Resources#
Use reproducible inputs and independent numerical checks to assess a study.
Set up a development environment, check packages, and understand documentation versions.
Compare spatial resolution, angular integration, ensemble sampling, and finite-size effects.
Reproducibility#
Record the BornSim version, vacuum wavelength, background refractive index, fluctuation statistics, material geometry, grid, angular quadrature, Born order, and random seeds. Retain ensemble sample counts and standard errors alongside the coefficients. Result archives retain settings and amplitudes, but do not contain the original voxel field; keep that field or its seeded construction separately.
Decreasing Born terms alone do not establish convergence. Use the Numerical validation examples to separate numerical errors and sampling uncertainty, and consult Theory and numerical methods for model limitations.