
    Instances For
      theorem Proofs.phiCp1650 :
      phiGridUB (1 / 1000) 1650 = 950676870825 / 1000000000000

      Checkpoint: the grid value at panel 1650, read off the verified scan.

      theorem Proofs.phiChunkEq4 :
      phiScanRevFrom (1 / 1000) 1650 (950676870825 / 1000000000000) 550 = phiChunkLit4

      Chunk 4: panels 1650…2200, kernel-evaluated FROM the checkpoint.

      Panels 0…2200: chunk 4 glued onto the verified prefix.