xtensor-interpolate

xtensor-interpolate is an unofficial extension to the xtensor library offering spline interpolation via internal bindings to Paul Dierckx's Fortran package, FITPACK.

Introduction

The FITPACK library is used in SciPy to build the back-end of scipy.interpolate;xtensor-interpolate aims to provide a familiar API to anyone experienced with this Python package.

Dependencies

• xtensor

If you would like to build the test suite, you will also need a copy of googletest. Optionally this can be downloaded and linked during the build step if you do not already have a copy.

If compiling on Windows you will also need a complete MinGW development environment. This is required to compile the bundled Fortran component with gfortran.

Installation

From Source

Installation from source is streamlined with CMake.

mkdir build
cd build
cmake ..
make install

The cmake step of the process can be customized with a number of flags.

• -DCMAKE_INSTALLATION_PREFIX=... to change where you want xtensor-interpolate to install to.

• -DBUILD_TESTS=ON to build the unit test suite. This requires googletest.

  • You can target the unit tests directly with make xtest, or just build the executable with make test_xtensor_interpolate.

• -DDOWNLOAD_GTEST=ON to download and compile googletest as part of the build step. Note that googletest will not actually be installed to your system directories, and is just statically linked into the test executable.

  • Use Dgtest_disable_pthreads=ON to compile a single-threaded version of googletest. This is required when using Windows and MinGW.

Usage

Basic spline interpolation

#include <cmath>  // M_PI
#include "xtensor/xbuilder.hpp"  // linspace
#include "xtensor/xtensor.hpp"  // xtensor
#include "xtensor-interpolate/xinterpolate.hpp"  // splrep, splev

xt::xtensor<double, 1> x =
    { -3.141, -2.443, -1.745, -1.047, -0.349, 0.349,  1.047,  1.745,  2.443,  3.141 };

xt::xtensor<double, 1> y =
    { 0., -0.643, -0.985, -0.866, -0.342,  0.342,  0.866,  0.985, 0.643,  0. };

// Find the B-spline representation of the above points.

auto tck = xt::interpolate::splrep(x, y);

// Evaluate the B-spline over a new range of data points.

xt::xtensor<double, 1> xs = xt::linspace<double>(-M_PI, M_PI, 100);

xt::xtensor<double, 1> ys = xt::interpolate::splev(xs, tck);

Plotted, this data looks like:

License

This software is licensed under the BSD-2-Clause license. See the LICENSE file for details.