编写自己的 ufunc#

我拥有力量!
希-曼

创建新的通用函数#

在阅读本节之前,建议您先熟悉 Python C 扩展的基础知识,方法是阅读/浏览 扩展和嵌入 Python 解释器 第 1 部分以及 如何扩展 NumPy 中的教程。

umath 模块是一个计算机生成的 C 模块,它创建了许多 ufunc。它提供了许多关于如何创建通用函数的示例。创建您自己的 ufunc 来利用 ufunc 机制也并不困难。假设您有一个函数,您希望对其输入逐元素进行操作。通过创建一个新的 ufunc,您将获得一个处理以下功能的函数:

  • 广播

  • N 维循环

  • 自动类型转换,并最大限度地减少内存使用

  • 可选输出数组

创建自己的 ufunc 并不难。所需做的只是为每个您想要支持的数据类型创建一个 1 维循环。每个 1 维循环必须具有特定的签名,并且只能使用固定大小数据类型的 ufunc。用于创建新的 ufunc 以处理内置数据类型的函数调用如下所示。用于注册用户定义数据类型的 ufunc 的机制有所不同。

在接下来的几个部分中,我们将提供一些示例代码,这些代码可以轻松修改以创建您自己的 ufunc。这些示例是 logit 函数(统计建模中常见的函数)的逐步完善或复杂的版本。logit 也很有趣,因为由于 IEEE 标准(特别是 IEEE 754)的魔力,下面创建的所有 logit 函数都自动具有以下行为。

>>> logit(0)
-inf
>>> logit(1)
inf
>>> logit(2)
nan
>>> logit(-2)
nan

这很棒,因为函数编写者无需手动传播 inf 或 nan。

非 ufunc 扩展示例#

为了进行比较并让读者全面了解,我们提供了一个简单的 logit C 扩展实现,该实现不使用 numpy。

为此,我们需要两个文件。第一个是包含实际代码的 C 文件,第二个是用于创建模块的 setup.py 文件。

#define PY_SSIZE_T_CLEAN
#include <Python.h>
#include <math.h>

/*
 * spammodule.c
 * This is the C code for a non-numpy Python extension to
 * define the logit function, where logit(p) = log(p/(1-p)).
 * This function will not work on numpy arrays automatically.
 * numpy.vectorize must be called in python to generate
 * a numpy-friendly function.
 *
 * Details explaining the Python-C API can be found under
 * 'Extending and Embedding' and 'Python/C API' at
 * docs.python.org .
 */


/* This declares the logit function */
static PyObject *spam_logit(PyObject *self, PyObject *args);

/*
 * This tells Python what methods this module has.
 * See the Python-C API for more information.
 */
static PyMethodDef SpamMethods[] = {
    {"logit",
        spam_logit,
        METH_VARARGS, "compute logit"},
    {NULL, NULL, 0, NULL}
};

/*
 * This actually defines the logit function for
 * input args from Python.
 */

static PyObject *spam_logit(PyObject *self, PyObject *args)
{
    double p;

    /* This parses the Python argument into a double */
    if(!PyArg_ParseTuple(args, "d", &p)) {
        return NULL;
    }

    /* THE ACTUAL LOGIT FUNCTION */
    p = p/(1-p);
    p = log(p);

    /*This builds the answer back into a python object */
    return Py_BuildValue("d", p);
}

/* This initiates the module using the above definitions. */
static struct PyModuleDef moduledef = {
    PyModuleDef_HEAD_INIT,
    "spam",
    NULL,
    -1,
    SpamMethods,
    NULL,
    NULL,
    NULL,
    NULL
};

PyMODINIT_FUNC PyInit_spam(void)
{
    PyObject *m;
    m = PyModule_Create(&moduledef);
    if (!m) {
        return NULL;
    }
    return m;
}

要使用 setup.py file,请将 setup.pyspammodule.c 放置在同一个文件夹中。然后 python setup.py build 将构建要导入的模块,或者 python setup.py install 将模块安装到您的 site-packages 目录中。

'''
    setup.py file for spammodule.c

    Calling
    $python setup.py build_ext --inplace
    will build the extension library in the current file.

    Calling
    $python setup.py build
    will build a file that looks like ./build/lib*, where
    lib* is a file that begins with lib. The library will
    be in this file and end with a C library extension,
    such as .so

    Calling
    $python setup.py install
    will install the module in your site-packages file.

