Grazing endpoint continuation within the error budget

← Verification · Cylinders · Point dipole
Reference
grazing endpoint continuation within the error budget
Kind
method: an independent formulation inside this repo (a scattering matrix, a plane-wave superposition, a quadrature): proves the formulation
Calculated
rates of lossy cylinders
Agreement
per tolerance
An electric dipole 5 nm from a 10 nm metal wire (n = 0.97 + 1.87i), asked for 1e-9: the grazing error it cannot pass The light-line caps leave an error the integration cannot shrink; asked for 1e-9, PyStratify charges it to the total error and reports the result as not converged (16 of 16 wavelengths), rather than claiming an accuracy it does not have. The grazing error is all but 0.3% of the reported error, so the two lines coincide. The same wire as the grazing-caps rows; 40 orders; no dispersion.
40045050055060065070010⁻⁷10⁻⁶10⁻⁵10⁻⁴λ [nm]error in Γ / Γ₀
grazing error charged, PyStratify total error reported, PyStratify 1e-9 × max Γ / Γ₀, the accuracy asked for
Hover to read values.
  • smallest grazing error / accuracy asked (must stay above 1): 11.3203
  • PyStratify the lines, the reference the circles

Run it yourself: unified_checks.py · test_unified.py · figures.py · UNIFIED_CHECKS.md on GitHub, PyStratify 97e52a9