Version 1.1.3

• Correction in device.cu to avoid an NVCC crash on Windows

• Correction and update in README

• Update of the default cache_dir value to improve cross-platform compatibility

• Correction in the extraction for the reptran auxdata .tar archive

Version 1.1.2

• Remove hitran-api

• Correction in MANIFEST.in file

• Require geoclide >= 3 and remove python maximum version constraint

• Correction in test_smartg_jax

• Cleaning before preparing a release for conda-forge

Version 1.1.1

• Corrections in AerOPAC and Cloud classes

  • Proper handling of the phase parameter
  • Docstring updated and corrected

• Correction of bugs while using RoughSurface for simulations in spherical shell geometry

  • Fix bug when using reflectance=false (for both BRDF=true and BRDF=False)
  • Fig bug with showdowing-masking function. Completely fixed for BRDF=true.
    Partially for BRDF=false, still very small bias for very high VZA values with
    sensor at TOA.

• Use of template for all geometric classes in the source CUDA code

• Remove an unused notebook

• Only use of pathlib to improve cross-platform compatibility

Version 1.1.0

• Smartg can be used with Jax (see demo_notebook)

• New way to download and consider auxiliary data

• Manual pycuda context init now possible

• Adaptation to numpy 2

• A pyproject.toml is now available i.e., installation with pip now possible

• The polarization can be desactivated by setting pol_off=True in smartg run method.

• The intern python geometric modules are now depracated, the geoclide package is used instead

  • See https://github.com/hygeos/geoclide
  • Several bugs are corrected
  • Best performance thanks to numpy
  • More functionalities are available

• Update of several notebooks

• Results with only photons not scattered by aerosols can be added to the output by setting
  no_aer_output=True in smartg run method

• The OUTPUT_LAYERS parameter of the smartg run method improved

  • more options
  • not only remove useless outputs but avoid also useless photon counting

• New key for local estimate intput dictionary

  • the key 'count_level' to choose if we want to count only a particular level (in plan parallel),
    see the doctring of the smartg run method

• The smartg 3D atm mode is now validated, but still in dev!!

  • A huge change (modules, classes, ...) is planned for this mode, so it is not recommended to use it for now.
  • The IPRT phase B C2 and C3 test cases have been validated
  • The notebook validation_SMARTG_IPRT_phaseB-C2.ipynb is available with examples but will be
    obselete in future release!
  • Only clouds in 3D for the moment
  • Only 1 type of cloud per simulation for the moment

• Several improvements

  • Rewrite the python docstring of several functions and classes
  • Some corrections
  • ...