cmbant/CAMB: v2.0.3

Résumé fourni par la source

CAMB v2.0.3 This release rolls up fixes, performance improvements, and compatibility updates since v2.0.0. Highlights Neutrinos set_cosmology(mnu=...) now treats mnu as the physical sum of neutrino masses and computes Omega_nu h^2 from the actual thermal density, rather than the non-relativistic mass-density approximation. Added omnuh2_active as an alternative way to specify the active-neutrino density directly. Improved mass/density round-tripping, including very light neutrinos and models close to the minimum inverted-hierarchy mass. standard_neutrino_neff is now applied consistently to neutrino eigenstate degeneracies and mass splitting. Fixed higher-order initial conditions for the neutrino-velocity isocurvature mode. Non-linear matter power and lensing Faster HMCode/Halofit calculations through BAO-wiggle reuse, batched non-linear redshift evaluation, caching, and improved parallelization. Typical non-linear matter-power calculations can be up to ~15% faster. More efficient redshift sampling for non-linear lensing calculations. Improved consistency and robustness of the shared non-linear matter-power redshift grid. Fixed the non-linear 21cm calculation setup. Fixed halofit_casarini and made several Halofit root-finding and interpolation cleanups. Improved consistency of lensed gradient spectra and lmax handling. Python, build and compatibility Restored support for Python 3.10. Added experimental support for the LLVM flang Fortran compiler. Added support for the CAMB_BUILD_DIR environment variable. Fixed several camb check_accuracy, keyword-only argument, and .ini round-trip issues. Various compiler-compatibility, OpenMP, packaging, CI, documentation, and code-maintenance improvements. Compatibility note The updated neutrino mass-density mapping causes very small changes in omnuh2 at fixed mnu, especially for very light neutrinos. To reproduce a specified active-neutrino density directly, use omnuh2_active with mnu=None. Full changelog: v2.0.0...v2.0.3

Ce résumé expose les affirmations des auteurs. BNTIC ne l’interprète pas comme une validation indépendante des résultats.