Numpy Ipynb

{
"cells": [
{
"cell_type": "markdown",
"metadata": {},
"source": [
"# NumPy"
]
},
{
"metadata": {},
"source": [
"Numpy introduction\n",
"------------------\n",
"\n",
"The NumPy package (read as NUMerical PYthon) provides access to\n",
"\n",
"- a new data structure called àrray`s which allow\n",
"\n",
"- efficient vector and matrix operations. It also provides\n",
"\n",
"- a number of linear algebra operations (such as solving of systems of
linear equations, computation of Eigenvectors and Eigenvalues).\n",
"\n",
"### History\n",
"\n",
"Some background information: There are two other implementations that provide
nearly the same functionality as NumPy. These are called “Numeric” and “numarray”:\
n",
"\n",
"- Numeric was the first provision of a set of numerical methods (similar to
Matlab) for Python. It evolved from a PhD project.\n",
"\n",
"- Numarray is a re-implementation of Numeric with certain improvements (but
for our purposes both Numeric and Numarray behave virtually identical).\n",
"\n",
"- Early in 2006 it was decided to merge the best aspects of Numeric and
Numarray into the Scientific Python (<span>`scipy`</span>) package and to provide
(a hopefully “final”) àrray` data type under the module name “NumPy”.\n",
"\n",
"We will use in the following materials the “NumPy” package as provided by
(new) SciPy. If for some reason this doesn’t work for you, chances are that your
SciPy is too old. In that case, you will find that either “Numeric” or “numarray”
is installed and should provide nearly the same capabilities.[5]\n",
"\n",
"### Arrays\n",
"\n",
"We introduce a new data type (provided by NumPy) which is called “àrray`”. An
array *appears* to be very similar to a list but an array can keep only elements of
the same type (whereas a list can mix different kinds of objects). This means
arrays are more efficient to store (because we don’t need to store the type for
every element). It also makes arrays the data structure of choice for numerical
calculations where we often deal with vectors and matricies.\n",
"\n",
"Vectors and matrices (and matrices with more than two indices) are all called
“arrays” in NumPy.\n",
"\n",
"#### Vectors (1d-arrays)\n",
"\n",
"The data structure we will need most often is a vector. Here are a few
examples of how we can generate one:"
]
},
{
"metadata": {},
"source": [
"# Array Creation and Properties"
]
},
{
"metadata": {},
"source": [
"There are a lot of ways to create arrays. Let's look at a few"
]
},
{
"metadata": {},
"source": [
"Here we create an array using àrange` and then change its shape to be 3 rows
and 5 columns."
]
},
{
"cell_type": "code",
"execution_count": 1,
"metadata": {},
"outputs": [],
"source": [
"import numpy as np"
]
},
{
"metadata": {},
"outputs": [],
"source": [
"a = np.arange(15)\n",
"b = np.arange(20)\n"
]
},
{
"metadata": {},
"outputs": [
{
"data": {
"text/plain": [
"array([ 0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16,\
n",
" 17, 18, 19])"
]
},
"metadata": {},
"output_type": "execute_result"
}
],
"source": [
"a\n",
"b"
]
},
{
"metadata": {},
"outputs": [
{
"data": {
"text/plain": [
"array([[ 0, 1, 2, 3, 4],\n",
" [ 5, 6, 7, 8, 9],\n",
" [10, 11, 12, 13, 14]])"
]
},
"metadata": {},
}
],
"source": [
"a = np.arange(15).reshape(3,5)\n",
"a"
]
},
{
"metadata": {},
"outputs": [
{
"data": {
"text/plain": [
"array([[ 0, 1, 2, 3, 4],\n",
" [ 5, 6, 7, 8, 9],\n",
" [10, 11, 12, 13, 14]])"
]
},
"metadata": {},
}
],
"source": [
"a"
]
},
{
"metadata": {},
"outputs": [
{
"data": {
"text/plain": [
"array([[ 0, 1, 2, 3, 4],\n",
" [ 5, 6, 7, 8, 9],\n",
" [10, 11, 12, 13, 14]])"
]
},
"metadata": {},
}
],
"source": [
"a"
]
},
{
"metadata": {},
"source": [
"A NumPy array has a lot of meta-data associated with it describing its shape,
datatype, etc."
]
},
{
"metadata": {},
"outputs": [
{
"name": "stdout",
"output_type": "stream",
"text": [
"2\n",
"(3, 5)\n",
"15\n",
"int32\n",
"4\n",
"<class 'numpy.ndarray'>\n"
]
}
],
"source": [
"print(a.ndim)\n",
"print(a.shape)\n",
"print(a.size)\n",
"print(a.dtype)\n",
"print(a.itemsize)\n",
"print(type(a))"
]
},
{
"metadata": {},
"outputs": [
{
"name": "stdout",
"text": [
"Help on ndarray object:\n",
"\n",
"class ndarray(builtins.object)\n",
" | ndarray(shape, dtype=float, buffer=None, offset=0,\n",
" | strides=None, order=None)\n",
" | \n",
" | An array object represents a multidimensional, homogeneous array\n",
" | of fixed-size items. An associated data-type object describes the\n",
" | format of each element in the array (its byte-order, how many bytes it\
n",
" | occupies in memory, whether it is an integer, a floating point number,\
n",
" | or something else, etc.)\n",
" | \n",
" | Arrays should be constructed using àrray`, `zeros` or èmpty` (refer\
n",
" | to the See Also section below). The parameters given here refer to\n",
" | a low-level method (`ndarray(...)`) for instantiating an array.\n",
" | \n",
" | For more information, refer to the `numpy` module and examine the\n",
" | methods and attributes of an array.\n",
" | \n",
" | Parameters\n",
" | ----------\n",
" | (for the __new__ method; see Notes below)\n",
" | \n",
" | shape : tuple of ints\n",
" | Shape of created array.\n",
" | dtype : data-type, optional\n",
" | Any object that can be interpreted as a numpy data type.\n",
" | buffer : object exposing buffer interface, optional\n",
" | Used to fill the array with data.\n",
" | offset : int, optional\n",
" | Offset of array data in buffer.\n",
" | strides : tuple of ints, optional\n",
" | Strides of data in memory.\n",
" | order : {'C', 'F'}, optional\n",
" | Row-major (C-style) or column-major (Fortran-style) order.\n",
" | \n",
" | Attributes\n",
" | ----------\n",
" | T : ndarray\n",
" | Transpose of the array.\n",
" | data : buffer\n",
" | The array's elements, in memory.\n",
" | dtype : dtype object\n",
" | Describes the format of the elements in the array.\n",
" | flags : dict\n",
" | Dictionary containing information related to memory use, e.g.,\n",
" | 'C_CONTIGUOUS', 'OWNDATA', 'WRITEABLE', etc.\n",
" | flat : numpy.flatiter object\n",
" | Flattened version of the array as an iterator. The iterator\n",
" | allows assignments, e.g., ``x.flat = 3`` (See `ndarray.flat` for\n",
" | assignment examples; TODO).\n",
" | imag : ndarray\n",
" | Imaginary part of the array.\n",
" | real : ndarray\n",
" | Real part of the array.\n",
" | size : int\n",
" | Number of elements in the array.\n",
" | itemsize : int\n",
" | The memory use of each array element in bytes.\n",
" | nbytes : int\n",
" | The total number of bytes required to store the array data,\n",
" | i.e., `ìtemsize * size``.\n",
" | ndim : int\n",
" | The array's number of dimensions.\n",
" | Shape of the array.\n",
" | strides : tuple of ints\n",
" | The step-size required to move from one element to the next in\n",
" | memory. For example, a contiguous ``(3, 4)`` array of type\n",
" | `ìnt16`` in C-order has strides ``(8, 2)``. This implies that\n",
" | to move from element to element in memory requires jumps of 2
bytes.\n",
" | To move from row-to-row, one needs to jump 8 bytes at a time\n",
" | (``2 * 4``).\n",
" | ctypes : ctypes object\n",
" | Class containing properties of the array needed for interaction\n",
" | with ctypes.\n",
" | base : ndarray\n",
" | If the array is a view into another array, that array is its `base`\
n",
" | (unless that array is also a view). The `base` array is where the\
n",
" | array data is actually stored.\n",
" | \n",
" | See Also\n",
" | --------\n",
" | array : Construct an array.\n",
" | zeros : Create an array, each element of which is zero.\n",
" | empty : Create an array, but leave its allocated memory unchanged
(i.e.,\n",
" | it contains \"garbage\").\n",
" | dtype : Create a data-type.\n",
" | numpy.typing.NDArray : An ndarray alias :term:`generic <generic type>`\
n",
" | w.r.t. its `dtype.type <numpy.dtype.type>`.\n",
" | \n",
" | Notes\n",
" | -----\n",
" | There are two modes of creating an array using ``__new__``:\n",
" | \n",
" | 1. If `buffer` is None, then only `shape`, `dtype`, and òrder`\n",
" | are used.\n",
" | 2. If `buffer` is an object exposing the buffer interface, then\n",
" | all keywords are interpreted.\n",
" | \n",
" | No ``__init__`` method is needed because the array is fully initialized\
n",
" | after the ``__new__`` method.\n",
" | \n",
" | Examples\n",
" | --------\n",
" | These examples illustrate the low-level `ndarray` constructor. Refer\
n",
" | to the `See Also` section above for easier ways of constructing an\n",
" | ndarray.\n",
" | \n",
" | First mode, `buffer` is None:\n",
" | \n",
" | >>> np.ndarray(shape=(2,2), dtype=float, order='F')\n",
" | array([[0.0e+000, 0.0e+000], # random\n",
" | [ nan, 2.5e-323]])\n",
" | \n",
" | Second mode:\n",
" | \n",
" | >>> np.ndarray((2,), buffer=np.array([1,2,3]),\n",
" | ... offset=np.int_().itemsize,\n",
" | ... dtype=int) # offset = 1*itemsize, i.e. skip first
element\n",
" | array([2, 3])\n",
" | \n",
" | Methods defined here:\n",
" | \n",
" | __abs__(self, /)\n",
" | abs(self)\n",
" | \n",
" | __add__(self, value, /)\n",
" | Return self+value.\n",
" | \n",
" | __and__(self, value, /)\n",
" | Return self&value.\n",
" | \n",
" | __array__(...)\n",
" | a.__array__([dtype], /) -> reference if type unchanged, copy
otherwise.\n",
" | \n",
" | Returns either a new reference to self if dtype is not given or a
new array\n",
" | of provided data type if dtype is different from the current dtype
of the\n",
" | array.\n",
" | \n",
" | __array_finalize__(...)\n",
" | a.__array_finalize__(obj, /)\n",
" | \n",
" | Present so subclasses can call super. Does nothing.\n",
" | \n",
" | __array_function__(...)\n",
" | \n",
" | __array_prepare__(...)\n",
" | a.__array_prepare__(array[, context], /)\n",
" | \n",
" | Returns a view of àrray` with the same type as self.\n",
" | \n",
" | __array_ufunc__(...)\n",
" | \n",
" | __array_wrap__(...)\n",
" | a.__array_wrap__(array[, context], /)\n",
" | \n",
" | \n",
" | __bool__(self, /)\n",
" | True if self else False\n",
" | \n",
" | __complex__(...)\n",
" | \n",
" | __contains__(self, key, /)\n",
" | Return key in self.\n",
" | \n",
" | __copy__(...)\n",
" | a.__copy__()\n",
" | \n",
" | Used if :func:`copy.copy` is called on an array. Returns a copy of
the array.\n",
" | \n",
" | Equivalent to `à.copy(order='K')``.\n",
" | \n",
" | __deepcopy__(...)\n",
" | a.__deepcopy__(memo, /) -> Deep copy of array.\n",
" | \n",
" | Used if :func:`copy.deepcopy` is called on an array.\n",
" | \n",
" | __delitem__(self, key, /)\n",
" | Delete self[key].\n",
" | \n",
" | __divmod__(self, value, /)\n",
" | Return divmod(self, value).\n",
" | \n",
" | __dlpack__(...)\n",
" | a.__dlpack__(*, stream=None)\n",
" | \n",
" | DLPack Protocol: Part of the Array API.\n",
" | \n",
" | __dlpack_device__(...)\n",
" | a.__dlpack_device__()\n",
" | \n",
" | \n",
" | __eq__(self, value, /)\n",
" | Return self==value.\n",
" | \n",
" | __float__(self, /)\n",
" | float(self)\n",
" | \n",
" | __floordiv__(self, value, /)\n",
" | Return self//value.\n",
" | \n",
" | __format__(...)\n",
" | Default object formatter.\n",
" | \n",
" | __ge__(self, value, /)\n",
" | Return self>=value.\n",
" | \n",
" | __getitem__(self, key, /)\n",
" | Return self[key].\n",
" | \n",
" | __gt__(self, value, /)\n",
" | Return self>value.\n",
" | \n",
" | __iadd__(self, value, /)\n",
" | Return self+=value.\n",
" | \n",
" | __iand__(self, value, /)\n",
" | Return self&=value.\n",
" | \n",
" | __ifloordiv__(self, value, /)\n",
" | Return self//=value.\n",
" | \n",
" | __ilshift__(self, value, /)\n",
" | Return self<<=value.\n",
" | \n",
" | __imatmul__(self, value, /)\n",
" | Return self@=value.\n",
" | \n",
" | __imod__(self, value, /)\n",
" | Return self%=value.\n",
" | \n",
" | __imul__(self, value, /)\n",
" | Return self*=value.\n",
" | \n",
" | __index__(self, /)\n",
" | Return self converted to an integer, if self is suitable for use as
an index into a list.\n",
" | \n",
" | __int__(self, /)\n",
" | int(self)\n",
" | \n",
" | __invert__(self, /)\n",
" | ~self\n",
" | \n",
" | __ior__(self, value, /)\n",
" | Return self|=value.\n",
" | \n",
" | __ipow__(self, value, /)\n",
" | Return self**=value.\n",
" | \n",
" | __irshift__(self, value, /)\n",
" | Return self>>=value.\n",
" | \n",
" | __isub__(self, value, /)\n",
" | Return self-=value.\n",
" | \n",
" | __iter__(self, /)\n",
" | Implement iter(self).\n",
" | \n",
" | __itruediv__(self, value, /)\n",
" | Return self/=value.\n",
" | \n",
" | __ixor__(self, value, /)\n",
" | Return self^=value.\n",
" | \n",
" | __le__(self, value, /)\n",
" | Return self<=value.\n",
" | \n",
" | __len__(self, /)\n",
" | Return len(self).\n",
" | \n",
" | __lshift__(self, value, /)\n",
" | Return self<<value.\n",
" | \n",
" | __lt__(self, value, /)\n",
" | Return self<value.\n",
" | \n",
" | __matmul__(self, value, /)\n",
" | Return self@value.\n",
" | \n",
" | __mod__(self, value, /)\n",
" | Return self%value.\n",
" | \n",
" | __mul__(self, value, /)\n",
" | Return self*value.\n",
" | \n",
" | __ne__(self, value, /)\n",
" | Return self!=value.\n",
" | \n",
" | __neg__(self, /)\n",
" | -self\n",
" | \n",
" | __or__(self, value, /)\n",
" | Return self|value.\n",
" | \n",
" | __pos__(self, /)\n",
" | +self\n",
" | \n",
" | __pow__(self, value, mod=None, /)\n",
" | Return pow(self, value, mod).\n",
" | \n",
" | __radd__(self, value, /)\n",
" | Return value+self.\n",
" | \n",
" | __rand__(self, value, /)\n",
" | Return value&self.\n",
" | \n",
" | __rdivmod__(self, value, /)\n",
" | Return divmod(value, self).\n",
" | \n",
" | __reduce__(...)\n",
" | a.__reduce__()\n",
" | \n",
" | For pickling.\n",
" | \n",
" | __reduce_ex__(...)\n",
" | Helper for pickle.\n",
" | \n",
" | __repr__(self, /)\n",
" | Return repr(self).\n",
" | \n",
" | __rfloordiv__(self, value, /)\n",
" | Return value//self.\n",
" | \n",
" | __rlshift__(self, value, /)\n",
" | Return value<<self.\n",
" | \n",
" | __rmatmul__(self, value, /)\n",
" | Return value@self.\n",
" | \n",
" | __rmod__(self, value, /)\n",
" | Return value%self.\n",
" | \n",
" | __rmul__(self, value, /)\n",
" | Return value*self.\n",
" | \n",
" | __ror__(self, value, /)\n",
" | Return value|self.\n",
" | \n",
" | __rpow__(self, value, mod=None, /)\n",
" | Return pow(value, self, mod).\n",
" | \n",
" | __rrshift__(self, value, /)\n",
" | Return value>>self.\n",
" | \n",
" | __rshift__(self, value, /)\n",
" | Return self>>value.\n",
" | \n",
" | __rsub__(self, value, /)\n",
" | Return value-self.\n",
" | \n",
" | __rtruediv__(self, value, /)\n",
" | Return value/self.\n",
" | \n",
" | __rxor__(self, value, /)\n",
" | Return value^self.\n",
" | \n",
" | __setitem__(self, key, value, /)\n",
" | Set self[key] to value.\n",
" | \n",
" | __setstate__(...)\n",
" | a.__setstate__(state, /)\n",
" | \n",
" | For unpickling.\n",
" | \n",
" | The `state` argument must be a sequence that contains the following\
n",
" | elements:\n",
" | \n",
" | Parameters\n",
" | ----------\n",
" | version : int\n",
" | optional pickle version. If omitted defaults to 0.\n",
" | shape : tuple\n",
" | dtype : data-type\n",
" | isFortran : bool\n",
" | rawdata : string or list\n",
" | a binary string with the data (or a list if 'a' is an object
array)\n",
" | \n",
" | __sizeof__(...)\n",
" | Size of object in memory, in bytes.\n",
" | \n",
" | __str__(self, /)\n",
" | Return str(self).\n",
" | \n",
" | __sub__(self, value, /)\n",
" | Return self-value.\n",
" | \n",
" | __truediv__(self, value, /)\n",
" | Return self/value.\n",
" | \n",
" | __xor__(self, value, /)\n",
" | Return self^value.\n",
" | \n",
" | all(...)\n",
" | a.all(axis=None, out=None, keepdims=False, *, where=True)\n",
" | \n",
" | Returns True if all elements evaluate to True.\n",
" | \n",
" | Refer to `numpy.all` for full documentation.\n",
" | \n",
" | See Also\n",
" | --------\n",
" | numpy.all : equivalent function\n",
" | \n",
" | any(...)\n",
" | a.any(axis=None, out=None, keepdims=False, *, where=True)\n",
" | \n",
" | Returns True if any of the elements of à` evaluate to True.\n",
" | \n",
" | Refer to `numpy.any` for full documentation.\n",
" | \n",
" | See Also\n",
" | --------\n",
" | numpy.any : equivalent function\n",
" | \n",
" | argmax(...)\n",
" | a.argmax(axis=None, out=None, *, keepdims=False)\n",
" | \n",
" | Return indices of the maximum values along the given axis.\n",
" | \n",
" | Refer to `numpy.argmax` for full documentation.\n",
" | \n",
" | See Also\n",
" | --------\n",
" | numpy.argmax : equivalent function\n",
" | \n",
" | argmin(...)\n",
" | a.argmin(axis=None, out=None, *, keepdims=False)\n",
" | \n",
" | Return indices of the minimum values along the given axis.\n",
" | \n",
" | Refer to `numpy.argmin` for detailed documentation.\n",
" | \n",
" | See Also\n",
" | --------\n",
" | numpy.argmin : equivalent function\n",
" | \n",
" | argpartition(...)\n",
" | a.argpartition(kth, axis=-1, kind='introselect', order=None)\n",
" | \n",
" | Returns the indices that would partition this array.\n",
" | \n",
" | Refer to `numpy.argpartition` for full documentation.\n",
" | \n",
" | .. versionadded:: 1.8.0\n",
" | \n",
" | See Also\n",
" | --------\n",
" | numpy.argpartition : equivalent function\n",
" | \n",
" | argsort(...)\n",
" | a.argsort(axis=-1, kind=None, order=None)\n",
" | \n",
" | Returns the indices that would sort this array.\n",
" | \n",
" | Refer to `numpy.argsort` for full documentation.\n",
" | \n",
" | See Also\n",
" | --------\n",
" | numpy.argsort : equivalent function\n",
" | \n",
" | astype(...)\n",
" | a.astype(dtype, order='K', casting='unsafe', subok=True, copy=True)\
n",
" | \n",
" | Copy of the array, cast to a specified type.\n",
" | \n",
" | Parameters\n",
" | ----------\n",
" | dtype : str or dtype\n",
" | Typecode or data-type to which the array is cast.\n",
" | order : {'C', 'F', 'A', 'K'}, optional\n",
" | Controls the memory layout order of the result.\n",
" | 'C' means C order, 'F' means Fortran order, 'A'\n",
" | means 'F' order if all the arrays are Fortran contiguous,\n",
" | 'C' order otherwise, and 'K' means as close to the\n",
" | order the array elements appear in memory as possible.\n",
" | Default is 'K'.\n",
" | casting : {'no', 'equiv', 'safe', 'same_kind', 'unsafe'}, optional\
n",
" | Controls what kind of data casting may occur. Defaults to
'unsafe'\n",
" | for backwards compatibility.\n",
" | \n",
" | * 'no' means the data types should not be cast at all.\n",
" | * 'equiv' means only byte-order changes are allowed.\n",
" | * 'safe' means only casts which can preserve values are
allowed.\n",
" | * 'same_kind' means only safe casts or casts within a kind,\
n",
" | like float64 to float32, are allowed.\n",
" | * 'unsafe' means any data conversions may be done.\n",
" | subok : bool, optional\n",
" | If True, then sub-classes will be passed-through (default),
otherwise\n",
" | the returned array will be forced to be a base-class array.\n",
" | copy : bool, optional\n",
" | By default, astype always returns a newly allocated array. If
this\n",
" | is set to false, and the `dtype`, òrder`, and `subok`\n",
" | requirements are satisfied, the input array is returned instead\
n",
" | of a copy.\n",
" | \n",
" | Returns\n",
" | -------\n",
" | arr_t : ndarray\n",
" | Unless `copy` is False and the other conditions for returning
the input\n",
" | array are satisfied (see description for `copy` input
parameter), àrr_t`\n",
" | is a new array of the same shape as the input array, with dtype,
order\n",
" | given by `dtype`, òrder`.\n",
" | \n",
" | Notes\n",
" | -----\n",
" | .. versionchanged:: 1.17.0\n",
" | Casting between a simple data type and a structured one is
possible only\n",
" | for \"unsafe\" casting. Casting to multiple fields is allowed,
but\n",
" | casting from multiple fields is not.\n",
" | \n",
" | Casting from numeric to string types in 'safe' casting mode
requires\n",
" | that the string dtype length is long enough to store the max\n",
" | integer/float value converted.\n",
" | \n",
" | Raises\n",
" | ------\n",
" | ComplexWarning\n",
" | When casting from complex to float or int. To avoid this,\n",
" | one should use `à.real.astype(t)``.\n",
" | \n",
" | Examples\n",
" | --------\n",
" | >>> x = np.array([1, 2, 2.5])\n",
" | >>> x\n",
" | array([1. , 2. , 2.5])\n",
" | \n",
" | >>> x.astype(int)\n",
" | array([1, 2, 2])\n",
" | \n",
" | byteswap(...)\n",
" | a.byteswap(inplace=False)\n",
" | \n",
" | Swap the bytes of the array elements\n",
" | \n",
" | Toggle between low-endian and big-endian data representation by\n",
" | returning a byteswapped array, optionally swapped in-place.\n",
" | Arrays of byte-strings are not swapped. The real and imaginary\n",
" | parts of a complex number are swapped individually.\n",
" | \n",
" | Parameters\n",
" | ----------\n",
" | inplace : bool, optional\n",
" | If ``True``, swap bytes in-place, default is ``False``.\n",
" | \n",
" | Returns\n",
" | -------\n",
" | out : ndarray\n",
" | The byteswapped array. If ìnplace` is ``True``, this is\n",
" | a view to self.\n",
" | \n",
" | Examples\n",
" | --------\n",
" | >>> A = np.array([1, 256, 8755], dtype=np.int16)\n",
" | >>> list(map(hex, A))\n",
" | ['0x1', '0x100', '0x2233']\n",
" | >>> A.byteswap(inplace=True)\n",
" | array([ 256, 1, 13090], dtype=int16)\n",
" | ['0x100', '0x1', '0x3322']\n",
" | \n",
" | Arrays of byte-strings are not swapped\n",
" | \n",
" | >>> A = np.array([b'ceg', b'fac'])\n",
" | >>> A.byteswap()\n",
" | array([b'ceg', b'fac'], dtype='|S3')\n",
" | \n",
" | `À.newbyteorder().byteswap()`` produces an array with the same
values\n",
" | but different representation in memory\n",
" | \n",
" | >>> A = np.array([1, 2, 3])\n",
" | >>> A.view(np.uint8)\n",
" | array([1, 0, 0, 0, 0, 0, 0, 0, 2, 0, 0, 0, 0, 0, 0, 0, 3, 0, 0, 0,
0, 0,\n",
" | 0, 0], dtype=uint8)\n",
" | >>> A.newbyteorder().byteswap(inplace=True)\n",
" | array([1, 2, 3])\n",
" | array([0, 0, 0, 0, 0, 0, 0, 1, 0, 0, 0, 0, 0, 0, 0, 2, 0, 0, 0, 0,
0, 0,\n",
" | 0, 3], dtype=uint8)\n",
" | \n",
" | choose(...)\n",
" | a.choose(choices, out=None, mode='raise')\n",
" | \n",
" | Use an index array to construct a new array from a set of choices.\
n",
" | \n",
" | Refer to `numpy.choose` for full documentation.\n",
" | \n",
" | See Also\n",
" | --------\n",
" | numpy.choose : equivalent function\n",
" | \n",
" | clip(...)\n",
" | a.clip(min=None, max=None, out=None, **kwargs)\n",
" | \n",
" | Return an array whose values are limited to ``[min, max]``.\n",
" | One of max or min must be given.\n",
" | \n",
" | Refer to `numpy.clip` for full documentation.\n",
" | \n",
" | See Also\n",
" | --------\n",
" | numpy.clip : equivalent function\n",
" | \n",
" | compress(...)\n",
" | a.compress(condition, axis=None, out=None)\n",
" | \n",
" | Return selected slices of this array along given axis.\n",
" | \n",
" | Refer to `numpy.compress` for full documentation.\n",
" | \n",
" | See Also\n",
" | --------\n",
" | numpy.compress : equivalent function\n",
" | \n",
" | conj(...)\n",
" | a.conj()\n",
" | \n",
" | Complex-conjugate all elements.\n",
" | \n",
" | Refer to `numpy.conjugate` for full documentation.\n",
" | \n",
" | See Also\n",
" | --------\n",
" | numpy.conjugate : equivalent function\n",
" | \n",
" | conjugate(...)\n",
" | a.conjugate()\n",
" | \n",
" | Return the complex conjugate, element-wise.\n",
" | \n",
" | \n",
" | See Also\n",
" | --------\n",
" | \n",
" | copy(...)\n",
" | a.copy(order='C')\n",
" | \n",
" | Return a copy of the array.\n",
" | \n",
" | Parameters\n",
" | ----------\n",
" | Controls the memory layout of the copy. 'C' means C-order,\n",
" | 'F' means F-order, 'A' means 'F' if à` is Fortran contiguous,\
n",
" | 'C' otherwise. 'K' means match the layout of à` as closely\n",
" | as possible. (Note that this function and :func:`numpy.copy` are
very\n",
" | similar but have different default values for their order=\n",
" | arguments, and this function always passes sub-classes
through.)\n",
" | \n",
" | See also\n",
" | --------\n",
" | numpy.copy : Similar function with different default behavior\n",
" | numpy.copyto\n",
" | \n",
" | Notes\n",
" | -----\n",
" | This function is the preferred method for creating an array copy.