    See the setuptools section 'Building Extension Modules'
    at setuptools.pypa.io for more information.
'''

from setuptools import setup, Extension
import numpy as np

module1 = Extension('spam', sources=['spammodule.c'])

setup(name='spam', version='1.0', ext_modules=[module1])

将 spam 模块导入到 python 后,您可以通过 spam.logit 调用 logit。请注意,上面使用的函数不能原样应用于 numpy 数组。为此,我们必须对其调用 numpy.vectorize。例如,如果在包含 spam 库的文件中打开了一个 python 解释器,或者 spam 已被安装,则可以执行以下命令:

>>> import numpy as np
>>> import spam
>>> spam.logit(0)
-inf
>>> spam.logit(1)
inf
>>> spam.logit(0.5)
0.0
>>> x = np.linspace(0,1,10)
>>> spam.logit(x)
TypeError: only length-1 arrays can be converted to Python scalars
>>> f = np.vectorize(spam.logit)
>>> f(x)
array([       -inf, -2.07944154, -1.25276297, -0.69314718, -0.22314355,
    0.22314355,  0.69314718,  1.25276297,  2.07944154,         inf])

生成的 LOGIT 函数速度不快!numpy.vectorize 只是循环遍历 spam.logit。循环在 C 级别完成,但 numpy 数组不断被解析和重建。这是很昂贵的。当作者将 numpy.vectorize(spam.logit) 与下面构建的 logit ufunc 进行比较时,logit ufunc 的速度几乎快了 4 倍。当然,具体的速度提升可能更大或更小,具体取决于函数的性质。

一个 dtype 的 NumPy ufunc 示例#

为简单起见,我们提供了一个单个 dtype 的 ufunc,即 'f8' double。与上一节一样,我们首先给出 .c 文件,然后给出用于创建包含 ufunc 的模块的 setup.py 文件。

对应于 ufunc 的实际计算的代码部分用 /\* BEGIN main ufunc computation \*//\* END main ufunc computation \*/ 标记。这两行之间的代码是创建您自己的 ufunc 时需要主要更改的部分。

#define PY_SSIZE_T_CLEAN
#include <Python.h>
#include "numpy/ndarraytypes.h"
#include "numpy/ufuncobject.h"
#include "numpy/npy_3kcompat.h"
#include <math.h>

/*
 * single_type_logit.c
 * This is the C code for creating your own
 * NumPy ufunc for a logit function.
 *
 * In this code we only define the ufunc for
 * a single dtype. The computations that must
 * be replaced to create a ufunc for
 * a different function are marked with BEGIN
 * and END.
 *
 * Details explaining the Python-C API can be found under
 * 'Extending and Embedding' and 'Python/C API' at
 * docs.python.org .
 */

static PyMethodDef LogitMethods[] = {
    {NULL, NULL, 0, NULL}
};

/* The loop definition must precede the PyMODINIT_FUNC. */

static void double_logit(char **args, const npy_intp *dimensions,
                         const npy_intp *steps, void *data)
{
    npy_intp i;
    npy_intp n = dimensions[0];
    char *in = args[0], *out = args[1];
    npy_intp in_step = steps[0], out_step = steps[1];

    double tmp;

    for (i = 0; i < n; i++) {
        /* BEGIN main ufunc computation */
        tmp = *(double *)in;
        tmp /= 1 - tmp;
        *((double *)out) = log(tmp);
        /* END main ufunc computation */

        in += in_step;
        out += out_step;
    }
}

/* This a pointer to the above function */
PyUFuncGenericFunction funcs[1] = {&double_logit};

/* These are the input and return dtypes of logit.*/
static const char types[2] = {NPY_DOUBLE, NPY_DOUBLE};

static struct PyModuleDef moduledef = {
    PyModuleDef_HEAD_INIT,
    "npufunc",
    NULL,
    -1,
    LogitMethods,
    NULL,
    NULL,
    NULL,
    NULL
};

PyMODINIT_FUNC PyInit_npufunc(void)
{
    PyObject *m, *logit, *d;

    import_array();
    import_umath();

    m = PyModule_Create(&moduledef);
    if (!m) {
        return NULL;
    }

    logit = PyUFunc_FromFuncAndData(funcs, NULL, types, 1, 1, 1,
                                    PyUFunc_None, "logit",
                                    "logit_docstring", 0);

    d = PyModule_GetDict(m);

    PyDict_SetItemString(d, "logit", logit);
    Py_DECREF(logit);

    return m;
}

这是一个用于上述代码的 setup.py file。与之前一样,可以通过在命令提示符下调用 python setup.py build 来构建模块,或者通过 python setup.py install 安装到 site-packages。模块也可以使用 python setup.py build_ext --inplace 放入本地文件夹(例如下面的 npufunc_directory)中。

'''
    setup.py file for single_type_logit.c
    Note that since this is a numpy extension
    we add an include_dirs=[get_include()] so that the
    extension is built with numpy's C/C++ header files.

