blob: eb5281fb1d13aab9a7b06786f23a655f7c0b380a [file] [log] [blame]
 /* tests/test_numpy_vectorize.cpp -- auto-vectorize functions over NumPy array arguments Copyright (c) 2016 Wenzel Jakob All rights reserved. Use of this source code is governed by a BSD-style license that can be found in the LICENSE file. */ #include "pybind11_tests.h" #include #include double my_func(int x, float y, double z) { py::print("my_func(x:int={}, y:float={:.0f}, z:float={:.0f})"_s.format(x, y, z)); return (float) x*y*z; } TEST_SUBMODULE(numpy_vectorize, m) { try { py::module_::import("numpy"); } catch (...) { return; } // test_vectorize, test_docs, test_array_collapse // Vectorize all arguments of a function (though non-vector arguments are also allowed) m.def("vectorized_func", py::vectorize(my_func)); // Vectorize a lambda function with a capture object (e.g. to exclude some arguments from the vectorization) m.def("vectorized_func2", [](py::array_t x, py::array_t y, float z) { return py::vectorize([z](int x, float y) { return my_func(x, y, z); })(std::move(x), std::move(y)); }); // Vectorize a complex-valued function m.def("vectorized_func3", py::vectorize( [](std::complex c) { return c * std::complex(2.f); } )); // test_type_selection // NumPy function which only accepts specific data types // A lot of these no lints could be replaced with const refs, and probably should at some point. m.def("selective_func", [](const py::array_t &) { return "Int branch taken."; }); m.def("selective_func", [](const py::array_t &) { return "Float branch taken."; }); m.def("selective_func", [](const py::array_t, py::array::c_style> &) { return "Complex float branch taken."; }); // test_passthrough_arguments // Passthrough test: references and non-pod types should be automatically passed through (in the // function definition below, only `b`, `d`, and `g` are vectorized): struct NonPODClass { explicit NonPODClass(int v) : value{v} {} int value; }; py::class_(m, "NonPODClass") .def(py::init()) .def_readwrite("value", &NonPODClass::value); m.def("vec_passthrough", py::vectorize([](const double *a, double b, // Changing this broke things // NOLINTNEXTLINE(performance-unnecessary-value-param) py::array_t c, const int &d, int &e, NonPODClass f, const double g) { return *a + b + c.at(0) + d + e + f.value + g; })); // test_method_vectorization struct VectorizeTestClass { explicit VectorizeTestClass(int v) : value{v} {}; float method(int x, float y) const { return y + (float) (x + value); } int value = 0; }; py::class_ vtc(m, "VectorizeTestClass"); vtc .def(py::init()) .def_readwrite("value", &VectorizeTestClass::value); // Automatic vectorizing of methods vtc.def("method", py::vectorize(&VectorizeTestClass::method)); // test_trivial_broadcasting // Internal optimization test for whether the input is trivially broadcastable: py::enum_(m, "trivial") .value("f_trivial", py::detail::broadcast_trivial::f_trivial) .value("c_trivial", py::detail::broadcast_trivial::c_trivial) .value("non_trivial", py::detail::broadcast_trivial::non_trivial); m.def("vectorized_is_trivial", [](const py::array_t &arg1, const py::array_t &arg2, const py::array_t &arg3) { py::ssize_t ndim = 0; std::vector shape; std::array buffers{ {arg1.request(), arg2.request(), arg3.request()}}; return py::detail::broadcast(buffers, ndim, shape); }); m.def("add_to", py::vectorize([](NonPODClass& x, int a) { x.value += a; })); }