The\n",
" | function :func:`numpy.copy` is similar, but it defaults to using
order 'K',\n",
" | and will not pass sub-classes through by default.\n",
" | \n",
" | Examples\n",
" | --------\n",
" | >>> x = np.array([[1,2,3],[4,5,6]], order='F')\n",
" | \n",
" | >>> y = x.copy()\n",
" | \n",
" | >>> x.fill(0)\n",
" | \n",
" | >>> x\n",
" | array([[0, 0, 0],\n",
" | [0, 0, 0]])\n",
" | \n",
" | >>> y\n",
" | array([[1, 2, 3],\n",
" | [4, 5, 6]])\n",
" | \n",
" | >>> y.flags['C_CONTIGUOUS']\n",
" | True\n",
" | \n",
" | cumprod(...)\n",
" | a.cumprod(axis=None, dtype=None, out=None)\n",
" | \n",
" | Return the cumulative product of the elements along the given axis.\
n",
" | \n",
" | Refer to `numpy.cumprod` for full documentation.\n",
" | \n",
" | See Also\n",
" | --------\n",
" | numpy.cumprod : equivalent function\n",
" | \n",
" | cumsum(...)\n",
" | a.cumsum(axis=None, dtype=None, out=None)\n",
" | \n",
" | Return the cumulative sum of the elements along the given axis.\n",
" | \n",
" | Refer to `numpy.cumsum` for full documentation.\n",
" | \n",
" | See Also\n",
" | --------\n",
" | numpy.cumsum : equivalent function\n",
" | \n",
" | diagonal(...)\n",
" | a.diagonal(offset=0, axis1=0, axis2=1)\n",
" | \n",
" | Return specified diagonals. In NumPy 1.9 the returned array is a\n",
" | read-only view instead of a copy as in previous NumPy versions. In\
n",
" | a future version the read-only restriction will be removed.\n",
" | \n",
" | Refer to :func:`numpy.diagonal` for full documentation.\n",
" | \n",
" | See Also\n",
" | --------\n",
" | numpy.diagonal : equivalent function\n",
" | \n",
" | dot(...)\n",
" | \n",
" | dump(...)\n",
" | a.dump(file)\n",
" | \n",
" | Dump a pickle of the array to the specified file.\n",
" | The array can be read back with pickle.load or numpy.load.\n",
" | \n",
" | Parameters\n",
" | ----------\n",
" | file : str or Path\n",
" | A string naming the dump file.\n",
" | \n",
" | `pathlib.Path` objects are now accepted.\n",
" | \n",
" | dumps(...)\n",
" | a.dumps()\n",
" | \n",
" | Returns the pickle of the array as a string.\n",
" | pickle.loads will convert the string back to an array.\n",
" | \n",
" | Parameters\n",
" | ----------\n",
" | None\n",
" | \n",
" | fill(...)\n",
" | a.fill(value)\n",
" | \n",
" | Fill the array with a scalar value.\n",
" | \n",
" | Parameters\n",
" | ----------\n",
" | value : scalar\n",
" | All elements of à` will be assigned this value.\n",
" | \n",
" | Examples\n",
" | --------\n",
" | >>> a = np.array([1, 2])\n",
" | >>> a.fill(0)\n",
" | >>> a\n",
" | array([0, 0])\n",
" | >>> a = np.empty(2)\n",
" | >>> a.fill(1)\n",
" | >>> a\n",
" | array([1., 1.])\n",
" | \n",
" | Fill expects a scalar value and always behaves the same as
assigning\n",
" | to a single array element. The following is a rare example where
this\n",
" | distinction is important:\n",
" | \n",
" | >>> a = np.array([None, None], dtype=object)\n",
" | >>> a[0] = np.array(3)\n",
" | >>> a\n",
" | array([array(3), None], dtype=object)\n",
" | >>> a.fill(np.array(3))\n",
" | >>> a\n",
" | array([array(3), array(3)], dtype=object)\n",
" | \n",
" | Where other forms of assignments will unpack the array being
assigned:\n",
" | \n",
" | >>> a[...] = np.array(3)\n",
" | >>> a\n",
" | array([3, 3], dtype=object)\n",
" | \n",
" | flatten(...)\n",
" | a.flatten(order='C')\n",
" | \n",
" | Return a copy of the array collapsed into one dimension.\n",
" | \n",
" | Parameters\n",
" | ----------\n",
" | 'C' means to flatten in row-major (C-style) order.\n",
" | 'F' means to flatten in column-major (Fortran-\n",
" | style) order. 'A' means to flatten in column-major\n",
" | order if à` is Fortran *contiguous* in memory,\n",
" | row-major order otherwise. 'K' means to flatten\n",
" | à` in the order the elements occur in memory.\n",
" | The default is 'C'.\n",
" | \n",
" | Returns\n",
" | -------\n",
" | y : ndarray\n",
" | A copy of the input array, flattened to one dimension.\n",
" | \n",
" | See Also\n",
" | --------\n",
" | ravel : Return a flattened array.\n",
" | flat : A 1-D flat iterator over the array.\n",
" | \n",
" | Examples\n",
" | --------\n",
" | >>> a = np.array([[1,2], [3,4]])\n",
" | >>> a.flatten()\n",
" | array([1, 2, 3, 4])\n",
" | >>> a.flatten('F')\n",
" | array([1, 3, 2, 4])\n",
" | \n",
" | getfield(...)\n",
" | a.getfield(dtype, offset=0)\n",
" | \n",
" | Returns a field of the given array as a certain type.\n",
" | \n",
" | A field is a view of the array data with a given data-type. The
values in\n",
" | the view are determined by the given type and the offset into the
current\n",
" | array in bytes. The offset needs to be such that the view dtype fits
in the\n",
" | array dtype; for example an array of dtype complex128 has 16-byte
elements.\n",
" | If taking a view with a 32-bit integer (4 bytes), the offset needs
to be\n",
" | between 0 and 12 bytes.\n",
" | \n",
" | Parameters\n",
" | ----------\n",
" | The data type of the view. The dtype size of the view can not be
larger\n",
" | than that of the array itself.\n",
" | offset : int\n",
" | Number of bytes to skip before beginning the element view.\n",
" | \n",
" | Examples\n",
" | --------\n",
" | >>> x = np.diag([1.+1.j]*2)\n",
" | >>> x[1, 1] = 2 + 4.j\n",
" | >>> x\n",
" | array([[1.+1.j, 0.+0.j],\n",
" | [0.+0.j, 2.+4.j]])\n",
" | >>> x.getfield(np.float64)\n",
" | array([[1., 0.],\n",
" | [0., 2.]])\n",
" | \n",
" | By choosing an offset of 8 bytes we can select the complex part of
the\n",
" | array for our view:\n",
" | \n",
" | >>> x.getfield(np.float64, offset=8)\n",
" | array([[1., 0.],\n",
" | [0., 4.]])\n",
" | \n",
" | item(...)\n",
" | a.item(*args)\n",
" | \n",
" | Copy an element of an array to a standard Python scalar and return
it.\n",
" | \n",
" | Parameters\n",
" | ----------\n",
" | \\*args : Arguments (variable number and type)\n",
" | \n",
" | * none: in this case, the method only works for arrays\n",
" | with one element (à.size == 1`), which element is\n",
" | copied into a standard Python scalar object and returned.\n",
" | \n",
" | * int_type: this argument is interpreted as a flat index into\
n",
" | the array, specifying which element to copy and return.\n",
" | \n",
" | * tuple of int_types: functions as does a single int_type
argument,\n",
" | except that the argument is interpreted as an nd-index into
the\n",
" | array.\n",
" | \n",
" | Returns\n",
" | -------\n",
" | z : Standard Python scalar object\n",
" | A copy of the specified element of the array as a suitable\n",
" | Python scalar\n",
" | \n",
" | Notes\n",
" | -----\n",
" | When the data type of à` is longdouble or clongdouble, item()
returns\n",
" | a scalar array object because there is no available Python scalar
that\n",
" | would not lose information. Void arrays return a buffer object for
item(),\n",
" | unless fields are defined, in which case a tuple is returned.\n",
" | \n",
" | ìtem` is very similar to a[args], except, instead of an array
scalar,\n",
" | a standard Python scalar is returned. This can be useful for
speeding up\n",
" | access to elements of the array and doing arithmetic on elements of
the\n",
" | array using Python's optimized math.\n",
" | \n",
" | Examples\n",
" | --------\n",
" | >>> np.random.seed(123)\n",
" | >>> x = np.random.randint(9, size=(3, 3))\n",
" | >>> x\n",
" | array([[2, 2, 6],\n",
" | [1, 3, 6],\n",
" | [1, 0, 1]])\n",
" | >>> x.item(3)\n",
" | 1\n",
" | >>> x.item(7)\n",
" | 0\n",
" | >>> x.item((0, 1))\n",
" | 2\n",
" | >>> x.item((2, 2))\n",
" | 1\n",
" | \n",
" | itemset(...)\n",
" | a.itemset(*args)\n",
" | \n",
" | Insert scalar into an array (scalar is cast to array's dtype, if
possible)\n",
" | \n",
" | There must be at least 1 argument, and define the last argument\n",
" | as *item*. Then, `à.itemset(*args)`` is equivalent to but faster\
n",
" | than `à[args] = item``. The item should be a scalar value and
àrgs`\n",
" | must select a single item in the array à`.\n",
" | \n",
" | Parameters\n",
" | ----------\n",
" | \\*args : Arguments\n",
" | If one argument: a scalar, only used in case à` is of size 1.\
n",
" | If two arguments: the last argument is the value to be set\n",
" | and must be a scalar, the first argument specifies a single
array\n",
" | element location. It is either an int or a tuple.\n",
" | \n",
" | Notes\n",
" | -----\n",
" | Compared to indexing syntax, ìtemset` provides some speed increase\
n",
" | for placing a scalar into a particular location in an `ndarray`,\n",
" | if you must do this. However, generally this is discouraged:\n",
" | among other problems, it complicates the appearance of the code.\n",
" | Also, when using ìtemset` (and ìtem`) inside a loop, be sure\n",
" | to assign the methods to a local variable to avoid the attribute\n",
" | look-up at each loop iteration.\n",
" | \n",
" | Examples\n",
" | --------\n",
" | >>> x\n",
" | array([[2, 2, 6],\n",
" | [1, 3, 6],\n",
" | [1, 0, 1]])\n",
" | >>> x.itemset(4, 0)\n",
" | >>> x.itemset((2, 2), 9)\n",
" | >>> x\n",
" | array([[2, 2, 6],\n",
" | [1, 0, 6],\n",
" | [1, 0, 9]])\n",
" | \n",
" | max(...)\n",
" | a.max(axis=None, out=None, keepdims=False, initial=<no value>,
where=True)\n",
" | \n",
" | Return the maximum along a given axis.\n",
" | \n",
" | Refer to `numpy.amax` for full documentation.\n",
" | \n",
" | See Also\n",
" | --------\n",
" | numpy.amax : equivalent function\n",
" | \n",
" | mean(...)\n",
" | a.mean(axis=None, dtype=None, out=None, keepdims=False, *,
where=True)\n",
" | \n",
" | Returns the average of the array elements along given axis.\n",
" | \n",
" | Refer to `numpy.mean` for full documentation.\n",
" | \n",
" | See Also\n",
" | --------\n",
" | numpy.mean : equivalent function\n",
" | \n",
" | min(...)\n",
" | a.min(axis=None, out=None, keepdims=False, initial=<no value>,
where=True)\n",
" | \n",
" | Return the minimum along a given axis.\n",
" | \n",
" | Refer to `numpy.amin` for full documentation.\n",
" | \n",
" | See Also\n",
" | --------\n",
" | numpy.amin : equivalent function\n",
" | \n",
" | newbyteorder(...)\n",
" | arr.newbyteorder(new_order='S', /)\n",
" | \n",
" | Return the array with the same data viewed with a different byte
order.\n",
" | \n",
" | Equivalent to::\n",
" | \n",
" | arr.view(arr.dtype.newbytorder(new_order))\n",
" | \n",
" | Changes are also made in all fields and sub-arrays of the array
data\n",
" | type.\n",
" | \n",
" | \n",
" | \n",
" | Parameters\n",
" | ----------\n",
" | new_order : string, optional\n",
" | Byte order to force; a value from the byte order specifications\
n",
" | below. `new_order` codes can be any of:\n",
" | \n",
" | * 'S' - swap dtype from current to opposite endian\n",
" | * {'<', 'little'} - little endian\n",
" | * {'>', 'big'} - big endian\n",
" | * {'=', 'native'} - native order, equivalent to `sys.byteorder`\
n",
" | * {'|', 'I'} - ignore (no change to byte order)\n",
" | \n",
" | The default value ('S') results in swapping the current\n",
" | byte order.\n",
" | \n",
" | \n",
" | Returns\n",
" | -------\n",
" | new_arr : array\n",
" | New array object with the dtype reflecting given change to the\
n",
" | byte order.\n",
" | \n",
" | nonzero(...)\n",
" | a.nonzero()\n",
" | \n",
" | Return the indices of the elements that are non-zero.\n",
" | \n",
" | Refer to `numpy.nonzero` for full documentation.\n",
" | \n",
" | See Also\n",
" | --------\n",
" | numpy.nonzero : equivalent function\n",
" | \n",
" | partition(...)\n",
" | a.partition(kth, axis=-1, kind='introselect', order=None)\n",
" | \n",
" | Rearranges the elements in the array in such a way that the value of
the\n",
" | element in kth position is in the position it would be in a sorted
array.\n",
" | All elements smaller than the kth element are moved before this
element and\n",
" | all equal or greater are moved behind it. The ordering of the
elements in\n",
" | the two partitions is undefined.\n",
" | \n",
" | \n",
" | Parameters\n",
" | ----------\n",
" | kth : int or sequence of ints\n",
" | Element index to partition by. The kth element value will be in
its\n",
" | final sorted position and all smaller elements will be moved
before it\n",
" | and all equal or greater elements behind it.\n",
" | The order of all elements in the partitions is undefined.\n",
" | If provided with a sequence of kth it will partition all
elements\n",
" | indexed by kth of them into their sorted position at once.\n",
" | \n",
" | .. deprecated:: 1.22.0\n",
" | Passing booleans as index is deprecated.\n",
" | axis : int, optional\n",
" | Axis along which to sort. Default is -1, which means sort along
the\n",
" | last axis.\n",
" | kind : {'introselect'}, optional\n",
" | Selection algorithm. Default is 'introselect'.\n",
" | order : str or list of str, optional\n",
" | When à` is an array with fields defined, this argument
specifies\n",
" | which fields to compare first, second, etc. A single field can\
n",
" | be specified as a string, and not all fields need to be
specified,\n",
" | but unspecified fields will still be used, in the order in
which\n",
" | they come up in the dtype, to break ties.\n",
" | \n",
" | See Also\n",
" | --------\n",
" | numpy.partition : Return a partitioned copy of an array.\n",
" | argpartition : Indirect partition.\n",
" | sort : Full sort.\n",
" | \n",
" | Notes\n",
" | -----\n",
" | See ``np.partition`` for notes on the different algorithms.\n",
" | \n",
" | Examples\n",
" | --------\n",
" | >>> a = np.array([3, 4, 2, 1])\n",
" | >>> a.partition(3)\n",
" | >>> a\n",
" | array([2, 1, 3, 4])\n",
" | \n",
" | >>> a.partition((1, 3))\n",
" | >>> a\n",
" | array([1, 2, 3, 4])\n",
" | \n",
" | prod(...)\n",
" | a.prod(axis=None, dtype=None, out=None, keepdims=False, initial=1,
where=True)\n",
" | \n",
" | Return the product of the array elements over the given axis\n",
" | \n",
" | Refer to `numpy.prod` for full documentation.\n",
" | \n",
" | See Also\n",
" | --------\n",
" | numpy.prod : equivalent function\n",
" | \n",
" | ptp(...)\n",
" | a.ptp(axis=None, out=None, keepdims=False)\n",
" | \n",
" | Peak to peak (maximum - minimum) value along a given axis.\n",
" | \n",
" | Refer to `numpy.ptp` for full documentation.\n",
" | \n",
" | See Also\n",
" | --------\n",
" | numpy.ptp : equivalent function\n",
" | \n",
" | put(...)\n",
" | a.put(indices, values, mode='raise')\n",
" | \n",
" | Set `à.flat[n] = values[n]`` for all `n` in indices.\n",
" | \n",
" | Refer to `numpy.put` for full documentation.\n",
" | \n",
" | See Also\n",
" | --------\n",
" | numpy.put : equivalent function\n",
" | \n",
" | ravel(...)\n",
" | a.ravel([order])\n",
" | \n",
" | Return a flattened array.\n",
" | \n",
" | Refer to `numpy.ravel` for full documentation.\n",
" | \n",
" | See Also\n",
" | --------\n",
" | numpy.ravel : equivalent function\n",
" | \n",
" | ndarray.flat : a flat iterator on the array.\n",
" | \n",
" | repeat(...)\n",
" | a.repeat(repeats, axis=None)\n",
" | \n",
" | Repeat elements of an array.\n",
" | \n",
" | Refer to `numpy.repeat` for full documentation.\n",
" | \n",
" | See Also\n",
" | --------\n",
" | numpy.repeat : equivalent function\n",
" | \n",
" | reshape(...)\n",
" | a.reshape(shape, order='C')\n",
" | \n",
" | Returns an array containing the same data with a new shape.\n",
" | \n",
" | Refer to `numpy.reshape` for full documentation.\n",
" | \n",
" | See Also\n",
" | --------\n",
" | numpy.reshape : equivalent function\n",
" | \n",
" | Notes\n",
" | -----\n",
" | Unlike the free function `numpy.reshape`, this method on `ndarray`
allows\n",
" | the elements of the shape parameter to be passed in as separate
arguments.\n",
" | For example, `à.reshape(10, 11)`` is equivalent to\n",
" | `à.reshape((10, 11))``.\n",
" | \n",
" | resize(...)\n",
" | a.resize(new_shape, refcheck=True)\n",
" | \n",
" | Change shape and size of array in-place.\n",
" | \n",
" | Parameters\n",
" | ----------\n",
" | new_shape : tuple of ints, or `n` ints\n",
" | Shape of resized array.\n",
" | refcheck : bool, optional\n",
" | If False, reference count will not be checked. Default is True.\
n",
" | \n",
" | Returns\n",
" | -------\n",
" | None\n",
" | \n",
" | Raises\n",
" | ------\n",
" | ValueError\n",
" | If à` does not own its own data or references or views to it
exist,\n",
" | and the data memory must be changed.\n",
" | PyPy only: will always raise if the data memory must be changed,
since\n",
" | there is no reliable way to determine if references or views to
it\n",
" | exist.\n",
" | \n",
" | SystemError\n",
" | If the òrder` keyword argument is specified. This behaviour is
a\n",
" | bug in NumPy.\n",
" | \n",
" | See Also\n",
" | --------\n",
" | resize : Return a new array with the specified shape.\n",
" | \n",
" | Notes\n",
" | -----\n",
" | This reallocates space for the data area if necessary.\n",
" | \n",
" | Only contiguous arrays (data elements consecutive in memory) can be\
n",
" | resized.\n",
" | \n",
" | The purpose of the reference count check is to make sure you\n",