    Calling
    $python setup.py build_ext --inplace
    will build the extension library in the npufunc_directory.

    Calling
    $python setup.py build
    will build a file that looks like ./build/lib*, where
    lib* is a file that begins with lib. The library will
    be in this file and end with a C library extension,
    such as .so

    Calling
    $python setup.py install
    will install the module in your site-packages file.

    See the setuptools section 'Building Extension Modules'
    at setuptools.pypa.io for more information.
'''

from setuptools import setup, Extension
from numpy import get_include

npufunc = Extension('npufunc',
                    sources=['single_type_logit.c'],
                    include_dirs=[get_include()])

setup(name='npufunc', version='1.0', ext_modules=[npufunc])

安装上述内容后,可以按如下方式导入和使用它。

>>> import numpy as np
>>> import npufunc
>>> npufunc.logit(0.5)
np.float64(0.0)
>>> a = np.linspace(0,1,5)
>>> npufunc.logit(a)
array([       -inf, -1.09861229,  0.        ,  1.09861229,         inf])

多个 dtype 的 NumPy ufunc 示例#

最后,我们提供了一个完整 ufunc 的示例,其中包含半浮点数、浮点数、双精度数和长双精度数的内部循环。与前面的部分一样,我们首先给出 .c 文件,然后给出相应的 setup.py 文件。

对应于 ufunc 的实际计算的代码部分用 /\* BEGIN main ufunc computation \*//\* END main ufunc computation \*/ 标记。这两行之间的代码是创建您自己的 ufunc 时需要主要更改的部分。

#define PY_SSIZE_T_CLEAN
#include <Python.h>
#include "numpy/ndarraytypes.h"
#include "numpy/ufuncobject.h"
#include "numpy/halffloat.h"
#include <math.h>

/*
 * multi_type_logit.c
 * This is the C code for creating your own
 * NumPy ufunc for a logit function.
 *
 * Each function of the form type_logit defines the
 * logit function for a different numpy dtype. Each
 * of these functions must be modified when you
 * create your own ufunc. The computations that must
 * be replaced to create a ufunc for
 * a different function are marked with BEGIN
 * and END.
 *
 * Details explaining the Python-C API can be found under
 * 'Extending and Embedding' and 'Python/C API' at
 * docs.python.org .
 *
 */

static PyMethodDef LogitMethods[] = {
    {NULL, NULL, 0, NULL}
};

/* The loop definitions must precede the PyMODINIT_FUNC. */

static void long_double_logit(char **args, const npy_intp *dimensions,
                              const npy_intp *steps, void *data)
{
    npy_intp i;
    npy_intp n = dimensions[0];
    char *in = args[0], *out = args[1];
    npy_intp in_step = steps[0], out_step = steps[1];

    long double tmp;

    for (i = 0; i < n; i++) {
        /* BEGIN main ufunc computation */
        tmp = *(long double *)in;
        tmp /= 1 - tmp;
        *((long double *)out) = logl(tmp);
        /* END main ufunc computation */

        in += in_step;
        out += out_step;
    }
}

static void double_logit(char **args, const npy_intp *dimensions,
                         const npy_intp *steps, void *data)
{
    npy_intp i;
    npy_intp n = dimensions[0];
    char *in = args[0], *out = args[1];
    npy_intp in_step = steps[0], out_step = steps[1];

    double tmp;

    for (i = 0; i < n; i++) {
        /* BEGIN main ufunc computation */
        tmp = *(double *)in;
        tmp /= 1 - tmp;
        *((double *)out) = log(tmp);
        /* END main ufunc computation */

        in += in_step;
        out += out_step;
    }
}

static void float_logit(char **args, const npy_intp *dimensions,
                       const npy_intp *steps, void *data)
{
    npy_intp i;
    npy_intp n = dimensions[0];
    char *in = args[0], *out = args[1];
    npy_intp in_step = steps[0], out_step = steps[1];

