The Expiry Corner as a Logarithmic Clock
A boundary-level error law for Carr randomization of the American put, with every numerical claim re-derivable from the published verifier stack.
Reference boundary value B(1) = 0.80875056967553 ± 5e-14, cross-checked across two independent solver lineages.
A boundary-level error law for Carr randomization of the American put, working paper, August 2026.
Every numerical claim in the paper is re-derivable from the companion repository. The paper’s §9 is the authoritative manifest, mirrored in the repo README.
What’s in the repo
Kim integral-equation collocation as the reference method, an independent CN–LCP cross-check, the boundary-error-law campaign across τ = 1 … 2⁻¹⁸ and n = 32 … 4096, a separate Howard policy-iteration lineage for two-lineage comparison, and a method-independent exact-identity certification harness.
Also shipped: a 10⁻¹³-grade reference boundary table on a dyadic τ-grid over 28 octaves, a 1,350-rung ladder archive with per-rung diagnostics, six audited campaign reports released verbatim, and an erratum ledger recording all nine events.
Reproducibility
Python ≥ 3.10 with numpy, scipy and mpmath. Some scripts carry hardcoded data paths from the campaign machine.