" | do not use this array as a buffer for another Python object and
then\n",
" | reallocate the memory. However, reference counts can increase in\n",
" | other ways so if you are sure that you have not shared the memory\
n",
" | for this array with another Python object, then you may safely set\
n",
" | `refcheck` to False.\n",
" | \n",
" | Examples\n",
" | --------\n",
" | Shrinking an array: array is flattened (in the order that the data
are\n",
" | stored in memory), resized, and reshaped:\n",
" | \n",
" | >>> a = np.array([[0, 1], [2, 3]], order='C')\n",
" | >>> a.resize((2, 1))\n",
" | >>> a\n",
" | array([[0],\n",
" | [1]])\n",
" | \n",
" | >>> a = np.array([[0, 1], [2, 3]], order='F')\n",
" | >>> a.resize((2, 1))\n",
" | >>> a\n",
" | array([[0],\n",
" | [2]])\n",
" | \n",
" | Enlarging an array: as above, but missing entries are filled with
zeros:\n",
" | \n",
" | >>> b = np.array([[0, 1], [2, 3]])\n",
" | >>> b.resize(2, 3) # new_shape parameter doesn't have to be a tuple\
n",
" | >>> b\n",
" | array([[0, 1, 2],\n",
" | [3, 0, 0]])\n",
" | \n",
" | Referencing an array prevents resizing...\n",
" | \n",
" | >>> c = a\n",
" | >>> a.resize((1, 1))\n",
" | Traceback (most recent call last):\n",
" | ...\n",
" | ValueError: cannot resize an array that references or is
referenced ...\n",
" | \n",
" | Unless `refcheck` is False:\n",
" | \n",
" | >>> a.resize((1, 1), refcheck=False)\n",
" | >>> a\n",
" | array([[0]])\n",
" | >>> c\n",
" | array([[0]])\n",
" | \n",
" | round(...)\n",
" | a.round(decimals=0, out=None)\n",
" | \n",
" | Return à` with each element rounded to the given number of
decimals.\n",
" | \n",
" | Refer to `numpy.around` for full documentation.\n",
" | \n",
" | See Also\n",
" | --------\n",
" | numpy.around : equivalent function\n",
" | \n",
" | searchsorted(...)\n",
" | a.searchsorted(v, side='left', sorter=None)\n",
" | \n",
" | Find indices where elements of v should be inserted in a to maintain
order.\n",
" | \n",
" | For full documentation, see `numpy.searchsorted`\n",
" | \n",
" | See Also\n",
" | --------\n",
" | numpy.searchsorted : equivalent function\n",
" | \n",
" | setfield(...)\n",
" | a.setfield(val, dtype, offset=0)\n",
" | \n",
" | Put a value into a specified place in a field defined by a data-
type.\n",
" | \n",
" | Place `val` into à`'s field defined by `dtype` and beginning
òffset`\n",
" | bytes into the field.\n",
" | \n",
" | Parameters\n",
" | ----------\n",
" | val : object\n",
" | Value to be placed in field.\n",
" | Data-type of the field in which to place `val`.\n",
" | The number of bytes into the field at which to place `val`.\n",
" | \n",
" | Returns\n",
" | -------\n",
" | None\n",
" | \n",
" | See Also\n",
" | --------\n",
" | getfield\n",
" | \n",
" | Examples\n",
" | --------\n",
" | >>> x = np.eye(3)\n",
" | array([[1., 0., 0.],\n",
" | [0., 1., 0.],\n",
" | [0., 0., 1.]])\n",
" | >>> x.setfield(3, np.int32)\n",
" | >>> x.getfield(np.int32)\n",
" | array([[3, 3, 3],\n",
" | [3, 3, 3],\n",
" | [3, 3, 3]], dtype=int32)\n",
" | >>> x\n",
" | array([[1.0e+000, 1.5e-323, 1.5e-323],\n",
" | [1.5e-323, 1.0e+000, 1.5e-323],\n",
" | [1.5e-323, 1.5e-323, 1.0e+000]])\n",
" | >>> x.setfield(np.eye(3), np.int32)\n",
" | >>> x\n",
" | array([[1., 0., 0.],\n",
" | [0., 1., 0.],\n",
" | [0., 0., 1.]])\n",
" | \n",
" | setflags(...)\n",
" | a.setflags(write=None, align=None, uic=None)\n",
" | \n",
" | Set array flags WRITEABLE, ALIGNED, WRITEBACKIFCOPY,\n",
" | respectively.\n",
" | \n",
" | These Boolean-valued flags affect how numpy interprets the memory\
n",
" | area used by à` (see Notes below). The ALIGNED flag can only\n",
" | be set to True if the data is actually aligned according to the
type.\n",
" | The WRITEBACKIFCOPY and flag can never be set\n",
" | to True. The flag WRITEABLE can only be set to True if the array
owns its\n",
" | own memory, or the ultimate owner of the memory exposes a writeable
buffer\n",
" | interface, or is a string. (The exception for string is made so
that\n",
" | unpickling can be done without copying memory.)\n",
" | \n",
" | Parameters\n",
" | ----------\n",
" | write : bool, optional\n",
" | Describes whether or not à` can be written to.\n",
" | align : bool, optional\n",
" | Describes whether or not à` is aligned properly for its type.\
n",
" | uic : bool, optional\n",
" | Describes whether or not à` is a copy of another \"base\"
array.\n",
" | \n",
" | Notes\n",
" | -----\n",
" | Array flags provide information about how the memory area used\n",
" | for the array is to be interpreted. There are 7 Boolean flags\n",
" | in use, only four of which can be changed by the user:\n",
" | WRITEBACKIFCOPY, WRITEABLE, and ALIGNED.\n",
" | \n",
" | WRITEABLE (W) the data area can be written to;\n",
" | \n",
" | ALIGNED (A) the data and strides are aligned appropriately for the
hardware\n",
" | (as determined by the compiler);\n",
" | \n",
" | WRITEBACKIFCOPY (X) this array is a copy of some other array
(referenced\n",
" | by .base). When the C-API function PyArray_ResolveWritebackIfCopy
is\n",
" | called, the base array will be updated with the contents of this
array.\n",
" | \n",
" | All flags can be accessed using the single (upper case) letter as
well\n",
" | as the full name.\n",
" | \n",
" | Examples\n",
" | --------\n",
" | >>> y = np.array([[3, 1, 7],\n",
" | ... [2, 0, 0],\n",
" | ... [8, 5, 9]])\n",
" | >>> y\n",
" | array([[3, 1, 7],\n",
" | [2, 0, 0],\n",
" | [8, 5, 9]])\n",
" | >>> y.flags\n",
" | C_CONTIGUOUS : True\n",
" | F_CONTIGUOUS : False\n",
" | OWNDATA : True\n",
" | WRITEABLE : True\n",
" | ALIGNED : True\n",
" | WRITEBACKIFCOPY : False\n",
" | >>> y.setflags(write=0, align=0)\n",
" | >>> y.flags\n",
" | WRITEABLE : False\n",
" | ALIGNED : False\n",
" | >>> y.setflags(uic=1)\n",
" | File \"<stdin>\", line 1, in <module>\n",
" | ValueError: cannot set WRITEBACKIFCOPY flag to True\n",
" | \n",
" | sort(...)\n",
" | a.sort(axis=-1, kind=None, order=None)\n",
" | \n",
" | Sort an array in-place. Refer to `numpy.sort` for full
documentation.\n",
" | \n",
" | Parameters\n",
" | ----------\n",
the\n",
" | last axis.\n",
" | kind : {'quicksort', 'mergesort', 'heapsort', 'stable'}, optional\
n",
" | Sorting algorithm. The default is 'quicksort'. Note that both
'stable'\n",
" | and 'mergesort' use timsort under the covers and, in general,
the\n",
" | actual implementation will vary with datatype. The 'mergesort'
option\n",
" | is retained for backwards compatibility.\n",
" | \n",
" | The 'stable' option was added.\n",
" | \n",
specifies\n",
n",
" | be specified as a string, and not all fields need be specified,\
n",
which\n",
" | \n",
" | See Also\n",
" | --------\n",
" | numpy.sort : Return a sorted copy of an array.\n",
" | numpy.argsort : Indirect sort.\n",
" | numpy.lexsort : Indirect stable sort on multiple keys.\n",
" | numpy.searchsorted : Find elements in sorted array.\n",
" | numpy.partition: Partial sort.\n",
" | \n",
" | Notes\n",
" | -----\n",
" | See `numpy.sort` for notes on the different sorting algorithms.\n",
" | \n",
" | Examples\n",
" | --------\n",
" | >>> a = np.array([[1,4], [3,1]])\n",
" | >>> a.sort(axis=1)\n",
" | >>> a\n",
" | array([[1, 4],\n",
" | [1, 3]])\n",
" | >>> a.sort(axis=0)\n",
" | >>> a\n",
" | array([[1, 3],\n",
" | [1, 4]])\n",
" | \n",
" | Use the òrder` keyword to specify a field to use when sorting a\n",
" | structured array:\n",
" | \n",
" | >>> a = np.array([('a', 2), ('c', 1)], dtype=[('x', 'S1'), ('y',
int)])\n",
" | >>> a.sort(order='y')\n",
" | >>> a\n",
" | array([(b'c', 1), (b'a', 2)],\n",
" | dtype=[('x', 'S1'), ('y', '<i8')])\n",
" | \n",
" | squeeze(...)\n",
" | a.squeeze(axis=None)\n",
" | \n",
" | Remove axes of length one from à`.\n",
" | \n",
" | Refer to `numpy.squeeze` for full documentation.\n",
" | \n",
" | See Also\n",
" | --------\n",
" | numpy.squeeze : equivalent function\n",
" | \n",
" | std(...)\n",
" | a.std(axis=None, dtype=None, out=None, ddof=0, keepdims=False, *,
where=True)\n",
" | \n",
" | Returns the standard deviation of the array elements along given
axis.\n",
" | \n",
" | Refer to `numpy.std` for full documentation.\n",
" | \n",
" | See Also\n",
" | --------\n",
" | numpy.std : equivalent function\n",
" | \n",
" | sum(...)\n",
" | a.sum(axis=None, dtype=None, out=None, keepdims=False, initial=0,
where=True)\n",
" | \n",
" | Return the sum of the array elements over the given axis.\n",
" | \n",
" | Refer to `numpy.sum` for full documentation.\n",
" | \n",
" | See Also\n",
" | --------\n",
" | numpy.sum : equivalent function\n",
" | \n",
" | swapaxes(...)\n",
" | a.swapaxes(axis1, axis2)\n",
" | \n",
" | Return a view of the array with àxis1` and àxis2` interchanged.\
n",
" | \n",
" | Refer to `numpy.swapaxes` for full documentation.\n",
" | \n",
" | See Also\n",
" | --------\n",
" | numpy.swapaxes : equivalent function\n",
" | \n",
" | take(...)\n",
" | a.take(indices, axis=None, out=None, mode='raise')\n",
" | \n",
" | Return an array formed from the elements of à` at the given
indices.\n",
" | \n",
" | Refer to `numpy.take` for full documentation.\n",
" | \n",
" | See Also\n",
" | --------\n",
" | numpy.take : equivalent function\n",
" | \n",
" | tobytes(...)\n",
" | a.tobytes(order='C')\n",
" | \n",
" | Construct Python bytes containing the raw data bytes in the array.\
n",
" | \n",
" | Constructs Python bytes showing a copy of the raw contents of\n",
" | data memory. The bytes object is produced in C-order by default.\n",
" | This behavior is controlled by the `òrder`` parameter.\n",
" | \n",
" | \n",
" | Parameters\n",
" | ----------\n",
" | order : {'C', 'F', 'A'}, optional\n",
" | Controls the memory layout of the bytes object. 'C' means C-
order,\n",
" | 'F' means F-order, 'A' (short for *Any*) means 'F' if à` is\n",
" | Fortran contiguous, 'C' otherwise. Default is 'C'.\n",
" | \n",
" | Returns\n",
" | -------\n",
" | s : bytes\n",
" | Python bytes exhibiting a copy of à`'s raw data.\n",
" | \n",
" | See also\n",
" | --------\n",
" | frombuffer\n",
" | Inverse of this operation, construct a 1-dimensional array from
Python\n",
" | bytes.\n",
" | \n",
" | Examples\n",
" | --------\n",
" | >>> x = np.array([[0, 1], [2, 3]], dtype='<u2')\n",
" | >>> x.tobytes()\n",
" | b'\\x00\\x00\\x01\\x00\\x02\\x00\\x03\\x00'\n",
" | >>> x.tobytes('C') == x.tobytes()\n",
" | True\n",
" | >>> x.tobytes('F')\n",
" | b'\\x00\\x00\\x02\\x00\\x01\\x00\\x03\\x00'\n",
" | \n",
" | tofile(...)\n",
" | a.tofile(fid, sep=\"\", format=\"%s\")\n",
" | \n",
" | Write array to a file as text or binary (default).\n",
" | \n",
" | Data is always written in 'C' order, independent of the order of
à`.\n",
" | The data produced by this method can be recovered using the
function\n",
" | fromfile().\n",
" | \n",
" | Parameters\n",
" | ----------\n",
" | fid : file or str or Path\n",
" | An open file object, or a string containing a filename.\n",
" | \n",
" | \n",
" | sep : str\n",
" | Separator between array items for text output.\n",
" | If \"\" (empty), a binary file is written, equivalent to\n",
" | ``file.write(a.tobytes())``.\n",
" | format : str\n",
" | Format string for text file output.\n",
" | Each entry in the array is formatted to text by first
converting\n",
" | it to the closest Python type, and then using \"format\" %
item.\n",
" | \n",
" | Notes\n",
" | -----\n",
" | This is a convenience function for quick storage of array data.\n",
" | Information on endianness and precision is lost, so this method is
not a\n",
" | good choice for files intended to archive data or transport data
between\n",
" | machines with different endianness. Some of these problems can be
overcome\n",
" | by outputting the data as text files, at the expense of speed and
file\n",
" | size.\n",
" | \n",
" | When fid is a file object, array contents are directly written to
the\n",
" | file, bypassing the file object's ``write`` method. As a result,
tofile\n",
" | cannot be used with files objects supporting compression (e.g.,
GzipFile)\n",
" | or file-like objects that do not support ``fileno()`` (e.g.,
BytesIO).\n",
" | \n",
" | tolist(...)\n",
" | a.tolist()\n",
" | \n",
" | Return the array as an `à.ndim``-levels deep nested list of Python
scalars.\n",
" | \n",
" | Return a copy of the array data as a (nested) Python list.\n",
" | Data items are converted to the nearest compatible builtin Python
type, via\n",
" | the `~numpy.ndarray.item` function.\n",
" | \n",
" | If `à.ndim`` is 0, then since the depth of the nested list is 0, it
will\n",
" | not be a list at all, but a simple Python scalar.\n",
" | \n",
" | Parameters\n",
" | ----------\n",
" | none\n",
" | \n",
" | Returns\n",
" | -------\n",
" | y : object, or list of object, or list of list of object, or ...\n",
" | The possibly nested list of array elements.\n",
" | \n",
" | Notes\n",
" | -----\n",
" | The array may be recreated via `à = np.array(a.tolist())``,
although this\n",
" | may sometimes lose precision.\n",
" | \n",
" | Examples\n",
" | --------\n",
" | For a 1D array, `à.tolist()`` is almost the same as ``list(a)``,\
n",
" | except that ``tolist`` changes numpy scalars to Python scalars:\n",
" | \n",
" | >>> a = np.uint32([1, 2])\n",
" | >>> a_list = list(a)\n",
" | >>> a_list\n",
" | [1, 2]\n",
" | >>> type(a_list[0])\n",
" | <class 'numpy.uint32'>\n",
" | >>> a_tolist = a.tolist()\n",
" | >>> a_tolist\n",
" | [1, 2]\n",
" | >>> type(a_tolist[0])\n",
" | <class 'int'>\n",
" | \n",
" | Additionally, for a 2D array, ``tolist`` applies recursively:\n",
" | \n",
" | >>> a = np.array([[1, 2], [3, 4]])\n",
" | >>> list(a)\n",
" | [array([1, 2]), array([3, 4])]\n",
" | >>> a.tolist()\n",
" | [[1, 2], [3, 4]]\n",
" | \n",
" | The base case for this recursion is a 0D array:\n",
" | \n",
" | >>> a = np.array(1)\n",
" | >>> list(a)\n",
" | ...\n",
" | TypeError: iteration over a 0-d array\n",
" | >>> a.tolist()\n",
" | 1\n",
" | \n",
" | tostring(...)\n",
" | a.tostring(order='C')\n",
" | \n",
" | A compatibility alias for `tobytes`, with exactly the same
behavior.\n",
" | \n",
" | Despite its name, it returns `bytes` not `str`\\ s.\n",
" | \n",
" | .. deprecated:: 1.19.0\n",
" | \n",
" | trace(...)\n",
" | a.trace(offset=0, axis1=0, axis2=1, dtype=None, out=None)\n",
" | \n",
" | Return the sum along diagonals of the array.\n",
" | \n",
" | Refer to `numpy.trace` for full documentation.\n",
" | \n",
" | See Also\n",
" | --------\n",
" | numpy.trace : equivalent function\n",
" | \n",
" | transpose(...)\n",
" | a.transpose(*axes)\n",
" | \n",
" | Returns a view of the array with axes transposed.\n",
" | \n",
" | Refer to `numpy.transpose` for full documentation.\n",
" | \n",
" | Parameters\n",
" | ----------\n",
" | axes : None, tuple of ints, or `n` ints\n",
" | \n",
" | * None or no argument: reverses the order of the axes.\n",
" | \n",
" | * tuple of ints: ì` in the `j`-th place in the tuple means that
the\n",
" | array's ì`-th axis becomes the transposed array's `j`-th axis.\
n",
" | \n",
" | * `n` ints: same as an n-tuple of the same ints (this form is\n",
" | intended simply as a \"convenience\" alternative to the tuple
form).\n",
" | \n",
" | Returns\n",
" | -------\n",
" | p : ndarray\n",
" | View of the array with its axes suitably permuted.\n",
" | \n",
" | See Also\n",
" | --------\n",
" | transpose : Equivalent function.\n",
" | ndarray.T : Array property returning the array transposed.\n",
" | ndarray.reshape : Give a new shape to an array without changing its
data.\n",
" | \n",
" | Examples\n",
" | --------\n",
" | >>> a = np.array([[1, 2], [3, 4]])\n",
" | >>> a\n",
" | array([[1, 2],\n",
" | [3, 4]])\n",
" | >>> a.transpose()\n",
" | array([[1, 3],\n",
" | [2, 4]])\n",
" | >>> a.transpose((1, 0))\n",
" | array([[1, 3],\n",
" | [2, 4]])\n",
" | >>> a.transpose(1, 0)\n",
" | array([[1, 3],\n",
" | [2, 4]])\n",
" | \n",
" | >>> a = np.array([1, 2, 3, 4])\n",
" | >>> a\n",
" | array([1, 2, 3, 4])\n",
" | array([1, 2, 3, 4])\n",
" | \n",
" | var(...)\n",
" | a.var(axis=None, dtype=None, out=None, ddof=0, keepdims=False, *,
where=True)\n",
" | \n",
" | Returns the variance of the array elements, along given axis.\n",
" | \n",
" | Refer to `numpy.var` for full documentation.\n",
" | \n",
" | See Also\n",
" | --------\n",
" | numpy.var : equivalent function\n",
" | \n",
" | view(...)\n",
" | a.view([dtype][, type])\n",
" | \n",
" | New view of array with the same data.\n",
" | \n",
" | .. note::\n",
" | Passing None for ``dtype`` is different from omitting the
parameter,\n",
" | since the former invokes ``dtype(None)`` which is an alias for\
n",
" | ``dtype('float_')``.\n",
" | \n",
" | Parameters\n",
" | ----------\n",
" | dtype : data-type or ndarray sub-class, optional\n",
" | Data-type descriptor of the returned view, e.g., float32 or
int16.\n",
" | Omitting it results in the view having the same data-type as
à`.\n",
" | This argument can also be specified as an ndarray sub-class,
which\n",
" | then specifies the type of the returned object (this is