    float tmp;

    for (i = 0; i < n; i++) {
        /* BEGIN main ufunc computation */
        tmp = *(float *)in;
        tmp /= 1 - tmp;
        *((float *)out) = logf(tmp);
        /* END main ufunc computation */

        in += in_step;
        out += out_step;
    }
}


static void half_float_logit(char **args, const npy_intp *dimensions,
                            const npy_intp *steps, void *data)
{
    npy_intp i;
    npy_intp n = dimensions[0];
    char *in = args[0], *out = args[1];
    npy_intp in_step = steps[0], out_step = steps[1];

    float tmp;

    for (i = 0; i < n; i++) {

        /* BEGIN main ufunc computation */
        tmp = npy_half_to_float(*(npy_half *)in);
        tmp /= 1 - tmp;
        tmp = logf(tmp);
        *((npy_half *)out) = npy_float_to_half(tmp);
        /* END main ufunc computation */

        in += in_step;
        out += out_step;
    }
}


/*This gives pointers to the above functions*/
PyUFuncGenericFunction funcs[4] = {&half_float_logit,
                                   &float_logit,
                                   &double_logit,
                                   &long_double_logit};

static const char types[8] = {NPY_HALF, NPY_HALF,
                              NPY_FLOAT, NPY_FLOAT,
                              NPY_DOUBLE, NPY_DOUBLE,
                              NPY_LONGDOUBLE, NPY_LONGDOUBLE};

static struct PyModuleDef moduledef = {
    PyModuleDef_HEAD_INIT,
    "npufunc",
    NULL,
    -1,
    LogitMethods,
    NULL,
    NULL,
    NULL,
    NULL
};

PyMODINIT_FUNC PyInit_npufunc(void)
{
    PyObject *m, *logit, *d;

    import_array();
    import_umath();

    m = PyModule_Create(&moduledef);
    if (!m) {
        return NULL;
    }

    logit = PyUFunc_FromFuncAndData(funcs, NULL, types, 4, 1, 1,
                                    PyUFunc_None, "logit",
                                    "logit_docstring", 0);

    d = PyModule_GetDict(m);

    PyDict_SetItemString(d, "logit", logit);
    Py_DECREF(logit);

    return m;
}

这是一个用于上述代码的 setup.py file。与之前一样,可以通过在命令提示符下调用 python setup.py build 来构建模块,或者通过 python setup.py install 安装到 site-packages。

'''
    setup.py file for multi_type_logit.c
    Note that since this is a numpy extension
    we add an include_dirs=[get_include()] so that the
    extension is built with numpy's C/C++ header files.
    Furthermore, we also have to include the npymath
    lib for half-float d-type.

    Calling
    $python setup.py build_ext --inplace
    will build the extension library in the current file.

    Calling
    $python setup.py build
    will build a file that looks like ./build/lib*, where
    lib* is a file that begins with lib. The library will
    be in this file and end with a C library extension,
    such as .so

    Calling
    $python setup.py install
    will install the module in your site-packages file.

    See the setuptools section 'Building Extension Modules'
    at setuptools.pypa.io for more information.
'''

from setuptools import setup, Extension
from numpy import get_include
from os import path

path_to_npymath = path.join(get_include(), '..', 'lib')
npufunc = Extension('npufunc',
                    sources=['multi_type_logit.c'],
                    include_dirs=[get_include()],
                    # Necessary for the half-float d-type.
                    library_dirs=[path_to_npymath],
                    libraries=["npymath"])

setup(name='npufunc', version='1.0', ext_modules=[npufunc])

安装上述内容后,可以按如下方式导入和使用它。

>>> import numpy as np
>>> import npufunc
>>> npufunc.logit(0.5)
np.float64(0.0)
>>> a = np.linspace(0,1,5)
>>> npufunc.logit(a)
array([       -inf, -1.09861229,  0.        ,  1.09861229,         inf])

具有多个参数/返回值的 NumPy ufunc 示例#

我们的最后一个示例是一个具有多个参数的 ufunc。它是针对单个 dtype 数据的 logit ufunc 代码的修改版本。我们计算 (A * B, logit(A * B))

我们只给出 C 代码,因为 setup.py 文件与 一个 dtype 的 NumPy ufunc 示例 中的 setup.py 文件完全相同,只是将以下行:

npufunc = Extension('npufunc',
                    sources=['single_type_logit.c'],
                    include_dirs=[get_include()])

