[WIP] Polynomial python wrappers for boost polynomials.
Exposed polynomials:
Legendre
Laguerre
Hermite
Chebyshev
LegendreStieltjes
Every polynomials has additional wrappers for:
<Polynomial>Sequence
- polynomial value generator at a specified abscissa valuex
up to a specifiedorder
<Polynomial>Series
- dense 1D polynomial series with mutable coefficients<Polynomial>Product
- N-dimensional polynomial product<Polynomial>ProductSet
- list of N-dimensional weighed polynomial products
Python only wrappers:
<Polynomial>PCE
- polynomial chaos expansion fitting and prediction, can be used with anysklearn
linear model thoughsklearn.linear_model.Lasso
works bestAdaptivePCE
- adaptive version of polynomial chaos expansion using difference of Sobol' indices or variances between successive fits as convergence metric
Cmake, pybind11 and boost are required for building. Tested with pybind11 2.4.3 and boost 1.73.0 using MSVC 19.27.28826.0 but should compile with GCC and Clang.
To build:
python setup.py build
Additional cmake options can be passed as
python setup.py build --cmake-options="-DCMAKE_CXX_COMPILER=/usr/bin/g++"
or in setup.user.cfg
:
[build_ext]
vcpkg_dir = ~/vcpkg
vcpkg_triplet =
vcpkg_manifest = true
cxx_compiler = /usr/bin/g++
cmake_generator = Ninja
cmake_options =
python setup.py install --user
This project has been set up using PyScaffold 3.2.3. For details and usage information on PyScaffold see https://pyscaffold.org/.