equivalent to\n",
" | setting the ``type`` parameter).\n",
" | type : Python type, optional\n",
" | Type of the returned view, e.g., ndarray or matrix. Again,
omission\n",
" | of the parameter results in type preservation.\n",
" | \n",
" | Notes\n",
" | -----\n",
" | `à.view()`` is used two different ways:\n",
" | \n",
" | `à.view(some_dtype)`` or `à.view(dtype=some_dtype)`` constructs a
view\n",
" | of the array's memory with a different data-type. This can cause a\
n",
" | reinterpretation of the bytes of memory.\n",
" | \n",
" | `à.view(ndarray_subclass)`` or `à.view(type=ndarray_subclass)``
just\n",
" | returns an instance of `ndarray_subclass` that looks at the same
array\n",
" | (same shape, dtype, etc.) This does not cause a reinterpretation of
the\n",
" | memory.\n",
" | \n",
" | For `à.view(some_dtype)``, if ``some_dtype`` has a different number
of\n",
" | bytes per entry than the previous dtype (for example, converting a
regular\n",
" | array to a structured array), then the last axis of `à`` must be\
n",
" | contiguous. This axis will be resized in the result.\n",
" | \n",
" | Only the last axis needs to be contiguous. Previously, the entire
array\n",
" | had to be C-contiguous.\n",
" | \n",
" | Examples\n",
" | --------\n",
" | >>> x = np.array([(1, 2)], dtype=[('a', np.int8), ('b', np.int8)])\
n",
" | \n",
" | Viewing array data using a different type and dtype:\n",
" | \n",
" | >>> y = x.view(dtype=np.int16, type=np.matrix)\n",
" | >>> y\n",
" | matrix([[513]], dtype=int16)\n",
" | >>> print(type(y))\n",
" | <class 'numpy.matrix'>\n",
" | \n",
" | Creating a view on a structured array so it can be used in
calculations\n",
" | \n",
" | >>> x = np.array([(1, 2),(3,4)], dtype=[('a', np.int8), ('b',
np.int8)])\n",
" | >>> xv = x.view(dtype=np.int8).reshape(-1,2)\n",
" | >>> xv\n",
" | array([[1, 2],\n",
" | [3, 4]], dtype=int8)\n",
" | >>> xv.mean(0)\n",
" | array([2., 3.])\n",
" | \n",
" | Making changes to the view changes the underlying array\n",
" | \n",
" | >>> xv[0,1] = 20\n",
" | >>> x\n",
" | array([(1, 20), (3, 4)], dtype=[('a', 'i1'), ('b', 'i1')])\n",
" | \n",
" | Using a view to convert an array to a recarray:\n",
" | \n",
" | >>> z = x.view(np.recarray)\n",
" | >>> z.a\n",
" | array([1, 3], dtype=int8)\n",
" | \n",
" | Views share data:\n",
" | \n",
" | >>> x[0] = (9, 10)\n",
" | >>> z[0]\n",
" | (9, 10)\n",
" | \n",
" | Views that change the dtype size (bytes per entry) should normally
be\n",
" | avoided on arrays defined by slices, transposes, fortran-ordering,
etc.:\n",
" | \n",
" | >>> x = np.array([[1, 2, 3], [4, 5, 6]], dtype=np.int16)\n",
" | >>> y = x[:, ::2]\n",
" | >>> y\n",
" | array([[1, 3],\n",
" | [4, 6]], dtype=int16)\n",
" | >>> y.view(dtype=[('width', np.int16), ('length', np.int16)])\n",
" | ...\n",
" | ValueError: To change to a dtype of a different size, the last axis
must be contiguous\n",
" | >>> z = y.copy()\n",
" | >>> z.view(dtype=[('width', np.int16), ('length', np.int16)])\n",
" | array([[(1, 3)],\n",
" | [(4, 6)]], dtype=[('width', '<i2'), ('length', '<i2')])\n",
" | \n",
" | However, views that change dtype are totally fine for arrays with a\
n",
" | contiguous last axis, even if the rest of the axes are not C-
contiguous:\n",
" | \n",
" | >>> x = np.arange(2 * 3 * 4, dtype=np.int8).reshape(2, 3, 4)\n",
" | >>> x.transpose(1, 0, 2).view(np.int16)\n",
" | array([[[ 256, 770],\n",
" | [3340, 3854]],\n",
" | <BLANKLINE>\n",
" | [[1284, 1798],\n",
" | [4368, 4882]],\n",
" | <BLANKLINE>\n",
" | [[2312, 2826],\n",
" | [5396, 5910]]], dtype=int16)\n",
" | \n",
" | ----------------------------------------------------------------------\
n",
" | Class methods defined here:\n",
" | \n",
" | __class_getitem__(...) from builtins.type\n",
" | a.__class_getitem__(item, /)\n",
" | \n",
" | Return a parametrized wrapper around the `~numpy.ndarray` type.\n",
" | \n",
" | .. versionadded:: 1.22\n",
" | \n",
" | Returns\n",
" | -------\n",
" | alias : types.GenericAlias\n",
" | A parametrized `~numpy.ndarray` type.\n",
" | \n",
" | Examples\n",
" | --------\n",
" | >>> from typing import Any\n",
" | >>> import numpy as np\n",
" | \n",
" | >>> np.ndarray[Any, np.dtype[Any]]\n",
" | numpy.ndarray[typing.Any, numpy.dtype[typing.Any]]\n",
" | \n",
" | Notes\n",
" | -----\n",
" | This method is only available for python 3.9 and later.\n",
" | \n",
" | See Also\n",
" | --------\n",
" | :pep:`585` : Type hinting generics in standard collections.\n",
" | numpy.typing.NDArray : An ndarray alias :term:`generic <generic
type>`\n",
" | \n",
" | ----------------------------------------------------------------------\
n",
" | Static methods defined here:\n",
" | \n",
" | __new__(*args, **kwargs) from builtins.type\n",
" | Create and return a new object. See help(type) for accurate
signature.\n",
" | \n",
" | ----------------------------------------------------------------------\
n",
" | Data descriptors defined here:\n",
" | \n",
" | T\n",
" | View of the transposed array.\n",
" | \n",
" | Same as ``self.transpose()``.\n",
" | \n",
" | Examples\n",
" | --------\n",
" | >>> a = np.array([[1, 2], [3, 4]])\n",
" | >>> a\n",
" | array([[1, 2],\n",
" | [3, 4]])\n",
" | >>> a.T\n",
" | array([[1, 3],\n",
" | [2, 4]])\n",
" | \n",
" | >>> a = np.array([1, 2, 3, 4])\n",
" | >>> a\n",
" | array([1, 2, 3, 4])\n",
" | >>> a.T\n",
" | array([1, 2, 3, 4])\n",
" | \n",
" | See Also\n",
" | --------\n",
" | transpose\n",
" | \n",
" | __array_interface__\n",
" | Array protocol: Python side.\n",
" | \n",
" | __array_priority__\n",
" | Array priority.\n",
" | \n",
" | __array_struct__\n",
" | Array protocol: C-struct side.\n",
" | \n",
" | base\n",
" | Base object if memory is from some other object.\n",
" | \n",
" | Examples\n",
" | --------\n",
" | The base of an array that owns its memory is None:\n",
" | \n",
" | >>> x = np.array([1,2,3,4])\n",
" | >>> x.base is None\n",
" | True\n",
" | \n",
" | Slicing creates a view, whose memory is shared with x:\n",
" | \n",
" | >>> y = x[2:]\n",
" | >>> y.base is x\n",
" | True\n",
" | \n",
" | ctypes\n",
" | An object to simplify the interaction of the array with the ctypes\
n",
" | module.\n",
" | \n",
" | This attribute creates an object that makes it easier to use arrays\
n",
" | when calling shared libraries with the ctypes module. The returned\
n",
" | object has, among others, data, shape, and strides attributes (see\
n",
" | Notes below) which themselves return ctypes objects that can be
used\n",
" | as arguments to a shared library.\n",
" | \n",
" | Parameters\n",
" | ----------\n",
" | None\n",
" | \n",
" | Returns\n",
" | -------\n",
" | c : Python object\n",
" | Possessing attributes data, shape, strides, etc.\n",
" | \n",
" | See Also\n",
" | --------\n",
" | numpy.ctypeslib\n",
" | \n",
" | Notes\n",
" | -----\n",
" | Below are the public attributes of this object which were
documented\n",
" | in \"Guide to NumPy\" (we have omitted undocumented public
attributes,\n",
" | as well as documented private attributes):\n",
" | \n",
" | .. autoattribute:: numpy.core._internal._ctypes.data\n",
" | :noindex:\n",
" | \n",
" | .. autoattribute:: numpy.core._internal._ctypes.shape\n",
" | :noindex:\n",
" | \n",
" | .. autoattribute:: numpy.core._internal._ctypes.strides\n",
" | :noindex:\n",
" | \n",
" | .. automethod:: numpy.core._internal._ctypes.data_as\n",
" | :noindex:\n",
" | \n",
" | .. automethod:: numpy.core._internal._ctypes.shape_as\n",
" | :noindex:\n",
" | \n",
" | .. automethod:: numpy.core._internal._ctypes.strides_as\n",
" | :noindex:\n",
" | \n",
" | If the ctypes module is not available, then the ctypes attribute\n",
" | of array objects still returns something useful, but ctypes objects\
n",
" | are not returned and errors may be raised instead. In particular,\
n",
" | the object will still have the `às_parameter`` attribute which
will\n",
" | return an integer equal to the data attribute.\n",
" | \n",
" | Examples\n",
" | --------\n",
" | >>> import ctypes\n",
" | >>> x = np.array([[0, 1], [2, 3]], dtype=np.int32)\n",
" | >>> x\n",
" | array([[0, 1],\n",
" | [2, 3]], dtype=int32)\n",
" | >>> x.ctypes.data\n",
" | 31962608 # may vary\n",
" | >>> x.ctypes.data_as(ctypes.POINTER(ctypes.c_uint32))\n",
" | <__main__.LP_c_uint object at 0x7ff2fc1fc200> # may vary\n",
" | >>> x.ctypes.data_as(ctypes.POINTER(ctypes.c_uint32)).contents\n",
" | c_uint(0)\n",
" | c_ulong(4294967296)\n",
" | >>> x.ctypes.shape\n",
" | <numpy.core._internal.c_long_Array_2 object at 0x7ff2fc1fce60> # may
vary\n",
" | >>> x.ctypes.strides\n",
" | <numpy.core._internal.c_long_Array_2 object at 0x7ff2fc1ff320> # may
vary\n",
" | \n",
" | data\n",
" | Python buffer object pointing to the start of the array's data.\n",
" | \n",
" | dtype\n",
" | Data-type of the array's elements.\n",
" | \n",
" | .. warning::\n",
" | \n",
" | Setting `àrr.dtype`` is discouraged and may be deprecated in
the\n",
" | future. Setting will replace the ``dtype`` without modifying
the\n",
" | memory (see also `ndarray.view` and `ndarray.astype`).\n",
" | \n",
" | Parameters\n",
" | ----------\n",
" | None\n",
" | \n",
" | Returns\n",
" | -------\n",
" | d : numpy dtype object\n",
" | \n",
" | See Also\n",
" | --------\n",
" | ndarray.astype : Cast the values contained in the array to a new
data-type.\n",
" | ndarray.view : Create a view of the same data but a different data-
type.\n",
" | numpy.dtype\n",
" | \n",
" | Examples\n",
" | --------\n",
" | >>> x\n",
" | array([[0, 1],\n",
" | [2, 3]])\n",
" | >>> x.dtype\n",
" | dtype('int32')\n",
" | >>> type(x.dtype)\n",
" | <type 'numpy.dtype'>\n",
" | \n",
" | flags\n",
" | Information about the memory layout of the array.\n",
" | \n",
" | Attributes\n",
" | ----------\n",
" | C_CONTIGUOUS (C)\n",
" | The data is in a single, C-style contiguous segment.\n",
" | F_CONTIGUOUS (F)\n",
" | The data is in a single, Fortran-style contiguous segment.\n",
" | OWNDATA (O)\n",
" | The array owns the memory it uses or borrows it from another
object.\n",
" | WRITEABLE (W)\n",
" | The data area can be written to. Setting this to False locks\
n",
" | the data, making it read-only. A view (slice, etc.) inherits
WRITEABLE\n",
" | from its base array at creation time, but a view of a writeable\
n",
" | array may be subsequently locked while the base array remains
writeable.\n",
" | (The opposite is not true, in that a view of a locked array may
not\n",
" | be made writeable. However, currently, locking a base object
does not\n",
" | lock any views that already reference it, so under that
circumstance it\n",
" | is possible to alter the contents of a locked array via a
previously\n",
" | created writeable view onto it.) Attempting to change a non-
writeable\n",
" | array raises a RuntimeError exception.\n",
" | ALIGNED (A)\n",
" | The data and all elements are aligned appropriately for the
hardware.\n",
" | WRITEBACKIFCOPY (X)\n",
" | This array is a copy of some other array. The C-API function\n",
" | PyArray_ResolveWritebackIfCopy must be called before
deallocating\n",
" | to the base array will be updated with the contents of this
array.\n",
" | FNC\n",
" | F_CONTIGUOUS and not C_CONTIGUOUS.\n",
" | FORC\n",
" | F_CONTIGUOUS or C_CONTIGUOUS (one-segment test).\n",
" | BEHAVED (B)\n",
" | ALIGNED and WRITEABLE.\n",
" | CARRAY (CA)\n",
" | BEHAVED and C_CONTIGUOUS.\n",
" | FARRAY (FA)\n",
" | BEHAVED and F_CONTIGUOUS and not C_CONTIGUOUS.\n",
" | \n",
" | Notes\n",
" | -----\n",
" | The `flags` object can be accessed dictionary-like (as in
`à.flags['WRITEABLE']``),\n",
" | or by using lowercased attribute names (as in
`à.flags.writeable``). Short flag\n",
" | names are only supported in dictionary access.\n",
" | \n",
" | Only the WRITEBACKIFCOPY, WRITEABLE, and ALIGNED flags can be\n",
" | changed by the user, via direct assignment to the attribute or
dictionary\n",
" | entry, or by calling `ndarray.setflags`.\n",
" | \n",
" | The array flags cannot be set arbitrarily:\n",
" | \n",
" | - WRITEBACKIFCOPY can only be set ``False``.\n",
" | - ALIGNED can only be set ``True`` if the data is truly aligned.\n",
" | - WRITEABLE can only be set ``True`` if the array owns its own
memory\n",
" | or the ultimate owner of the memory exposes a writeable buffer\n",
" | interface or is a string.\n",
" | \n",
" | Arrays can be both C-style and Fortran-style contiguous
simultaneously.\n",
" | This is clear for 1-dimensional arrays, but can also be true for
higher\n",
" | dimensional arrays.\n",
" | \n",
" | Even for contiguous arrays a stride for a given dimension\n",
" | `àrr.strides[dim]`` may be *arbitrary* if `àrr.shape[dim] == 1``\
n",
" | or the array has no elements.\n",
" | It does *not* generally hold that ``self.strides[-1] ==
self.itemsize``\n",
" | for C-style contiguous arrays or ``self.strides[0] ==
self.itemsize`` for\n",
" | Fortran-style contiguous arrays is true.\n",
" | \n",
" | flat\n",
" | A 1-D iterator over the array.\n",
" | \n",
" | This is a `numpy.flatiter` instance, which acts similarly to, but is
not\n",
" | a subclass of, Python's built-in iterator object.\n",
" | \n",
" | See Also\n",
" | --------\n",
" | flatten : Return a copy of the array collapsed into one dimension.\
n",
" | \n",
" | flatiter\n",
" | \n",
" | Examples\n",
" | --------\n",
" | >>> x = np.arange(1, 7).reshape(2, 3)\n",
" | >>> x\n",
" | array([[1, 2, 3],\n",
" | [4, 5, 6]])\n",
" | >>> x.flat[3]\n",
" | 4\n",
" | >>> x.T\n",
" | array([[1, 4],\n",
" | [2, 5],\n",
" | [3, 6]])\n",
" | >>> x.T.flat[3]\n",
" | 5\n",
" | >>> type(x.flat)\n",
" | <class 'numpy.flatiter'>\n",
" | \n",
" | An assignment example:\n",
" | \n",
" | >>> x.flat = 3; x\n",
" | array([[3, 3, 3],\n",
" | [3, 3, 3]])\n",
" | >>> x.flat[[1,4]] = 1; x\n",
" | array([[3, 1, 3],\n",
" | [3, 1, 3]])\n",
" | \n",
" | imag\n",
" | The imaginary part of the array.\n",
" | \n",
" | Examples\n",
" | --------\n",
" | >>> x = np.sqrt([1+0j, 0+1j])\n",
" | >>> x.imag\n",
" | array([ 0. , 0.70710678])\n",
" | >>> x.imag.dtype\n",
" | dtype('float64')\n",
" | \n",
" | itemsize\n",
" | Length of one array element in bytes.\n",
" | \n",
" | Examples\n",
" | --------\n",
" | >>> x = np.array([1,2,3], dtype=np.float64)\n",
" | >>> x.itemsize\n",
" | 8\n",
" | >>> x = np.array([1,2,3], dtype=np.complex128)\n",
" | 16\n",
" | \n",
" | nbytes\n",
" | Total bytes consumed by the elements of the array.\n",
" | \n",
" | Notes\n",
" | -----\n",
" | Does not include memory consumed by non-element attributes of the\
n",
" | array object.\n",
" | \n",
" | Examples\n",
" | --------\n",
" | >>> x = np.zeros((3,5,2), dtype=np.complex128)\n",
" | >>> x.nbytes\n",
" | 480\n",
" | >>> np.prod(x.shape) * x.itemsize\n",
" | 480\n",
" | \n",
" | ndim\n",
" | Number of array dimensions.\n",
" | \n",
" | Examples\n",
" | --------\n",
" | >>> x = np.array([1, 2, 3])\n",
" | >>> x.ndim\n",
" | 1\n",
" | >>> y = np.zeros((2, 3, 4))\n",
" | >>> y.ndim\n",
" | 3\n",
" | \n",
" | real\n",
" | The real part of the array.\n",
" | \n",
" | Examples\n",
" | --------\n",
" | >>> x = np.sqrt([1+0j, 0+1j])\n",
" | >>> x.real\n",
" | array([ 1. , 0.70710678])\n",
" | >>> x.real.dtype\n",
" | \n",
" | See Also\n",
" | --------\n",
" | numpy.real : equivalent function\n",
" | \n",
" | shape\n",
" | Tuple of array dimensions.\n",
" | \n",
" | The shape property is usually used to get the current shape of an
array,\n",
" | but may also be used to reshape the array in-place by assigning a
tuple of\n",
" | array dimensions to it. As with `numpy.reshape`, one of the new
shape\n",
" | dimensions can be -1, in which case its value is inferred from the
size of\n",
" | the array and the remaining dimensions. Reshaping an array in-place
will\n",
" | fail if a copy is required.\n",
" | \n",
" | \n",
" | Setting `àrr.shape`` is discouraged and may be deprecated in
the\n",
" | future. Using `ndarray.reshape` is the preferred approach.\n",
" | \n",
" | Examples\n",
" | --------\n",
" | >>> x = np.array([1, 2, 3, 4])\n",
" | >>> x.shape\n",
" | (4,)\n",
" | >>> y = np.zeros((2, 3, 4))\n",
" | >>> y.shape\n",
" | (2, 3, 4)\n",
" | >>> y.shape = (3, 8)\n",
" | >>> y\n",
" | array([[ 0., 0., 0., 0., 0., 0., 0., 0.],\n",
" | [ 0., 0., 0., 0., 0., 0., 0., 0.],\n",
" | [ 0., 0., 0., 0., 0., 0., 0., 0.]])\n",
" | >>> y.shape = (3, 6)\n",
" | ValueError: total size of new array must be unchanged\n",
" | >>> np.zeros((4,2))[::2].shape = (-1,)\n",
" | AttributeError: Incompatible shape for in-place modification. Use\
n",
" | `.reshape()` to make a copy with the desired shape.\n",
" | \n",
" | See Also\n",
" | --------\n",
" | numpy.shape : Equivalent getter function.\n",
" | numpy.reshape : Function similar to setting ``shape``.\n",
" | ndarray.reshape : Method similar to setting ``shape``.\n",
" | \n",
" | size\n",
" | \n",
" | Equal to ``np.prod(a.shape)``, i.e., the product of the array's\n",
" | dimensions.\n",
" | \n",
" | Notes\n",
" | -----\n",
" | à.size` returns a standard arbitrary precision Python integer.