替换为:

npufunc = Extension('npufunc',
                    sources=['multi_arg_logit.c'],
                    include_dirs=[get_include()])

C 文件如下所示。生成的 ufunc 接受两个参数 AB。它返回一个元组,其第一个元素是 A * B,第二个元素是 logit(A * B)。请注意,它自动支持广播,以及 ufunc 的所有其他属性。

#define PY_SSIZE_T_CLEAN
#include <Python.h>
#include "numpy/ndarraytypes.h"
#include "numpy/ufuncobject.h"
#include "numpy/halffloat.h"
#include <math.h>

/*
 * multi_arg_logit.c
 * This is the C code for creating your own
 * NumPy ufunc for a multiple argument, multiple
 * return value ufunc. The places where the
 * ufunc computation is carried out are marked
 * with comments.
 *
 * Details explaining the Python-C API can be found under
 * 'Extending and Embedding' and 'Python/C API' at
 * docs.python.org.
 */

static PyMethodDef LogitMethods[] = {
    {NULL, NULL, 0, NULL}
};

/* The loop definition must precede the PyMODINIT_FUNC. */

static void double_logitprod(char **args, const npy_intp *dimensions,
                             const npy_intp *steps, void *data)
{
    npy_intp i;
    npy_intp n = dimensions[0];
    char *in1 = args[0], *in2 = args[1];
    char *out1 = args[2], *out2 = args[3];
    npy_intp in1_step = steps[0], in2_step = steps[1];
    npy_intp out1_step = steps[2], out2_step = steps[3];

    double tmp;

    for (i = 0; i < n; i++) {
        /* BEGIN main ufunc computation */
        tmp = *(double *)in1;
        tmp *= *(double *)in2;
        *((double *)out1) = tmp;
        *((double *)out2) = log(tmp / (1 - tmp));
        /* END main ufunc computation */

        in1 += in1_step;
        in2 += in2_step;
        out1 += out1_step;
        out2 += out2_step;
    }
}

/*This a pointer to the above function*/
PyUFuncGenericFunction funcs[1] = {&double_logitprod};

/* These are the input and return dtypes of logit.*/

static const char types[4] = {NPY_DOUBLE, NPY_DOUBLE,
                              NPY_DOUBLE, NPY_DOUBLE};

static struct PyModuleDef moduledef = {
    PyModuleDef_HEAD_INIT,
    "npufunc",
    NULL,
    -1,
    LogitMethods,
    NULL,
    NULL,
    NULL,
    NULL
};

PyMODINIT_FUNC PyInit_npufunc(void)
{
    PyObject *m, *logit, *d;

    import_array();
    import_umath();

    m = PyModule_Create(&moduledef);
    if (!m) {
        return NULL;
    }

    logit = PyUFunc_FromFuncAndData(funcs, NULL, types, 1, 2, 2,
                                    PyUFunc_None, "logit",
                                    "logit_docstring", 0);

    d = PyModule_GetDict(m);

    PyDict_SetItemString(d, "logit", logit);
    Py_DECREF(logit);

    return m;
}

具有结构化数组 dtype 参数的 NumPy ufunc 示例#

此示例演示了如何为结构化数组 dtype 创建 ufunc。对于示例,我们展示了一个用于添加两个 dtype 为 'u8,u8,u8' 的数组的简单 ufunc。该过程与其他示例略有不同,因为对 PyUFunc_FromFuncAndData 的调用不会完全注册自定义 dtype 和结构化数组 dtype 的 ufunc。我们需要再调用 PyUFunc_RegisterLoopForDescr 来完成 ufunc 的设置。

我们只给出 C 代码,因为 setup.py 文件与 一个 dtype 的 NumPy ufunc 示例 中的 setup.py 文件完全相同,只是将以下行:

npufunc = Extension('npufunc',
                    sources=['single_type_logit.c'],
                    include_dirs=[get_include()])

替换为:

npufunc = Extension('npufunc',
                    sources=['add_triplet.c'],
                    include_dirs=[get_include()])

C 文件如下所示。

#define PY_SSIZE_T_CLEAN
#include <Python.h>
#include "numpy/ndarraytypes.h"
#include "numpy/ufuncobject.h"
#include "numpy/npy_3kcompat.h"
#include <math.h>