This\n",
" | may not be the case with other methods of obtaining the same value\
n",
" | (like the suggested ``np.prod(a.shape)``, which returns an instance\
n",
" | of ``np.int_``), and may be relevant if the value is used further
in\n",
" | calculations that may overflow a fixed size integer type.\n",
" | \n",
" | Examples\n",
" | --------\n",
" | >>> x = np.zeros((3, 5, 2), dtype=np.complex128)\n",
" | >>> x.size\n",
" | 30\n",
" | >>> np.prod(x.shape)\n",
" | 30\n",
" | \n",
" | strides\n",
" | Tuple of bytes to step in each dimension when traversing an array.\
n",
" | \n",
" | The byte offset of element ``(i[0], i[1], ..., i[n])`` in an array
à`\n",
" | is::\n",
" | \n",
" | offset = sum(np.array(i) * a.strides)\n",
" | \n",
" | A more detailed explanation of strides can be found in the\n",
" | \"ndarray.rst\" file in the NumPy reference guide.\n",
" | \n",
" | \n",
" | Setting `àrr.strides`` is discouraged and may be deprecated in
the\n",
" | future. `numpy.lib.stride_tricks.as_strided` should be
preferred\n",
" | to create a new view of the same data in a safer way.\n",
" | \n",
" | Notes\n",
" | -----\n",
" | Imagine an array of 32-bit integers (each 4 bytes)::\n",
" | \n",
" | x = np.array([[0, 1, 2, 3, 4],\n",
" | [5, 6, 7, 8, 9]], dtype=np.int32)\n",
" | \n",
" | This array is stored in memory as 40 bytes, one after the other\n",
" | (known as a contiguous block of memory). The strides of an array
tell\n",
" | us how many bytes we have to skip in memory to move to the next
position\n",
" | along a certain axis. For example, we have to skip 4 bytes (1
value) to\n",
" | move to the next column, but 20 bytes (5 values) to get to the same\
n",
" | position in the next row. As such, the strides for the array `x`
will be\n",
" | ``(20, 4)``.\n",
" | \n",
" | See Also\n",
" | --------\n",
" | numpy.lib.stride_tricks.as_strided\n",
" | \n",
" | Examples\n",
" | --------\n",
" | >>> y = np.reshape(np.arange(2*3*4), (2,3,4))\n",
" | >>> y\n",
" | array([[[ 0, 1, 2, 3],\n",
" | [ 4, 5, 6, 7],\n",
" | [ 8, 9, 10, 11]],\n",
" | [[12, 13, 14, 15],\n",
" | [16, 17, 18, 19],\n",
" | [20, 21, 22, 23]]])\n",
" | >>> y.strides\n",
" | (48, 16, 4)\n",
" | >>> y[1,1,1]\n",
" | 17\n",
" | >>> offset=sum(y.strides * np.array((1,1,1)))\n",
" | >>> offset/y.itemsize\n",
" | 17\n",
" | \n",
" | >>> x = np.reshape(np.arange(5*6*7*8),
(5,6,7,8)).transpose(2,3,1,0)\n",
" | >>> x.strides\n",
" | (32, 4, 224, 1344)\n",
" | >>> i = np.array([3,5,2,2])\n",
" | >>> offset = sum(i * x.strides)\n",
" | >>> x[3,5,2,2]\n",
" | 813\n",
" | >>> offset / x.itemsize\n",
" | 813\n",
" | \n",
" | ----------------------------------------------------------------------\
n",
" | Data and other attributes defined here:\n",
" | \n",
" | __hash__ = None\n",
"\n"
]
}
],
"source": [
"help(a)"
]
},
{
"metadata": {},
"source": [
"we can create an array from a list"
]
},
{
"metadata": {},
"outputs": [
{
"name": "stdout",
"text": [
"[1. 2. 3. 4.]\n",
"float64\n",
"<class 'numpy.ndarray'>\n"
]
}
],
"source": [
"b = np.array([1.0,2.0,3.0,4.0])\n",
"print(b)\n",
"print(b.dtype)\n",
"print(type(b))"
]
},
{
"metadata": {},
"source": [
"we can create a multi-dimensional array of a specified size initialized all to
0 easily. There is also an analogous ones() and empty() array routine. Note that
here we explicitly set the datatype for the array. \n",
"\n",
"Unlike lists in python, all of the elements of a numpy array are of the same
datatype"
]
},
{
"metadata": {},
"outputs": [
{
"data": {
"text/plain": [
"array([[1., 0., 0., 0., 0., 0., 0., 0., 0., 0.],\n",
" [0., 1., 0., 0., 0., 0., 0., 0., 0., 0.],\n",
" [0., 0., 1., 0., 0., 0., 0., 0., 0., 0.],\n",
" [0., 0., 0., 1., 0., 0., 0., 0., 0., 0.],\n",
" [0., 0., 0., 0., 1., 0., 0., 0., 0., 0.],\n",
" [0., 0., 0., 0., 0., 1., 0., 0., 0., 0.],\n",
" [0., 0., 0., 0., 0., 0., 1., 0., 0., 0.],\n",
" [0., 0., 0., 0., 0., 0., 0., 1., 0., 0.],\n",
" [0., 0., 0., 0., 0., 0., 0., 0., 1., 0.],\n",
" [0., 0., 0., 0., 0., 0., 0., 0., 0., 1.]])"
]
},
"metadata": {},
}
],
"source": [
"c = np.eye(10)\n",
"dtype=np.float64\n",
"c"
]
},
{
"metadata": {},
"source": [
"`linspace` (and `logspace`) create arrays with evenly space (in log) numbers.
For `logspace`, you specify the start and ending powers (`base**start` to
`base**stop`)"
]
},
{
"metadata": {},
"outputs": [
{
"name": "stdout",
"text": [
"[-1. -0.78571429 -0.57142857 -0.35714286 -0.14285714 0.07142857\n",
" 0.28571429 0.5 0.71428571 0.92857143 1.14285714 1.35714286\n",
" 1.57142857 1.78571429 2. ]\n"
]
}
],
"source": [
"d = np.linspace(-1, 2, 15, endpoint=True)\n",
"print(d)"
]
},
{
"metadata": {},
"outputs": [
{
"name": "stdout",
"text": [
"[ 0.1 0.16378937 0.26826958 0.43939706 0.71968567\n",
" 1.17876863 1.93069773 3.16227766 5.17947468 8.48342898\n",
" 13.89495494 22.75845926 37.2759372 61.05402297 100. ]\n"
]
}
],
"source": [
"e = np.logspace(-1, 2, 15, endpoint=True, base=10)\n",
"print(e)"
]
},
{
"metadata": {},
"source": [
"As always, as for help -- the numpy functions have very nice docstrings"
]
},
{
"metadata": {},
"outputs": [
{
"name": "stdout",
"text": [
"Help on function logspace in module numpy:\n",
"\n",
"logspace(start, stop, num=50, endpoint=True, base=10.0, dtype=None, axis=0)\
n",
" Return numbers spaced evenly on a log scale.\n",
" \n",
" In linear space, the sequence starts at ``base ** start``\n",
" (`base` to the power of `start`) and ends with ``base ** stop``\n",
" (see èndpoint` below).\n",
" \n",
" .. versionchanged:: 1.16.0\n",
" Non-scalar `start` and `stop` are now supported.\n",
" \n",
" Parameters\n",
" ----------\n",
" start : array_like\n",
" ``base ** start`` is the starting value of the sequence.\n",
" stop : array_like\n",
" ``base ** stop`` is the final value of the sequence, unless
èndpoint`\n",
" is False. In that case, ``num + 1`` values are spaced over the\n",
" interval in log-space, of which all but the last (a sequence of\n",
" length `num`) are returned.\n",
" num : integer, optional\n",
" Number of samples to generate. Default is 50.\n",
" endpoint : boolean, optional\n",
" If true, `stop` is the last sample. Otherwise, it is not included.\
n",
" Default is True.\n",
" base : array_like, optional\n",
" The base of the log space. The step size between the elements in\n",
" ``ln(samples) / ln(base)`` (or ``log_base(samples)``) is uniform.\
n",
" Default is 10.0.\n",
" dtype : dtype\n",
" The type of the output array. If `dtype` is not given, the data
type\n",
" is inferred from `start` and `stop`. The inferred type will never
be\n",
" an integer; `float` is chosen even if the arguments would produce
an\n",
" array of integers.\n",
" axis : int, optional\n",
" The axis in the result to store the samples. Relevant only if
start\n",
" or stop are array-like. By default (0), the samples will be along
a\n",
" new axis inserted at the beginning. Use -1 to get an axis at the
end.\n",
" \n",
" .. versionadded:: 1.16.0\n",
" \n",
" \n",
" Returns\n",
" -------\n",
" samples : ndarray\n",
" `num` samples, equally spaced on a log scale.\n",
" \n",
" See Also\n",
" --------\n",
" arange : Similar to linspace, with the step size specified instead of
the\n",
" number of samples. Note that, when used with a float endpoint,
the\n",
" endpoint may or may not be included.\n",
" linspace : Similar to logspace, but with the samples uniformly
distributed\n",
" in linear space, instead of log space.\n",
" geomspace : Similar to logspace, but with endpoints specified directly.\
n",
" :ref:`how-to-partition`\n",
" \n",
" Notes\n",
" -----\n",
" Logspace is equivalent to the code\n",
" \n",
" >>> y = np.linspace(start, stop, num=num, endpoint=endpoint)\n",
" ... # doctest: +SKIP\n",
" >>> power(base, y).astype(dtype)\n",
" ... # doctest: +SKIP\n",
" \n",
" Examples\n",
" --------\n",
" >>> np.logspace(2.0, 3.0, num=4)\n",
" array([ 100. , 215.443469 , 464.15888336, 1000. ])\n",
" >>> np.logspace(2.0, 3.0, num=4, endpoint=False)\n",
" array([100. , 177.827941 , 316.22776602, 562.34132519])\n",
" >>> np.logspace(2.0, 3.0, num=4, base=2.0)\n",
" array([4. , 5.0396842 , 6.34960421, 8. ])\n",
" \n",
" Graphical illustration:\n",
" \n",
" >>> import matplotlib.pyplot as plt\n",
" >>> N = 10\n",
" >>> x1 = np.logspace(0.1, 1, N, endpoint=True)\n",
" >>> x2 = np.logspace(0.1, 1, N, endpoint=False)\n",
" >>> y = np.zeros(N)\n",
" >>> plt.plot(x1, y, 'o')\n",
" [<matplotlib.lines.Line2D object at 0x...>]\n",
" >>> plt.plot(x2, y + 0.5, 'o')\n",
" [<matplotlib.lines.Line2D object at 0x...>]\n",
" >>> plt.ylim([-0.5, 1])\n",
" (-0.5, 1)\n",
" >>> plt.show()\n",
"\n"
]
}
],
"source": [
"help(np.logspace)"
]
},
{
"metadata": {},
"source": [
"we can also initialize an array based on a function"
]
},
{
"metadata": {},
"outputs": [
{
"data": {
"text/plain": [
"array([[ True, False, False],\n",
" [False, True, False],\n",
" [False, False, True]])"
]
},
"metadata": {},
}
],
"source": [
"f = np.fromfunction(lambda i, j: i == j, (3, 3), dtype=int)\n",
"f"
]
},
{
"metadata": {},
"source": [
"# Array Operations"
]
},
{
"metadata": {},
"source": [
"most operations (`+`, `-`, `*`, `/`) will work on an entire array at once,
element-by-element.\n",
"\n",
"Note that that the multiplication operator is not a matrix multiply (there is
a new operator in python 3.5+, `@`, to do matrix multiplicaiton.\n",
"\n",
"Let's create a simply array to start with"
]
},
{
"metadata": {},
"outputs": [
{
"name": "stdout",
"text": [
"[[ 0 1 2 3]\n",
" [ 4 5 6 7]\n",
" [ 8 9 10 11]]\n"
]
}
],
"source": [
"print(a)"
]
},
{
"metadata": {},
"source": [
"Multiplication by a scalar multiplies every element"
]
},
{
"metadata": {},
"outputs": [
{
"data": {
"text/plain": [
"array([[ 0, 2, 4, 6],\n",
" [ 8, 10, 12, 14],\n",
" [16, 18, 20, 22]])"
]
},
"metadata": {},
}
],
"source": [
"a*2"
]
},
{
"metadata": {},
"source": [
"adding two arrays adds element-by-element"
]
},
{
"metadata": {},
"outputs": [
{
"data": {
"text/plain": [
"array([[ 0, 2, 4, 6],\n",
" [ 8, 10, 12, 14],\n",
" [16, 18, 20, 22]])"
]
},
"metadata": {},
}
],
"source": [
"a+a"
]
},
{
"metadata": {},
"source": [
"multiplying two arrays multiplies element-by-element"
]
},
{
"metadata": {},
"outputs": [
{
"data": {
"text/plain": [
"array([[ 0, 1, 4, 9],\n",
" [ 16, 25, 36, 49],\n",
" [ 64, 81, 100, 121]])"
]
},
"metadata": {},
}
],
"source": [
"a*a"
]
},
{
"metadata": {},
"source": [
"We can think of our 2-d array a was a 3 x 5 matrix (3 rows, 5 columns). We
can take the transpose to geta 5 x 3 matrix, and then we can do a matrix
multiplication"
]
},
{
"metadata": {},
"outputs": [
{
"data": {
"text/plain": [
"array([[ 0, 4, 8],\n",
" [ 1, 5, 9],\n",
" [ 2, 6, 10],\n",
" [ 3, 7, 11]])"
]
},
"metadata": {},
}
],
"source": [
"b = a.transpose()\n",
"b"
]
},
{
"metadata": {},
"outputs": [
{
"data": {
"text/plain": [
"array([[ 14, 38, 62],\n",
" [ 38, 126, 214],\n",
" [ 62, 214, 366]])"
]
},
"metadata": {},
}
],
"source": [
"a@b"
]
},
{
"metadata": {},
"source": [
"We can sum along axes or the entire array"
]
},
{
"metadata": {},
"outputs": [
{
"name": "stdout",
"text": [
"[[ 0 1 2 3]\n",
" [ 4 5 6 7]\n",
" [ 8 9 10 11]]\n"
]
},
{
"data": {
"text/plain": [
"array([ 6, 22, 38])"
]
},
"metadata": {},
}
],
"source": [
"print(a)\n",
"a.sum(axis=1)"
]
},
{
"metadata": {},
"outputs": [
{
"data": {
"text/plain": [
"66"
]
},
"metadata": {},
}
],
"source": [
"a.sum()"
]
},
{
"metadata": {},
"source": [
"Also get the extrema"
]
},
{
"metadata": {},
"outputs": [
{
"name": "stdout",
"text": [
"0 11\n"
]
}
],
"source": [
"print(a.min(), a.max())"
]
},
{
"metadata": {},
"source": [
"### universal functions"
]
},
{
"metadata": {},
"source": [
"Up until now, we have been discussing some of the basic nuts and bolts of
NumPy; now, we will dive into the reasons that NumPy is so important in the Python
data science world.\n",
"Namely, it provides an easy and flexible interface to optimized computation
with arrays of data.\n",
"\n",
"Computation on NumPy arrays can be very fast, or it can be very slow.\n",
"The key to making it fast is to use *vectorized* operations, generally
implemented through NumPy's *universal functions* (ufuncs).\n",
"This section motivates the need for NumPy's ufuncs, which can be used to make
repeated calculations on array elements much more efficient.\n",
"It then introduces many of the most common and useful arithmetic ufuncs
available in the NumPy package."
]
},
{
"metadata": {},
"source": [
"universal functions work element-by-element. Let's create a new array scaled
by `pi`"
]
},
{
"metadata": {},
"outputs": [
{
"name": "stdout",
"text": [
"[[0. 0.02575076 0.05150152 0.07725228]\n",
" [0.10300304 0.1287538 0.15450456 0.18025532]\n",
" [0.20600608 0.23175684 0.25750759 0.28325835]]\n"
]
}
],
"source": [
"b = a*np.pi/122.0\n",
"print(b)\n"
]
},
{
"metadata": {},
"outputs": [
{
"name": "stdout",
"text": [
"[[1. 0.99966847 0.99867409 0.99701753]\n",
" [0.99469988 0.99172267 0.9880879 0.98379795]\n",
" [0.97885569 0.97326437 0.96702772 0.96014987]]\n"
]
}
],
"source": [
"c = np.cos(b)\n",
"print(c)"
]
},
{
"metadata": {},
"outputs": [],
"source": [
"d =b+c"
]
},
{
"metadata": {},
"outputs": [
{
"name": "stdout",
"text": [
"[[1. 1.02541923 1.05017561 1.0742698 ]\n",
" [1.09770291 1.12047647 1.14259245 1.16405327]\n",
" [1.18486176 1.20502121 1.22453532 1.24340823]]\n"
]
}
],
"source": [
"print(d)"
]
},
{
"metadata": {},
"source": [
"\n",
"## Array Slicing: Accessing Subarrays"
]
},
{
"metadata": {},
"source": [
"Just as we can use square brackets to access individual array elements, we can
also use them to access subarrays with the *slice* notation, marked by the colon
(``:``) character.\n",
"The NumPy slicing syntax follows that of the standard Python list; to access a
slice of an array ``x``, use this:\n",
"``` python\n",
"x[start:stop:step]\n",
"```\n",
"If any of these are unspecified, they default to the values ``start=0``,
``stop=``*``size of dimension``*, ``step=1``.\n",
"We'll take a look at accessing sub-arrays in one dimension and in multiple
dimensions."
]
},
{
"metadata": {},
"outputs": [
{
"name": "stdout",
"text": [
"[ 0 3 6 9 12 15 18 21 24 27 30 33 36 39 42 45 48 51\n",
" 54 57 60 63 66 69 72 75 78 81 84 87 90 93 96 99 102 105\n",
" 108 111 114 117 120 123 126 129 132 135 138 141 144 147 150 153 156 159\n",
" 162 165 168 171 174 177 180 183 186 189 192 195 198 201 204 207 210 213\n",
" 216 219 222 225 228 231 234 237 240 243 246 249 252 255 258 261 264 267\n",
" 270 273 276 279 282 285 288 291 294 297 300 303 306 309 312 315 318 321\n",
" 324 327 330 333 336 339 342 345 348 351 354 357 360 363 366 369 372 375\n",
" 378 381 384 387 390 393 396 399 402 405 408 411 414 417 420 423 426 429\n",
" 432 435 438 441 444 447 450 453 456 459 462 465 468 471 474 477 480 483\n",
" 486 489 492 495 498 501 504 507 510 513 516 519 522 525 528 531 534 537\n",
" 540 543 546 549 552 555 558 561 564 567 570 573 576 579 582 585 588 591\n",
" 594 597 600 603 606 609 612 615 618 621 624 627 630 633 636 639 642 645\n",
" 648 651 654 657 660 663 666 669 672 675 678 681 684 687 690 693 696 699\n",
" 702 705 708 711 714 717 720 723 726 729 732 735 738 741 744 747 750 753\n",
" 756 759 762 765 768 771 774 777 780 783 786 789 792 795 798 801 804 807\n",
" 810 813 816 819 822 825 828 831 834 837 840 843 846 849 852 855 858 861\n",
" 864 867 870 873 876 879 882 885 888 891 894 897 900 903 906 909 912 915\n",
" 918 921 924 927 930 933 936 939 942 945 948 951 954 957 960 963 966 969\n",
" 972 975 978 981 984 987 990 993 996 999]\n"
]
}
],
"source": [
"iron = np.arange(1000)\n",
"iron\n",
"even = iron[0:1000:3]\n",
"print(even)\n"
]
},
{
"metadata": {},
"source": []
},
{
"metadata": {},
"outputs": [
{
"data": {
"text/plain": [
"array([0, 1, 2, 3, 4, 5, 6, 7, 8])"
]
},
"metadata": {},
}
],
"source": [
"a"
]
},
{
"metadata": {},
"source": [
"Now look at accessing a single element vs. a range (using slicing)"
]
},
{
"metadata": {},
"source": [
"Giving a single (0-based) index just references a single value"
]
},
{
"metadata": {},
"outputs": [
{
"data": {
"text/plain": [
"3"
]
},
"metadata": {},
}
],
"source": [
"a[3]"
]
},
{
"metadata": {},
"outputs": [
{
"name": "stdout",
"text": [
"[2]\n"
]
}
],
"source": [
"print(a[2:3])"
]
},
{
"metadata": {},
"outputs": [
{
"data": {
"text/plain": [
"array([2])"
]
},
"metadata": {},
}
],
"source": [
"a[2:3]"
]
},
{
"metadata": {},
"outputs": [
{
"data": {
"text/plain": [
"array([0, 1, 2, 3, 4, 5, 6, 7, 8])"
]
},
"metadata": {},
}
],
"source": [
"a[:]"
]
},
{
"metadata": {},
"source": [
"## Multidimensional Arrays\n",
"\n",
"Multidimensional arrays are stored in a contiguous space in memory -- this
means that the columns / rows need to be unraveled (flattened) so that it can be
thought of as a single one-dimensional array. Different programming languages do
this via different conventions:\n",
"\n",
"\n",
"Storage order:\n",
"\n",
"* Python/C use *row-major* storage: rows are stored one after the other\n",
"* Fortran/matlab use *column-major* storage: columns are stored one after
another\n",
"\n",
"The ordering matters when \n",
"\n",
"* passing arrays between languages (we'll talk about this later this
semester)\n",
"* looping over arrays -- you want to access elements that are next to one-
another in memory\n",
" * e.g, in Fortran:\n",
" <pre>\n",
" double precision :: A(M,N)\n",
" do j = 1, N\n",
" do i = 1, M\n",
" A(i,j) = …\n",
" enddo\n",
" enddo\n",
" </pre>\n",
" \n",
" * in C\n",
" <pre>\n",
" double A[M][N];\n",
" for (i = 0; i < M; i++) {\n",
" for (j = 0; j < N; j++) {\n",
" A[i][j] = …\n",
" }\n",
" } \n",
" </pre>\n",
" \n",
"\n",
"In python, using NumPy, we'll try to avoid explicit loops over elements as
much as possible\n",
"\n",
"Let's look at multidimensional arrays:"
]
},
{
"metadata": {},
"outputs": [
{
"data": {
"text/plain": [
"array([[ 0, 1, 2, 3, 4],\n",
" [ 5, 6, 7, 8, 9],\n",
" [10, 11, 12, 13, 14]])"
]
},
"metadata": {},
}
],
"source": [
"a"
]
},
{
"metadata": {},
"source": [
"Notice that the output of à` shows the row-major storage. The rows are
grouped together in the inner `[...]`\n",
"\n",
"Giving a single index (0-based) for each dimension just references a single
value in the array"
]
},
{
"metadata": {},
"outputs": [
{
"data": {
"text/plain": [
"6"
]
},
"metadata": {},
}
],
"source": [
"a[1,1]"
]
},
{
"metadata": {},
"source": [
"Doing slices will access a range of elements. Think of the start and stop in
the slice as referencing the left-edge of the slots in the array."