/*
 * add_triplet.c
 * This is the C code for creating your own
 * NumPy ufunc for a structured array dtype.
 *
 * Details explaining the Python-C API can be found under
 * 'Extending and Embedding' and 'Python/C API' at
 * docs.python.org.
 */

static PyMethodDef StructUfuncTestMethods[] = {
    {NULL, NULL, 0, NULL}
};

/* The loop definition must precede the PyMODINIT_FUNC. */

static void add_uint64_triplet(char **args, const npy_intp *dimensions,
                               const npy_intp *steps, void *data)
{
    npy_intp i;
    npy_intp is1 = steps[0];
    npy_intp is2 = steps[1];
    npy_intp os = steps[2];
    npy_intp n = dimensions[0];
    uint64_t *x, *y, *z;

    char *i1 = args[0];
    char *i2 = args[1];
    char *op = args[2];

    for (i = 0; i < n; i++) {

        x = (uint64_t *)i1;
        y = (uint64_t *)i2;
        z = (uint64_t *)op;

        z[0] = x[0] + y[0];
        z[1] = x[1] + y[1];
        z[2] = x[2] + y[2];

        i1 += is1;
        i2 += is2;
        op += os;
    }
}

/* This a pointer to the above function */
PyUFuncGenericFunction funcs[1] = {&add_uint64_triplet};

/* These are the input and return dtypes of add_uint64_triplet. */
static const char types[3] = {NPY_UINT64, NPY_UINT64, NPY_UINT64};

static struct PyModuleDef moduledef = {
    PyModuleDef_HEAD_INIT,
    "struct_ufunc_test",
    NULL,
    -1,
    StructUfuncTestMethods,
    NULL,
    NULL,
    NULL,
    NULL
};

PyMODINIT_FUNC PyInit_npufunc(void)
{
    PyObject *m, *add_triplet, *d;
    PyObject *dtype_dict;
    PyArray_Descr *dtype;
    PyArray_Descr *dtypes[3];

    import_array();
    import_umath();

    m = PyModule_Create(&moduledef);
    if (m == NULL) {
        return NULL;
    }

    /* Create a new ufunc object */
    add_triplet = PyUFunc_FromFuncAndData(NULL, NULL, NULL, 0, 2, 1,
                                          PyUFunc_None, "add_triplet",
                                          "add_triplet_docstring", 0);

    dtype_dict = Py_BuildValue("[(s, s), (s, s), (s, s)]",
                               "f0", "u8", "f1", "u8", "f2", "u8");
    PyArray_DescrConverter(dtype_dict, &dtype);
    Py_DECREF(dtype_dict);

    dtypes[0] = dtype;
    dtypes[1] = dtype;
    dtypes[2] = dtype;

    /* Register ufunc for structured dtype */
    PyUFunc_RegisterLoopForDescr(add_triplet,
                                 dtype,
                                 &add_uint64_triplet,
                                 dtypes,
                                 NULL);

    d = PyModule_GetDict(m);

    PyDict_SetItemString(d, "add_triplet", add_triplet);
    Py_DECREF(add_triplet);
    return m;
}

返回的 ufunc 对象是一个可调用的 Python 对象。它应该放在(模块)字典中,名称与 ufunc 创建例程中 name 参数使用的名称相同。以下示例改编自 umath 模块

static PyUFuncGenericFunction atan2_functions[] = {
                      PyUFunc_ff_f, PyUFunc_dd_d,
                      PyUFunc_gg_g, PyUFunc_OO_O_method};
static void *atan2_data[] = {
                      (void *)atan2f, (void *)atan2,
                      (void *)atan2l, (void *)"arctan2"};
static const char atan2_signatures[] = {
              NPY_FLOAT, NPY_FLOAT, NPY_FLOAT,
              NPY_DOUBLE, NPY_DOUBLE, NPY_DOUBLE,
              NPY_LONGDOUBLE, NPY_LONGDOUBLE, NPY_LONGDOUBLE
              NPY_OBJECT, NPY_OBJECT, NPY_OBJECT};
...
/* in the module initialization code */
PyObject *f, *dict, *module;
...
dict = PyModule_GetDict(module);
...
f = PyUFunc_FromFuncAndData(atan2_functions,
    atan2_data, atan2_signatures, 4, 2, 1,
    PyUFunc_None, "arctan2",
    "a safe and correct arctan(x1/x2)", 0);
PyDict_SetItemString(dict, "arctan2", f);
Py_DECREF(f);
...