]
},
{
"metadata": {},
"outputs": [
{
"data": {
"text/plain": [
"array([[0, 1],\n",
" [5, 6]])"
]
},
"metadata": {},
}
],
"source": [
"a[0:2,0:2]"
]
},
{
"metadata": {},
"source": [
"Access a specific column"
]
},
{
"metadata": {},
"outputs": [
{
"data": {
"text/plain": [
"array([ 1, 6, 11])"
]
},
"metadata": {},
}
],
"source": [
"a[:,1]"
]
},
{
"metadata": {},
"source": [
"Sometimes we want a one-dimensional view into the array -- here we see the
memory layout (row-major) more explicitly"
]
},
{
"metadata": {},
"outputs": [
{
"name": "stdout",
"text": [
"[ 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14]\n"
]
}
],
"source": [
"a = a.flatten()\n",
"print(a)"
]
},
{
"metadata": {},
"source": [
"we can also iterate -- this is done over the first axis (rows)"
]
},
{
"metadata": {},
"outputs": [
{
"name": "stdout",
"text": [
"0\n",
"1\n",
"2\n",
"3\n",
"4\n",
"5\n",
"6\n",
"7\n",
"8\n",
"9\n",
"10\n",
"11\n",
"12\n",
"13\n",
"14\n",
"[[ 0 1 2 3 4]\n",
" [ 5 6 7 8 9]\n",
" [10 11 12 13 14]]\n"
]
}
],
"source": [
"\n",
"#print (a)\n",
"for tajmahal in a:\n",
" print(tajmahal)\n",
"print(a)"
]
},
{
"metadata": {},
"source": [
"or element by element"
]
},
{
"metadata": {},
"outputs": [
{
"name": "stdout",
"text": [
"0\n",
"1\n",
"2\n",
"3\n",
"4\n",
"5\n",
"6\n",
"7\n",
"8\n",
"9\n",
"10\n",
"11\n",
"12\n",
"13\n",
"14\n"
]
}
],
"source": [
"for e in a.flat:\n",
" print(e)"
]
},
{
"metadata": {},
"outputs": [
{
"name": "stdout",
"text": [
"Help on ndarray object:\n",
"\n",
"class ndarray(builtins.object)\n",
" | ndarray(shape, dtype=float, buffer=None, offset=0,\n",
" | strides=None, order=None)\n",
" | \n",
" | An array object represents a multidimensional, homogeneous array\n",
" | of fixed-size items. An associated data-type object describes the\n",
" | format of each element in the array (its byte-order, how many bytes it\
n",
" | occupies in memory, whether it is an integer, a floating point number,\
n",
" | or something else, etc.)\n",
" | \n",
" | Arrays should be constructed using àrray`, `zeros` or èmpty` (refer\
n",
" | to the See Also section below). The parameters given here refer to\n",
" | a low-level method (`ndarray(...)`) for instantiating an array.\n",
" | \n",
" | For more information, refer to the `numpy` module and examine the\n",
" | methods and attributes of an array.\n",
" | \n",
" | Parameters\n",
" | ----------\n",
" | (for the __new__ method; see Notes below)\n",
" | \n",
" | Shape of created array.\n",
" | dtype : data-type, optional\n",
" | Any object that can be interpreted as a numpy data type.\n",
" | buffer : object exposing buffer interface, optional\n",
" | Used to fill the array with data.\n",
" | Offset of array data in buffer.\n",
" | strides : tuple of ints, optional\n",
" | Strides of data in memory.\n",
" | order : {'C', 'F'}, optional\n",
" | Row-major (C-style) or column-major (Fortran-style) order.\n",
" | \n",
" | Attributes\n",
" | ----------\n",
" | T : ndarray\n",
" | Transpose of the array.\n",
" | data : buffer\n",
" | The array's elements, in memory.\n",
" | Describes the format of the elements in the array.\n",
" | flags : dict\n",
" | Dictionary containing information related to memory use, e.g.,\n",
" | 'C_CONTIGUOUS', 'OWNDATA', 'WRITEABLE', etc.\n",
" | flat : numpy.flatiter object\n",
" | Flattened version of the array as an iterator. The iterator\n",
" | allows assignments, e.g., ``x.flat = 3`` (See `ndarray.flat` for\n",
" | assignment examples; TODO).\n",
" | imag : ndarray\n",
" | Imaginary part of the array.\n",
" | real : ndarray\n",
" | Real part of the array.\n",
" | size : int\n",
" | itemsize : int\n",
" | The memory use of each array element in bytes.\n",
" | nbytes : int\n",
" | The total number of bytes required to store the array data,\n",
" | i.e., `ìtemsize * size``.\n",
" | ndim : int\n",
" | The array's number of dimensions.\n",
" | Shape of the array.\n",
" | strides : tuple of ints\n",
" | The step-size required to move from one element to the next in\n",
" | memory. For example, a contiguous ``(3, 4)`` array of type\n",
" | `ìnt16`` in C-order has strides ``(8, 2)``. This implies that\n",
" | to move from element to element in memory requires jumps of 2
bytes.\n",
" | To move from row-to-row, one needs to jump 8 bytes at a time\n",
" | (``2 * 4``).\n",
" | ctypes : ctypes object\n",
" | Class containing properties of the array needed for interaction\n",
" | with ctypes.\n",
" | base : ndarray\n",
" | If the array is a view into another array, that array is its `base`\
n",
" | (unless that array is also a view). The `base` array is where the\
n",
" | array data is actually stored.\n",
" | \n",
" | See Also\n",
" | --------\n",
" | array : Construct an array.\n",
" | zeros : Create an array, each element of which is zero.\n",
" | empty : Create an array, but leave its allocated memory unchanged
(i.e.,\n",
" | it contains \"garbage\").\n",
" | dtype : Create a data-type.\n",
" | numpy.typing.NDArray : An ndarray alias :term:`generic <generic type>`\
n",
" | \n",
" | Notes\n",
" | -----\n",
" | There are two modes of creating an array using ``__new__``:\n",
" | \n",
" | 1. If `buffer` is None, then only `shape`, `dtype`, and òrder`\n",
" | are used.\n",
" | 2. If `buffer` is an object exposing the buffer interface, then\n",
" | all keywords are interpreted.\n",
" | \n",
" | No ``__init__`` method is needed because the array is fully initialized\
n",
" | after the ``__new__`` method.\n",
" | \n",
" | Examples\n",
" | --------\n",
" | These examples illustrate the low-level `ndarray` constructor. Refer\
n",
" | to the `See Also` section above for easier ways of constructing an\n",
" | ndarray.\n",
" | \n",
" | First mode, `buffer` is None:\n",
" | \n",
" | >>> np.ndarray(shape=(2,2), dtype=float, order='F')\n",
" | array([[0.0e+000, 0.0e+000], # random\n",
" | [ nan, 2.5e-323]])\n",
" | \n",
" | Second mode:\n",
" | \n",
" | >>> np.ndarray((2,), buffer=np.array([1,2,3]),\n",
" | ... offset=np.int_().itemsize,\n",
" | ... dtype=int) # offset = 1*itemsize, i.e. skip first
element\n",
" | array([2, 3])\n",
" | \n",
" | Methods defined here:\n",
" | \n",
" | __abs__(self, /)\n",
" | abs(self)\n",
" | \n",
" | __add__(self, value, /)\n",
" | Return self+value.\n",
" | \n",
" | __and__(self, value, /)\n",
" | Return self&value.\n",
" | \n",
" | __array__(...)\n",
" | a.__array__([dtype], /) -> reference if type unchanged, copy
otherwise.\n",
" | \n",
" | Returns either a new reference to self if dtype is not given or a
new array\n",
" | of provided data type if dtype is different from the current dtype
of the\n",
" | array.\n",
" | \n",
" | __array_finalize__(...)\n",
" | a.__array_finalize__(obj, /)\n",
" | \n",
" | Present so subclasses can call super. Does nothing.\n",
" | \n",
" | __array_function__(...)\n",
" | \n",
" | __array_prepare__(...)\n",
" | a.__array_prepare__(array[, context], /)\n",
" | \n",
" | \n",
" | __array_ufunc__(...)\n",
" | \n",
" | __array_wrap__(...)\n",
" | a.__array_wrap__(array[, context], /)\n",
" | \n",
" | \n",
" | __bool__(self, /)\n",
" | True if self else False\n",
" | \n",
" | __complex__(...)\n",
" | \n",
" | __contains__(self, key, /)\n",
" | Return key in self.\n",
" | \n",
" | __copy__(...)\n",
" | a.__copy__()\n",
" | \n",
" | Used if :func:`copy.copy` is called on an array. Returns a copy of
the array.\n",
" | \n",
" | Equivalent to `à.copy(order='K')``.\n",
" | \n",
" | __deepcopy__(...)\n",
" | a.__deepcopy__(memo, /) -> Deep copy of array.\n",
" | \n",
" | Used if :func:`copy.deepcopy` is called on an array.\n",
" | \n",
" | __delitem__(self, key, /)\n",
" | Delete self[key].\n",
" | \n",
" | __divmod__(self, value, /)\n",
" | Return divmod(self, value).\n",
" | \n",
" | __dlpack__(...)\n",
" | a.__dlpack__(*, stream=None)\n",
" | \n",
" | \n",
" | __dlpack_device__(...)\n",
" | a.__dlpack_device__()\n",
" | \n",
" | \n",
" | __eq__(self, value, /)\n",
" | Return self==value.\n",
" | \n",
" | __float__(self, /)\n",
" | float(self)\n",
" | \n",
" | __floordiv__(self, value, /)\n",
" | Return self//value.\n",
" | \n",
" | __format__(...)\n",
" | Default object formatter.\n",
" | \n",
" | __ge__(self, value, /)\n",
" | Return self>=value.\n",
" | \n",
" | __getitem__(self, key, /)\n",
" | Return self[key].\n",
" | \n",
" | __gt__(self, value, /)\n",
" | Return self>value.\n",
" | \n",
" | __iadd__(self, value, /)\n",
" | Return self+=value.\n",
" | \n",
" | __iand__(self, value, /)\n",
" | Return self&=value.\n",
" | \n",
" | __ifloordiv__(self, value, /)\n",
" | Return self//=value.\n",
" | \n",
" | __ilshift__(self, value, /)\n",
" | Return self<<=value.\n",
" | \n",
" | __imatmul__(self, value, /)\n",
" | Return self@=value.\n",
" | \n",
" | __imod__(self, value, /)\n",
" | Return self%=value.\n",
" | \n",
" | __imul__(self, value, /)\n",
" | Return self*=value.\n",
" | \n",
" | __index__(self, /)\n",
" | Return self converted to an integer, if self is suitable for use as
an index into a list.\n",
" | \n",
" | __int__(self, /)\n",
" | int(self)\n",
" | \n",
" | __invert__(self, /)\n",
" | ~self\n",
" | \n",
" | __ior__(self, value, /)\n",
" | Return self|=value.\n",
" | \n",
" | __ipow__(self, value, /)\n",
" | Return self**=value.\n",
" | \n",
" | __irshift__(self, value, /)\n",
" | Return self>>=value.\n",
" | \n",
" | __isub__(self, value, /)\n",
" | Return self-=value.\n",
" | \n",
" | __iter__(self, /)\n",
" | Implement iter(self).\n",
" | \n",
" | __itruediv__(self, value, /)\n",
" | Return self/=value.\n",
" | \n",
" | __ixor__(self, value, /)\n",
" | Return self^=value.\n",
" | \n",
" | __le__(self, value, /)\n",
" | Return self<=value.\n",
" | \n",
" | __len__(self, /)\n",
" | Return len(self).\n",
" | \n",
" | __lshift__(self, value, /)\n",
" | Return self<<value.\n",
" | \n",
" | __lt__(self, value, /)\n",
" | Return self<value.\n",
" | \n",
" | __matmul__(self, value, /)\n",
" | Return self@value.\n",
" | \n",
" | __mod__(self, value, /)\n",
" | Return self%value.\n",
" | \n",
" | __mul__(self, value, /)\n",
" | Return self*value.\n",
" | \n",
" | __ne__(self, value, /)\n",
" | Return self!=value.\n",
" | \n",
" | __neg__(self, /)\n",
" | -self\n",
" | \n",
" | __or__(self, value, /)\n",
" | Return self|value.\n",
" | \n",
" | __pos__(self, /)\n",
" | +self\n",
" | \n",
" | __pow__(self, value, mod=None, /)\n",
" | Return pow(self, value, mod).\n",
" | \n",
" | __radd__(self, value, /)\n",
" | Return value+self.\n",
" | \n",
" | __rand__(self, value, /)\n",
" | Return value&self.\n",
" | \n",
" | __rdivmod__(self, value, /)\n",
" | Return divmod(value, self).\n",
" | \n",
" | __reduce__(...)\n",
" | a.__reduce__()\n",
" | \n",
" | For pickling.\n",
" | \n",
" | __reduce_ex__(...)\n",
" | Helper for pickle.\n",
" | \n",
" | __repr__(self, /)\n",
" | Return repr(self).\n",
" | \n",
" | __rfloordiv__(self, value, /)\n",
" | Return value//self.\n",
" | \n",
" | __rlshift__(self, value, /)\n",
" | Return value<<self.\n",
" | \n",
" | __rmatmul__(self, value, /)\n",
" | Return value@self.\n",
" | \n",
" | __rmod__(self, value, /)\n",
" | Return value%self.\n",
" | \n",
" | __rmul__(self, value, /)\n",
" | Return value*self.\n",
" | \n",
" | __ror__(self, value, /)\n",
" | Return value|self.\n",
" | \n",
" | __rpow__(self, value, mod=None, /)\n",
" | Return pow(value, self, mod).\n",
" | \n",
" | __rrshift__(self, value, /)\n",
" | Return value>>self.\n",
" | \n",
" | __rshift__(self, value, /)\n",
" | Return self>>value.\n",
" | \n",
" | __rsub__(self, value, /)\n",
" | Return value-self.\n",
" | \n",
" | __rtruediv__(self, value, /)\n",
" | Return value/self.\n",
" | \n",
" | __rxor__(self, value, /)\n",
" | Return value^self.\n",
" | \n",
" | __setitem__(self, key, value, /)\n",
" | Set self[key] to value.\n",
" | \n",
" | __setstate__(...)\n",
" | a.__setstate__(state, /)\n",
" | \n",
" | For unpickling.\n",
" | \n",
" | The `state` argument must be a sequence that contains the following\
n",
" | elements:\n",
" | \n",
" | Parameters\n",
" | ----------\n",
" | version : int\n",
" | optional pickle version. If omitted defaults to 0.\n",
" | shape : tuple\n",
" | dtype : data-type\n",
" | isFortran : bool\n",
" | rawdata : string or list\n",
" | a binary string with the data (or a list if 'a' is an object
array)\n",
" | \n",
" | __sizeof__(...)\n",
" | Size of object in memory, in bytes.\n",
" | \n",
" | __str__(self, /)\n",
" | Return str(self).\n",
" | \n",
" | __sub__(self, value, /)\n",
" | Return self-value.\n",
" | \n",
" | __truediv__(self, value, /)\n",
" | Return self/value.\n",
" | \n",
" | __xor__(self, value, /)\n",
" | Return self^value.\n",
" | \n",
" | all(...)\n",
" | a.all(axis=None, out=None, keepdims=False, *, where=True)\n",
" | \n",
" | Returns True if all elements evaluate to True.\n",
" | \n",
" | Refer to `numpy.all` for full documentation.\n",
" | \n",
" | See Also\n",
" | --------\n",
" | numpy.all : equivalent function\n",
" | \n",
" | any(...)\n",
" | a.any(axis=None, out=None, keepdims=False, *, where=True)\n",
" | \n",
" | Returns True if any of the elements of à` evaluate to True.\n",
" | \n",
" | Refer to `numpy.any` for full documentation.\n",
" | \n",
" | See Also\n",
" | --------\n",
" | numpy.any : equivalent function\n",
" | \n",
" | argmax(...)\n",
" | a.argmax(axis=None, out=None, *, keepdims=False)\n",
" | \n",
" | Return indices of the maximum values along the given axis.\n",
" | \n",
" | Refer to `numpy.argmax` for full documentation.\n",
" | \n",
" | See Also\n",
" | --------\n",
" | numpy.argmax : equivalent function\n",
" | \n",
" | argmin(...)\n",
" | a.argmin(axis=None, out=None, *, keepdims=False)\n",
" | \n",
" | Return indices of the minimum values along the given axis.\n",
" | \n",
" | Refer to `numpy.argmin` for detailed documentation.\n",
" | \n",
" | See Also\n",
" | --------\n",
" | numpy.argmin : equivalent function\n",
" | \n",
" | argpartition(...)\n",
" | a.argpartition(kth, axis=-1, kind='introselect', order=None)\n",
" | \n",
" | Returns the indices that would partition this array.\n",
" | \n",
" | Refer to `numpy.argpartition` for full documentation.\n",
" | \n",
" | \n",
" | See Also\n",
" | --------\n",
" | numpy.argpartition : equivalent function\n",
" | \n",
" | argsort(...)\n",
" | a.argsort(axis=-1, kind=None, order=None)\n",
" | \n",
" | Returns the indices that would sort this array.\n",
" | \n",
" | Refer to `numpy.argsort` for full documentation.\n",
" | \n",
" | See Also\n",
" | --------\n",
" | numpy.argsort : equivalent function\n",
" | \n",
" | astype(...)\n",
" | a.astype(dtype, order='K', casting='unsafe', subok=True, copy=True)\
n",
" | \n",
" | Copy of the array, cast to a specified type.\n",
" | \n",
" | Parameters\n",
" | ----------\n",
" | Typecode or data-type to which the array is cast.\n",
" | Controls the memory layout order of the result.\n",
" | 'C' means C order, 'F' means Fortran order, 'A'\n",
" | means 'F' order if all the arrays are Fortran contiguous,\n",
" | 'C' order otherwise, and 'K' means as close to the\n",
" | order the array elements appear in memory as possible.\n",
" | Default is 'K'.\n",
" | casting : {'no', 'equiv', 'safe', 'same_kind', 'unsafe'}, optional\
n",
" | Controls what kind of data casting may occur. Defaults to
'unsafe'\n",
" | for backwards compatibility.\n",
" | \n",
" | * 'no' means the data types should not be cast at all.\n",
" | * 'equiv' means only byte-order changes are allowed.\n",
" | * 'safe' means only casts which can preserve values are
allowed.\n",
" | * 'same_kind' means only safe casts or casts within a kind,\
n",
" | like float64 to float32, are allowed.\n",
" | * 'unsafe' means any data conversions may be done.\n",
" | subok : bool, optional\n",
" | If True, then sub-classes will be passed-through (default),
otherwise\n",
" | the returned array will be forced to be a base-class array.\n",
" | copy : bool, optional\n",
" | By default, astype always returns a newly allocated array. If
this\n",
" | is set to false, and the `dtype`, òrder`, and `subok`\n",
" | requirements are satisfied, the input array is returned instead\
n",
" | of a copy.\n",
" | \n",
" | Returns\n",
" | -------\n",
" | arr_t : ndarray\n",
" | Unless `copy` is False and the other conditions for returning
the input\n",
" | array are satisfied (see description for `copy` input
parameter), àrr_t`\n",
" | is a new array of the same shape as the input array, with dtype,
order\n",
" | given by `dtype`, òrder`.\n",
" | \n",
" | Notes\n",
" | -----\n",
" | Casting between a simple data type and a structured one is
possible only\n",
" | for \"unsafe\" casting. Casting to multiple fields is allowed,
but\n",
" | casting from multiple fields is not.\n",
" | \n",
" | Casting from numeric to string types in 'safe' casting mode
requires\n",
" | that the string dtype length is long enough to store the max\n",
" | integer/float value converted.\n",
" | \n",
" | Raises\n",
" | ------\n",
" | ComplexWarning\n",
" | When casting from complex to float or int. To avoid this,\n",
" | one should use `à.real.astype(t)``.\n",
" | \n",
" | Examples\n",
" | --------\n",
" | >>> x = np.array([1, 2, 2.5])\n",
" | >>> x\n",
" | array([1. , 2. , 2.5])\n",
" | \n",
" | >>> x.astype(int)\n",
" | array([1, 2, 2])\n",
" | \n",
" | byteswap(...)\n",
" | a.byteswap(inplace=False)\n",
" | \n",
" | Swap the bytes of the array elements\n",
" | \n",
" | Toggle between low-endian and big-endian data representation by\n",
" | returning a byteswapped array, optionally swapped in-place.\n",
" | Arrays of byte-strings are not swapped. The real and imaginary\n",
" | parts of a complex number are swapped individually.\n",
" | \n",
" | Parameters\n",
" | ----------\n",
" | inplace : bool, optional\n",
" | If ``True``, swap bytes in-place, default is ``False``.\n",
" | \n",
" | Returns\n",
" | -------\n",
" | out : ndarray\n",
" | The byteswapped array. If ìnplace` is ``True``, this is\n",
" | a view to self.\n",
" | \n",
" | Examples\n",
" | --------\n",
" | >>> A = np.array([1, 256, 8755], dtype=np.int16)\n",
" | ['0x1', '0x100', '0x2233']\n",
" | >>> A.byteswap(inplace=True)\n",
" | array([ 256, 1, 13090], dtype=int16)\n",
" | ['0x100', '0x1', '0x3322']\n",
" | \n",
" | Arrays of byte-strings are not swapped\n",
" | \n",
" | >>> A = np.array([b'ceg', b'fac'])\n",
" | >>> A.byteswap()\n",
" | array([b'ceg', b'fac'], dtype='|S3')\n",
" | \n",
" | `À.newbyteorder().byteswap()`` produces an array with the same
values\n",
" | but different representation in memory\n",
" | \n",
" | >>> A = np.array([1, 2, 3])\n",
" | array([1, 0, 0, 0, 0, 0, 0, 0, 2, 0, 0, 0, 0, 0, 0, 0, 3, 0, 0, 0,
0, 0,\n",
" | 0, 0], dtype=uint8)\n",
" | >>> A.newbyteorder().byteswap(inplace=True)\n",
" | array([1, 2, 3])\n",
" | array([0, 0, 0, 0, 0, 0, 0, 1, 0, 0, 0, 0, 0, 0, 0, 2, 0, 0, 0, 0,
0, 0,\n",
" | 0, 3], dtype=uint8)\n",
" | \n",
" | choose(...)\n",
" | a.choose(choices, out=None, mode='raise')\n",
" | \n",
" | Use an index array to construct a new array from a set of choices.\
n",
" | \n",
" | Refer to `numpy.choose` for full documentation.\n",
" | \n",
" | See Also\n",
" | --------\n",
" | numpy.choose : equivalent function\n",
" | \n",
" | clip(...)\n",
" | a.clip(min=None, max=None, out=None, **kwargs)\n",
" | \n",
" | Return an array whose values are limited to ``[min, max]``.\n",
" | One of max or min must be given.\n",
" | \n",
" | Refer to `numpy.clip` for full documentation.\n",
" | \n",
" | See Also\n",
" | --------\n",
" | numpy.clip : equivalent function\n",
" | \n",
" | compress(...)\n",
" | a.compress(condition, axis=None, out=None)\n",
" | \n",
" | Return selected slices of this array along given axis.\n",
" | \n",
" | Refer to `numpy.compress` for full documentation.\n",
" | \n",
" | See Also\n",
" | --------\n",
" | numpy.compress : equivalent function\n",
" | \n",
" | conj(...)\n",
" | a.conj()\n",
" | \n",
" | Complex-conjugate all elements.\n",
" | \n",
" | \n",
" | See Also\n",
" | --------\n",
" | \n",
" | conjugate(...)\n",
" | a.conjugate()\n",
" | \n",
" | Return the complex conjugate, element-wise.\n",
" | \n",
" | \n",
" | See Also\n",
" | --------\n",
" | \n",
" | copy(...)\n",
" | a.copy(order='C')\n",
" | \n",
" | Return a copy of the array.\n",
" | \n",
" | Parameters\n",
" | ----------\n",
" | Controls the memory layout of the copy. 'C' means C-order,\n",
" | 'F' means F-order, 'A' means 'F' if à` is Fortran contiguous,\
n",
" | 'C' otherwise. 'K' means match the layout of à` as closely\n",
" | as possible. (Note that this function and :func:`numpy.copy` are
very\n",
" | similar but have different default values for their order=\n",
" | arguments, and this function always passes sub-classes
through.)\n",
" | \n",
" | See also\n",
" | --------\n",
" | numpy.copy : Similar function with different default behavior\n",
" | numpy.copyto\n",
" | \n",
" | Notes\n",
" | -----\n",
" | This function is the preferred method for creating an array copy.
The\n",
" | function :func:`numpy.copy` is similar, but it defaults to using
order 'K',\n",
" | and will not pass sub-classes through by default.\n",
" | \n",
" | Examples\n",
" | --------\n",
" | >>> x = np.array([[1,2,3],[4,5,6]], order='F')\n",
" | \n",
" | >>> y = x.copy()\n",
" | \n",
" | >>> x.fill(0)\n",
" | \n",
" | >>> x\n",
" | array([[0, 0, 0],\n",
" | [0, 0, 0]])\n",
" | \n",
" | >>> y\n",
" | array([[1, 2, 3],\n",
" | [4, 5, 6]])\n",
" | \n",
" | >>> y.flags['C_CONTIGUOUS']\n",
" | True\n",
" | \n",
" | cumprod(...)\n",
" | a.cumprod(axis=None, dtype=None, out=None)\n",
" | \n",
" | Return the cumulative product of the elements along the given axis.\
n",
" | \n",
" | Refer to `numpy.cumprod` for full documentation.\n",
" | \n",
" | See Also\n",
" | --------\n",
" | numpy.cumprod : equivalent function\n",
" | \n",
" | cumsum(...)\n",
" | a.cumsum(axis=None, dtype=None, out=None)\n",
" | \n",
" | Return the cumulative sum of the elements along the given axis.\n",
" | \n",
" | Refer to `numpy.cumsum` for full documentation.\n",
" | \n",
" | See Also\n",
" | --------\n",
" | numpy.cumsum : equivalent function\n",
" | \n",
" | diagonal(...)\n",
" | a.diagonal(offset=0, axis1=0, axis2=1)\n",
" | \n",
" | Return specified diagonals. In NumPy 1.9 the returned array is a\n",
" | read-only view instead of a copy as in previous NumPy versions. In\
n",
" | a future version the read-only restriction will be removed.\n",
" | \n",
" | Refer to :func:`numpy.diagonal` for full documentation.\n",
" | \n",
" | See Also\n",
" | --------\n",
" | numpy.diagonal : equivalent function\n",
" | \n",
" | dot(...)\n",
" | \n",
" | dump(...)\n",
" | a.dump(file)\n",
" | \n",
" | Dump a pickle of the array to the specified file.\n",
" | The array can be read back with pickle.load or numpy.load.\n",
" | \n",
" | Parameters\n",
" | ----------\n",
" | file : str or Path\n",
" | A string naming the dump file.\n",
" | \n",
" | \n",
" | dumps(...)\n",
" | a.dumps()\n",
" | \n",
" | Returns the pickle of the array as a string.\n",
" | pickle.loads will convert the string back to an array.\n",
" | \n",
" | Parameters\n",
" | ----------\n",
" | None\n",
" | \n",
" | fill(...)\n",
" | a.fill(value)\n",
" | \n",
" | Fill the array with a scalar value.\n",
" | \n",
" | Parameters\n",
" | ----------\n",
" | value : scalar\n",
" | All elements of à` will be assigned this value.\n",
" | \n",
" | Examples\n",
" | --------\n",
" | >>> a = np.array([1, 2])\n",
" | >>> a.fill(0)\n",
" | >>> a\n",
" | array([0, 0])\n",
" | >>> a = np.empty(2)\n",
" | >>> a.fill(1)\n",
" | >>> a\n",
" | array([1., 1.])\n",
" | \n",
" | Fill expects a scalar value and always behaves the same as
assigning\n",
" | to a single array element. The following is a rare example where
this\n",
" | distinction is important:\n",
" | \n",
" | >>> a = np.array([None, None], dtype=object)\n",
" | >>> a[0] = np.array(3)\n",
" | >>> a\n",
" | array([array(3), None], dtype=object)\n",
" | >>> a.fill(np.array(3))\n",
" | >>> a\n",
" | array([array(3), array(3)], dtype=object)\n",
" | \n",
" | Where other forms of assignments will unpack the array being
assigned:\n",
" | \n",
" | >>> a[...] = np.array(3)\n",
" | >>> a\n",
" | array([3, 3], dtype=object)\n",
" | \n",
" | flatten(...)\n",
" | a.flatten(order='C')\n",
" | \n",
" | Return a copy of the array collapsed into one dimension.\n",
" | \n",
" | Parameters\n",
" | ----------\n",
" | 'C' means to flatten in row-major (C-style) order.\n",
" | 'F' means to flatten in column-major (Fortran-\n",
" | style) order. 'A' means to flatten in column-major\n",
" | order if à` is Fortran *contiguous* in memory,\n",
" | row-major order otherwise. 'K' means to flatten\n",
" | à` in the order the elements occur in memory.\n",
" | The default is 'C'.\n",
" | \n",
" | Returns\n",
" | -------\n",
" | y : ndarray\n",
" | A copy of the input array, flattened to one dimension.\n",
" | \n",
" | See Also\n",
" | --------\n",
" | ravel : Return a flattened array.\n",
" | flat : A 1-D flat iterator over the array.\n",
" | \n",
" | Examples\n",
" | --------\n",
" | >>> a = np.array([[1,2], [3,4]])\n",
" | >>> a.flatten()\n",
" | array([1, 2, 3, 4])\n",
" | >>> a.flatten('F')\n",
" | array([1, 3, 2, 4])\n",
" | \n",
" | getfield(...)\n",
" | a.getfield(dtype, offset=0)\n",
" | \n",
" | Returns a field of the given array as a certain type.\n",
" | \n",
" | A field is a view of the array data with a given data-type. The
values in\n",
" | the view are determined by the given type and the offset into the
current\n",
" | array in bytes. The offset needs to be such that the view dtype fits
in the\n",
" | array dtype; for example an array of dtype complex128 has 16-byte
elements.\n",
" | If taking a view with a 32-bit integer (4 bytes), the offset needs
to be\n",
" | between 0 and 12 bytes.\n",
" | \n",
" | Parameters\n",
" | ----------\n",
" | The data type of the view. The dtype size of the view can not be
larger\n",
" | than that of the array itself.\n",
" | offset : int\n",
" | Number of bytes to skip before beginning the element view.\n",
" | \n",
" | Examples\n",
" | --------\n",
" | >>> x = np.diag([1.+1.j]*2)\n",
" | >>> x[1, 1] = 2 + 4.j\n",
" | >>> x\n",
" | array([[1.+1.j, 0.+0.j],\n",
" | [0.+0.j, 2.+4.j]])\n",
" | array([[1., 0.],\n",
" | [0., 2.]])\n",
" | \n",
" | By choosing an offset of 8 bytes we can select the complex part of
the\n",
" | array for our view:\n",
" | \n",
" | >>> x.getfield(np.float64, offset=8)\n",
" | array([[1., 0.],\n",
" | [0., 4.]])\n",
" | \n",
" | item(...)\n",
" | a.item(*args)\n",
" | \n",
" | Copy an element of an array to a standard Python scalar and return
it.\n",
" | \n",
" | Parameters\n",
" | ----------\n",
" | \\*args : Arguments (variable number and type)\n",
" | \n",
" | * none: in this case, the method only works for arrays\n",
" | with one element (à.size == 1`), which element is\n",
" | copied into a standard Python scalar object and returned.\n",
" | \n",
" | * int_type: this argument is interpreted as a flat index into\
n",
" | the array, specifying which element to copy and return.\n",
" | \n",
" | * tuple of int_types: functions as does a single int_type
argument,\n",
" | except that the argument is interpreted as an nd-index into
the\n",
" | array.\n",
" | \n",
" | Returns\n",
" | -------\n",
" | z : Standard Python scalar object\n",
" | A copy of the specified element of the array as a suitable\n",
" | Python scalar\n",
" | \n",
" | Notes\n",
" | -----\n",
" | When the data type of à` is longdouble or clongdouble, item()
returns\n",
" | a scalar array object because there is no available Python scalar
that\n",
" | would not lose information. Void arrays return a buffer object for
item(),\n",
" | unless fields are defined, in which case a tuple is returned.\n",
" | \n",
" | ìtem` is very similar to a[args], except, instead of an array
scalar,\n",
" | a standard Python scalar is returned. This can be useful for
speeding up\n",
" | access to elements of the array and doing arithmetic on elements of
the\n",
" | array using Python's optimized math.\n",
" | \n",
" | Examples\n",
" | --------\n",
" | >>> x\n",
" | array([[2, 2, 6],\n",
" | [1, 3, 6],\n",
" | [1, 0, 1]])\n",
" | >>> x.item(3)\n",
" | 1\n",
" | >>> x.item(7)\n",
" | 0\n",
" | >>> x.item((0, 1))\n",
" | 2\n",
" | >>> x.item((2, 2))\n",
" | 1\n",
" | \n",
" | itemset(...)\n",
" | a.itemset(*args)\n",
" | \n",
" | Insert scalar into an array (scalar is cast to array's dtype, if
possible)\n",
" | \n",
" | There must be at least 1 argument, and define the last argument\n",
" | as *item*. Then, `à.itemset(*args)`` is equivalent to but faster\
n",
" | than `à[args] = item``. The item should be a scalar value and
àrgs`\n",
" | must select a single item in the array à`.\n",
" | \n",
" | Parameters\n",
" | ----------\n",
" | \\*args : Arguments\n",
" | If one argument: a scalar, only used in case à` is of size 1.\
n",
" | If two arguments: the last argument is the value to be set\n",
" | and must be a scalar, the first argument specifies a single
array\n",
" | element location. It is either an int or a tuple.\n",
" | \n",
" | Notes\n",
" | -----\n",
" | Compared to indexing syntax, ìtemset` provides some speed increase\
n",
" | for placing a scalar into a particular location in an `ndarray`,\n",
" | if you must do this. However, generally this is discouraged:\n",
" | among other problems, it complicates the appearance of the code.\n",
" | Also, when using ìtemset` (and ìtem`) inside a loop, be sure\n",
" | to assign the methods to a local variable to avoid the attribute\n",
" | look-up at each loop iteration.\n",
" | \n",
" | Examples\n",
" | --------\n",
" | >>> x\n",
" | array([[2, 2, 6],\n",
" | [1, 3, 6],\n",
" | [1, 0, 1]])\n",
" | >>> x.itemset(4, 0)\n",
" | >>> x.itemset((2, 2), 9)\n",
" | >>> x\n",
" | array([[2, 2, 6],\n",
" | [1, 0, 6],\n",
" | [1, 0, 9]])\n",
" | \n",
" | max(...)\n",
" | a.max(axis=None, out=None, keepdims=False, initial=<no value>,
where=True)\n",
" | \n",
" | Return the maximum along a given axis.\n",
" | \n",
" | Refer to `numpy.amax` for full documentation.\n",
" | \n",
" | See Also\n",
" | --------\n",
" | numpy.amax : equivalent function\n",
" | \n",
" | mean(...)\n",
" | a.mean(axis=None, dtype=None, out=None, keepdims=False, *,
where=True)\n",
" | \n",
" | Returns the average of the array elements along given axis.\n",
" | \n",
" | Refer to `numpy.mean` for full documentation.\n",
" | \n",
" | See Also\n",
" | --------\n",
" | numpy.mean : equivalent function\n",
" | \n",
" | min(...)\n",
" | a.min(axis=None, out=None, keepdims=False, initial=<no value>,
where=True)\n",
" | \n",
" | Return the minimum along a given axis.\n",
" | \n",
" | Refer to `numpy.amin` for full documentation.\n",
" | \n",
" | See Also\n",
" | --------\n",
" | numpy.amin : equivalent function\n",
" | \n",
" | newbyteorder(...)\n",
" | arr.newbyteorder(new_order='S', /)\n",
" | \n",
" | Return the array with the same data viewed with a different byte
order.\n",
" | \n",
" | Equivalent to::\n",
" | \n",
" | arr.view(arr.dtype.newbytorder(new_order))\n",
" | \n",
" | Changes are also made in all fields and sub-arrays of the array
data\n",
" | type.\n",
" | \n",
" | \n",
" | \n",
" | Parameters\n",
" | ----------\n",
" | new_order : string, optional\n",
" | Byte order to force; a value from the byte order specifications\
n",
" | below. `new_order` codes can be any of:\n",
" | \n",
" | * 'S' - swap dtype from current to opposite endian\n",
" | * {'<', 'little'} - little endian\n",
" | * {'>', 'big'} - big endian\n",
" | * {'=', 'native'} - native order, equivalent to `sys.byteorder`\
n",
" | * {'|', 'I'} - ignore (no change to byte order)\n",
" | \n",
" | The default value ('S') results in swapping the current\n",
" | byte order.\n",
" | \n",
" | \n",
" | Returns\n",
" | -------\n",
" | new_arr : array\n",
" | New array object with the dtype reflecting given change to the\
n",
" | byte order.\n",
" | \n",
" | nonzero(...)\n",
" | a.nonzero()\n",
" | \n",
" | Return the indices of the elements that are non-zero.\n",
" | \n",
" | Refer to `numpy.nonzero` for full documentation.\n",
" | \n",
" | See Also\n",
" | --------\n",
" | numpy.nonzero : equivalent function\n",
" | \n",
" | partition(...)\n",
" | a.partition(kth, axis=-1, kind='introselect', order=None)\n",
" | \n",
" | Rearranges the elements in the array in such a way that the value of
the\n",
" | element in kth position is in the position it would be in a sorted
array.\n",
" | All elements smaller than the kth element are moved before this
element and\n",
" | all equal or greater are moved behind it. The ordering of the
elements in\n",
" | the two partitions is undefined.\n",
" | \n",
" | \n",
" | Parameters\n",
" | ----------\n",
" | kth : int or sequence of ints\n",
" | Element index to partition by. The kth element value will be in
its\n",
" | final sorted position and all smaller elements will be moved
before it\n",
" | and all equal or greater elements behind it.\n",
" | The order of all elements in the partitions is undefined.\n",
" | If provided with a sequence of kth it will partition all
elements\n",
" | indexed by kth of them into their sorted position at once.\n",
" | \n",
" | .. deprecated:: 1.22.0\n",
" | Passing booleans as index is deprecated.\n",
the\n",
" | last axis.\n",
" | kind : {'introselect'}, optional\n",
" | Selection algorithm. Default is 'introselect'.\n",
specifies\n",
n",
" | be specified as a string, and not all fields need to be
specified,\n",
which\n",
" | \n",
" | See Also\n",
" | --------\n",
" | numpy.partition : Return a partitioned copy of an array.\n",
" | argpartition : Indirect partition.\n",
" | sort : Full sort.\n",
" | \n",
" | Notes\n",
" | -----\n",
" | See ``np.partition`` for notes on the different algorithms.\n",
" | \n",
" | Examples\n",
" | --------\n",
" | >>> a = np.array([3, 4, 2, 1])\n",
" | >>> a.partition(3)\n",
" | >>> a\n",
" | array([2, 1, 3, 4])\n",
" | \n",
" | >>> a.partition((1, 3))\n",
" | >>> a\n",
" | array([1, 2, 3, 4])\n",
" | \n",
" | prod(...)\n",
" | a.prod(axis=None, dtype=None, out=None, keepdims=False, initial=1,
where=True)\n",
" | \n",
" | Return the product of the array elements over the given axis\n",
" | \n",
" | Refer to `numpy.prod` for full documentation.\n",
" | \n",
" | See Also\n",
" | --------\n",
" | numpy.prod : equivalent function\n",
" | \n",
" | ptp(...)\n",
" | a.ptp(axis=None, out=None, keepdims=False)\n",
" | \n",
" | Peak to peak (maximum - minimum) value along a given axis.\n",
" | \n",
" | Refer to `numpy.ptp` for full documentation.\n",
" | \n",
" | See Also\n",
" | --------\n",
" | numpy.ptp : equivalent function\n",
" | \n",
" | put(...)\n",
" | a.put(indices, values, mode='raise')\n",
" | \n",
" | Set `à.flat[n] = values[n]`` for all `n` in indices.\n",
" | \n",
" | Refer to `numpy.put` for full documentation.\n",
" | \n",
" | See Also\n",
" | --------\n",
" | numpy.put : equivalent function\n",
" | \n",
" | ravel(...)\n",
" | a.ravel([order])\n",
" | \n",
" | Return a flattened array.\n",
" | \n",
" | Refer to `numpy.ravel` for full documentation.\n",
" | \n",
" | See Also\n",
" | --------\n",
" | numpy.ravel : equivalent function\n",
" | \n",
" | ndarray.flat : a flat iterator on the array.\n",
" | \n",
" | repeat(...)\n",
" | a.repeat(repeats, axis=None)\n",
" | \n",
" | Repeat elements of an array.\n",
" | \n",
" | Refer to `numpy.repeat` for full documentation.\n",
" | \n",
" | See Also\n",
" | --------\n",
" | numpy.repeat : equivalent function\n",
" | \n",
" | reshape(...)\n",
" | a.reshape(shape, order='C')\n",
" | \n",
" | Returns an array containing the same data with a new shape.\n",
" | \n",
" | Refer to `numpy.reshape` for full documentation.\n",
" | \n",
" | See Also\n",
" | --------\n",
" | numpy.reshape : equivalent function\n",
" | \n",
" | Notes\n",
" | -----\n",
" | Unlike the free function `numpy.reshape`, this method on `ndarray`
allows\n",
" | the elements of the shape parameter to be passed in as separate
arguments.\n",
" | For example, `à.reshape(10, 11)`` is equivalent to\n",
" | `à.reshape((10, 11))``.\n",
" | \n",
" | resize(...)\n",
" | a.resize(new_shape, refcheck=True)\n",
" | \n",
" | Change shape and size of array in-place.\n",
" | \n",
" | Parameters\n",
" | ----------\n",
" | new_shape : tuple of ints, or `n` ints\n",
" | Shape of resized array.\n",
" | refcheck : bool, optional\n",
" | If False, reference count will not be checked. Default is True.\
n",
" | \n",
" | Returns\n",
" | -------\n",
" | None\n",
" | \n",
" | Raises\n",
" | ------\n",
" | ValueError\n",
" | If à` does not own its own data or references or views to it
exist,\n",
" | and the data memory must be changed.\n",
" | PyPy only: will always raise if the data memory must be changed,
since\n",
" | there is no reliable way to determine if references or views to
it\n",
" | exist.\n",
" | \n",
" | SystemError\n",
" | If the òrder` keyword argument is specified. This behaviour is
a\n",
" | bug in NumPy.\n",
" | \n",
" | See Also\n",
" | --------\n",
" | resize : Return a new array with the specified shape.\n",
" | \n",
" | Notes\n",
" | -----\n",
" | This reallocates space for the data area if necessary.\n",
" | \n",
" | Only contiguous arrays (data elements consecutive in memory) can be\
n",
" | resized.\n",
" | \n",
" | The purpose of the reference count check is to make sure you\n",
" | do not use this array as a buffer for another Python object and
then\n",
" | reallocate the memory. However, reference counts can increase in\n",
" | other ways so if you are sure that you have not shared the memory\
n",
" | for this array with another Python object, then you may safely set\
n",
" | `refcheck` to False.\n",
" | \n",
" | Examples\n",
" | --------\n",
" | Shrinking an array: array is flattened (in the order that the data
are\n",
" | stored in memory), resized, and reshaped:\n",
" | \n",
" | >>> a = np.array([[0, 1], [2, 3]], order='C')\n",
" | >>> a.resize((2, 1))\n",
" | >>> a\n",
" | array([[0],\n",
" | [1]])\n",
" | \n",
" | >>> a = np.array([[0, 1], [2, 3]], order='F')\n",
" | >>> a.resize((2, 1))\n",
" | >>> a\n",
" | array([[0],\n",
" | [2]])\n",
" | \n",
" | Enlarging an array: as above, but missing entries are filled with
zeros:\n",
" | \n",
" | >>> b = np.array([[0, 1], [2, 3]])\n",
" | >>> b.resize(2, 3) # new_shape parameter doesn't have to be a tuple\
n",
" | >>> b\n",
" | array([[0, 1, 2],\n",
" | [3, 0, 0]])\n",
" | \n",
" | Referencing an array prevents resizing...\n",
" | \n",
" | >>> c = a\n",
" | >>> a.resize((1, 1))\n",
" | ...\n",
" | ValueError: cannot resize an array that references or is
referenced ...\n",
" | \n",
" | Unless `refcheck` is False:\n",
" | \n",
" | >>> a.resize((1, 1), refcheck=False)\n",
" | >>> a\n",
" | array([[0]])\n",
" | >>> c\n",
" | array([[0]])\n",
" | \n",
" | round(...)\n",
" | a.round(decimals=0, out=None)\n",
" | \n",
" | Return à` with each element rounded to the given number of
decimals.\n",
" | \n",
" | Refer to `numpy.around` for full documentation.\n",
" | \n",
" | See Also\n",
" | --------\n",
" | numpy.around : equivalent function\n",
" | \n",
" | searchsorted(...)\n",
" | a.searchsorted(v, side='left', sorter=None)\n",
" | \n",
" | Find indices where elements of v should be inserted in a to maintain
order.\n",
" | \n",
" | For full documentation, see `numpy.searchsorted`\n",
" | \n",
" | See Also\n",
" | --------\n",
" | numpy.searchsorted : equivalent function\n",
" | \n",
" | setfield(...)\n",
" | a.setfield(val, dtype, offset=0)\n",
" | \n",
" | Put a value into a specified place in a field defined by a data-
type.\n",
" | \n",
" | Place `val` into à`'s field defined by `dtype` and beginning
òffset`\n",
" | bytes into the field.\n",
" | \n",
" | Parameters\n",
" | ----------\n",
" | val : object\n",
" | Value to be placed in field.\n",
" | Data-type of the field in which to place `val`.\n",
" | The number of bytes into the field at which to place `val`.\n",
" | \n",
" | Returns\n",
" | -------\n",
" | None\n",
" | \n",
" | See Also\n",
" | --------\n",
" | getfield\n",
" | \n",
" | Examples\n",
" | --------\n",
" | >>> x = np.eye(3)\n",
" | array([[1., 0., 0.],\n",
" | [0., 1., 0.],\n",
" | [0., 0., 1.]])\n",
" | >>> x.setfield(3, np.int32)\n",
" | >>> x.getfield(np.int32)\n",
" | array([[3, 3, 3],\n",
" | [3, 3, 3],\n",
" | [3, 3, 3]], dtype=int32)\n",
" | >>> x\n",
" | array([[1.0e+000, 1.5e-323, 1.5e-323],\n",
" | [1.5e-323, 1.0e+000, 1.5e-323],\n",
" | [1.5e-323, 1.5e-323, 1.0e+000]])\n",
" | >>> x.setfield(np.eye(3), np.int32)\n",
" | >>> x\n",
" | array([[1., 0., 0.],\n",
" | [0., 1., 0.],\n",
" | [0., 0., 1.]])\n",
" | \n",
" | setflags(...)\n",
" | a.setflags(write=None, align=None, uic=None)\n",
" | \n",
" | Set array flags WRITEABLE, ALIGNED, WRITEBACKIFCOPY,\n",
" | respectively.\n",
" | \n",
" | These Boolean-valued flags affect how numpy interprets the memory\
n",
" | area used by à` (see Notes below). The ALIGNED flag can only\n",
" | be set to True if the data is actually aligned according to the
type.\n",
" | The WRITEBACKIFCOPY and flag can never be set\n",
" | to True. The flag WRITEABLE can only be set to True if the array
owns its\n",
" | own memory, or the ultimate owner of the memory exposes a writeable
buffer\n",
" | interface, or is a string. (The exception for string is made so
that\n",
" | unpickling can be done without copying memory.)\n",
" | \n",
" | Parameters\n",
" | ----------\n",
" | write : bool, optional\n",
" | Describes whether or not à` can be written to.\n",
" | align : bool, optional\n",
" | Describes whether or not à` is aligned properly for its type.\
n",
" | uic : bool, optional\n",
" | Describes whether or not à` is a copy of another \"base\"
array.\n",
" | \n",
" | Notes\n",
" | -----\n",
" | Array flags provide information about how the memory area used\n",
" | for the array is to be interpreted. There are 7 Boolean flags\n",
" | in use, only four of which can be changed by the user:\n",
" | WRITEBACKIFCOPY, WRITEABLE, and ALIGNED.\n",
" | \n",
" | WRITEABLE (W) the data area can be written to;\n",
" | \n",
" | ALIGNED (A) the data and strides are aligned appropriately for the
hardware\n",
" | (as determined by the compiler);\n",
" | \n",
" | WRITEBACKIFCOPY (X) this array is a copy of some other array
(referenced\n",
" | by .base). When the C-API function PyArray_ResolveWritebackIfCopy
is\n",
" | called, the base array will be updated with the contents of this
array.\n",
" | \n",
" | All flags can be accessed using the single (upper case) letter as
well\n",
" | as the full name.\n",
" | \n",
" | Examples\n",
" | --------\n",
" | >>> y = np.array([[3, 1, 7],\n",
" | ... [2, 0, 0],\n",
" | ... [8, 5, 9]])\n",
" | >>> y\n",
" | array([[3, 1, 7],\n",
" | [2, 0, 0],\n",
" | [8, 5, 9]])\n",
" | >>> y.flags\n",
" | WRITEABLE : True\n",
" | ALIGNED : True\n",
" | >>> y.setflags(write=0, align=0)\n",
" | >>> y.flags\n",
" | WRITEABLE : False\n",
" | ALIGNED : False\n",
" | >>> y.setflags(uic=1)\n",
" | ValueError: cannot set WRITEBACKIFCOPY flag to True\n",
" | \n",
" | sort(...)\n",
" | a.sort(axis=-1, kind=None, order=None)\n",
" | \n",
" | Sort an array in-place. Refer to `numpy.sort` for full
documentation.\n",
" | \n",
" | Parameters\n",
" | ----------\n",
the\n",
" | last axis.\n",
" | kind : {'quicksort', 'mergesort', 'heapsort', 'stable'}, optional\
n",
" | Sorting algorithm. The default is 'quicksort'. Note that both
'stable'\n",
" | and 'mergesort' use timsort under the covers and, in general,
the\n",
" | actual implementation will vary with datatype. The 'mergesort'
option\n",
" | is retained for backwards compatibility.\n",
" | \n",
" | The 'stable' option was added.\n",
" | \n",
specifies\n",
n",
" | be specified as a string, and not all fields need be specified,\
n",
which\n",
" | \n",
" | See Also\n",
" | --------\n",
" | numpy.sort : Return a sorted copy of an array.\n",
" | numpy.argsort : Indirect sort.\n",
" | numpy.lexsort : Indirect stable sort on multiple keys.\n",
" | numpy.searchsorted : Find elements in sorted array.\n",
" | numpy.partition: Partial sort.\n",
" | \n",
" | Notes\n",
" | -----\n",
" | See `numpy.sort` for notes on the different sorting algorithms.\n",
" | \n",
" | Examples\n",
" | --------\n",
" | >>> a = np.array([[1,4], [3,1]])\n",
" | >>> a.sort(axis=1)\n",
" | >>> a\n",
" | array([[1, 4],\n",
" | [1, 3]])\n",
" | >>> a.sort(axis=0)\n",
" | >>> a\n",
" | array([[1, 3],\n",
" | [1, 4]])\n",
" | \n",
" | Use the òrder` keyword to specify a field to use when sorting a\n",
" | structured array:\n",
" | \n",
" | >>> a = np.array([('a', 2), ('c', 1)], dtype=[('x', 'S1'), ('y',
int)])\n",
" | >>> a.sort(order='y')\n",
" | >>> a\n",
" | array([(b'c', 1), (b'a', 2)],\n",
" | dtype=[('x', 'S1'), ('y', '<i8')])\n",
" | \n",
" | squeeze(...)\n",
" | a.squeeze(axis=None)\n",
" | \n",
" | Remove axes of length one from à`.\n",
" | \n",
" | Refer to `numpy.squeeze` for full documentation.\n",
" | \n",
" | See Also\n",
" | --------\n",
" | numpy.squeeze : equivalent function\n",
" | \n",
" | std(...)\n",
" | a.std(axis=None, dtype=None, out=None, ddof=0, keepdims=False, *,
where=True)\n",
" | \n",
" | Returns the standard deviation of the array elements along given
axis.\n",
" | \n",
" | Refer to `numpy.std` for full documentation.\n",
" | \n",
" | See Also\n",
" | --------\n",
" | numpy.std : equivalent function\n",
" | \n",
" | sum(...)\n",
" | a.sum(axis=None, dtype=None, out=None, keepdims=False, initial=0,
where=True)\n",
" | \n",
" | Return the sum of the array elements over the given axis.\n",
" | \n",
" | Refer to `numpy.sum` for full documentation.\n",
" | \n",
" | See Also\n",
" | --------\n",
" | numpy.sum : equivalent function\n",
" | \n",
" | swapaxes(...)\n",
" | a.swapaxes(axis1, axis2)\n",
" | \n",
" | Return a view of the array with àxis1` and àxis2` interchanged.\
n",
" | \n",
" | Refer to `numpy.swapaxes` for full documentation.\n",
" | \n",
" | See Also\n",
" | --------\n",
" | numpy.swapaxes : equivalent function\n",
" | \n",
" | take(...)\n",
" | a.take(indices, axis=None, out=None, mode='raise')\n",
" | \n",
" | Return an array formed from the elements of à` at the given
indices.\n",
" | \n",
" | Refer to `numpy.take` for full documentation.\n",
" | \n",
" | See Also\n",
" | --------\n",
" | numpy.take : equivalent function\n",
" | \n",
" | tobytes(...)\n",
" | a.tobytes(order='C')\n",
" | \n",
" | Construct Python bytes containing the raw data bytes in the array.\
n",
" | \n",
" | Constructs Python bytes showing a copy of the raw contents of\n",
" | data memory. The bytes object is produced in C-order by default.\n",
" | This behavior is controlled by the `òrder`` parameter.\n",
" | \n",
" | \n",
" | Parameters\n",
" | ----------\n",
" | order : {'C', 'F', 'A'}, optional\n",
" | Controls the memory layout of the bytes object. 'C' means C-
order,\n",
" | 'F' means F-order, 'A' (short for *Any*) means 'F' if à` is\n",
" | Fortran contiguous, 'C' otherwise. Default is 'C'.\n",
" | \n",
" | Returns\n",
" | -------\n",
" | s : bytes\n",
" | Python bytes exhibiting a copy of à`'s raw data.\n",
" | \n",
" | See also\n",
" | --------\n",
" | frombuffer\n",
" | Inverse of this operation, construct a 1-dimensional array from
Python\n",
" | bytes.\n",
" | \n",
" | Examples\n",
" | --------\n",
" | >>> x = np.array([[0, 1], [2, 3]], dtype='<u2')\n",
" | >>> x.tobytes()\n",
" | b'\\x00\\x00\\x01\\x00\\x02\\x00\\x03\\x00'\n",
" | >>> x.tobytes('C') == x.tobytes()\n",
" | True\n",
" | >>> x.tobytes('F')\n",
" | b'\\x00\\x00\\x02\\x00\\x01\\x00\\x03\\x00'\n",
" | \n",
" | tofile(...)\n",
" | a.tofile(fid, sep=\"\", format=\"%s\")\n",
" | \n",
" | Write array to a file as text or binary (default).\n",
" | \n",
" | Data is always written in 'C' order, independent of the order of
à`.\n",
" | The data produced by this method can be recovered using the
function\n",
" | fromfile().\n",
" | \n",
" | Parameters\n",
" | ----------\n",
" | fid : file or str or Path\n",
" | An open file object, or a string containing a filename.\n",
" | \n",
" | \n",
" | sep : str\n",
" | Separator between array items for text output.\n",
" | If \"\" (empty), a binary file is written, equivalent to\n",
" | ``file.write(a.tobytes())``.\n",
" | format : str\n",
" | Format string for text file output.\n",
" | Each entry in the array is formatted to text by first
converting\n",
" | it to the closest Python type, and then using \"format\" %
item.\n",
" | \n",
" | Notes\n",
" | -----\n",
" | This is a convenience function for quick storage of array data.\n",
" | Information on endianness and precision is lost, so this method is
not a\n",
" | good choice for files intended to archive data or transport data
between\n",
" | machines with different endianness. Some of these problems can be
overcome\n",
" | by outputting the data as text files, at the expense of speed and
file\n",
" | size.\n",
" | \n",
" | When fid is a file object, array contents are directly written to
the\n",
" | file, bypassing the file object's ``write`` method. As a result,
tofile\n",
" | cannot be used with files objects supporting compression (e.g.,
GzipFile)\n",
" | or file-like objects that do not support ``fileno()`` (e.g.,
BytesIO).\n",
" | \n",
" | tolist(...)\n",
" | a.tolist()\n",
" | \n",
" | Return the array as an `à.ndim``-levels deep nested list of Python
scalars.\n",
" | \n",
" | Return a copy of the array data as a (nested) Python list.\n",
" | Data items are converted to the nearest compatible builtin Python
type, via\n",
" | the `~numpy.ndarray.item` function.\n",
" | \n",
" | If `à.ndim`` is 0, then since the depth of the nested list is 0, it
will\n",
" | not be a list at all, but a simple Python scalar.\n",
" | \n",
" | Parameters\n",
" | ----------\n",
" | none\n",
" | \n",
" | Returns\n",
" | -------\n",
" | y : object, or list of object, or list of list of object, or ...\n",
" | The possibly nested list of array elements.\n",
" | \n",
" | Notes\n",
" | -----\n",
" | The array may be recreated via `à = np.array(a.tolist())``,
although this\n",
" | may sometimes lose precision.\n",
" | \n",
" | Examples\n",
" | --------\n",
" | For a 1D array, `à.tolist()`` is almost the same as ``list(a)``,\
n",
" | except that ``tolist`` changes numpy scalars to Python scalars:\n",
" | \n",
" | >>> a = np.uint32([1, 2])\n",
" | >>> a_list = list(a)\n",
" | >>> a_list\n",
" | [1, 2]\n",
" | >>> type(a_list[0])\n",
" | <class 'numpy.uint32'>\n",
" | >>> a_tolist = a.tolist()\n",
" | >>> a_tolist\n",
" | [1, 2]\n",
" | >>> type(a_tolist[0])\n",
" | <class 'int'>\n",
" | \n",
" | Additionally, for a 2D array, ``tolist`` applies recursively:\n",
" | \n",
" | >>> a = np.array([[1, 2], [3, 4]])\n",
" | >>> list(a)\n",
" | [array([1, 2]), array([3, 4])]\n",
" | >>> a.tolist()\n",
" | [[1, 2], [3, 4]]\n",
" | \n",
" | The base case for this recursion is a 0D array:\n",
" | \n",
" | >>> a = np.array(1)\n",
" | >>> list(a)\n",
" | ...\n",
" | TypeError: iteration over a 0-d array\n",
" | >>> a.tolist()\n",
" | 1\n",
" | \n",
" | tostring(...)\n",
" | a.tostring(order='C')\n",
" | \n",
" | A compatibility alias for `tobytes`, with exactly the same
behavior.\n",
" | \n",
" | Despite its name, it returns `bytes` not `str`\\ s.\n",
" | \n",
" | .. deprecated:: 1.19.0\n",
" | \n",
" | trace(...)\n",
" | a.trace(offset=0, axis1=0, axis2=1, dtype=None, out=None)\n",
" | \n",
" | Return the sum along diagonals of the array.\n",
" | \n",
" | Refer to `numpy.trace` for full documentation.\n",
" | \n",
" | See Also\n",
" | --------\n",
" | numpy.trace : equivalent function\n",
" | \n",
" | transpose(...)\n",
" | a.transpose(*axes)\n",
" | \n",
" | Returns a view of the array with axes transposed.\n",
" | \n",
" | Refer to `numpy.transpose` for full documentation.\n",
" | \n",
" | Parameters\n",
" | ----------\n",
" | axes : None, tuple of ints, or `n` ints\n",
" | \n",
" | * None or no argument: reverses the order of the axes.\n",
" | \n",
" | * tuple of ints: ì` in the `j`-th place in the tuple means that
the\n",
" | array's ì`-th axis becomes the transposed array's `j`-th axis.\
n",
" | \n",
" | * `n` ints: same as an n-tuple of the same ints (this form is\n",
" | intended simply as a \"convenience\" alternative to the tuple
form).\n",
" | \n",
" | Returns\n",
" | -------\n",
" | p : ndarray\n",
" | View of the array with its axes suitably permuted.\n",
" | \n",
" | See Also\n",
" | --------\n",
" | transpose : Equivalent function.\n",
" | ndarray.T : Array property returning the array transposed.\n",
" | ndarray.reshape : Give a new shape to an array without changing its
data.\n",
" | \n",
" | Examples\n",
" | --------\n",
" | >>> a = np.array([[1, 2], [3, 4]])\n",
" | >>> a\n",
" | array([[1, 2],\n",
" | [3, 4]])\n",
" | array([[1, 3],\n",
" | [2, 4]])\n",
" | >>> a.transpose((1, 0))\n",
" | array([[1, 3],\n",
" | [2, 4]])\n",
" | >>> a.transpose(1, 0)\n",
" | array([[1, 3],\n",
" | [2, 4]])\n",
" | \n",
" | >>> a = np.array([1, 2, 3, 4])\n",
" | >>> a\n",
" | array([1, 2, 3, 4])\n",
" | array([1, 2, 3, 4])\n",
" | \n",
" | var(...)\n",
" | a.var(axis=None, dtype=None, out=None, ddof=0, keepdims=False, *,
where=True)\n",
" | \n",
" | Returns the variance of the array elements, along given axis.\n",
" | \n",
" | Refer to `numpy.var` for full documentation.\n",
" | \n",
" | See Also\n",
" | --------\n",
" | numpy.var : equivalent function\n",
" | \n",
" | view(...)\n",
" | a.view([dtype][, type])\n",
" | \n",
" | New view of array with the same data.\n",
" | \n",
" | .. note::\n",
" | Passing None for ``dtype`` is different from omitting the
parameter,\n",
" | since the former invokes ``dtype(None)`` which is an alias for\
n",
" | ``dtype('float_')``.\n",
" | \n",
" | Parameters\n",
" | ----------\n",
" | dtype : data-type or ndarray sub-class, optional\n",
" | Data-type descriptor of the returned view, e.g., float32 or
int16.\n",
" | Omitting it results in the view having the same data-type as
à`.\n",
" | This argument can also be specified as an ndarray sub-class,
which\n",
" | then specifies the type of the returned object (this is
equivalent to\n",
" | setting the ``type`` parameter).\n",
" | type : Python type, optional\n",
" | Type of the returned view, e.g., ndarray or matrix. Again,
omission\n",
" | of the parameter results in type preservation.\n",
" | \n",
" | Notes\n",
" | -----\n",
" | `à.view()`` is used two different ways:\n",
" | \n",
" | `à.view(some_dtype)`` or `à.view(dtype=some_dtype)`` constructs a
view\n",
" | of the array's memory with a different data-type. This can cause a\
n",
" | reinterpretation of the bytes of memory.\n",
" | \n",
" | `à.view(ndarray_subclass)`` or `à.view(type=ndarray_subclass)``
just\n",
" | returns an instance of `ndarray_subclass` that looks at the same
array\n",
" | (same shape, dtype, etc.) This does not cause a reinterpretation of
the\n",
" | memory.\n",
" | \n",
" | For `à.view(some_dtype)``, if ``some_dtype`` has a different number
of\n",
" | bytes per entry than the previous dtype (for example, converting a
regular\n",
" | array to a structured array), then the last axis of `à`` must be\
n",
" | contiguous. This axis will be resized in the result.\n",
" | \n",
" | Only the last axis needs to be contiguous. Previously, the entire
array\n",
" | had to be C-contiguous.\n",
" | \n",
" | Examples\n",
" | --------\n",
" | >>> x = np.array([(1, 2)], dtype=[('a', np.int8), ('b', np.int8)])\
n",
" | \n",
" | Viewing array data using a different type and dtype:\n",
" | \n",
" | >>> y = x.view(dtype=np.int16, type=np.matrix)\n",
" | >>> y\n",
" | matrix([[513]], dtype=int16)\n",
" | >>> print(type(y))\n",
" | <class 'numpy.matrix'>\n",
" | \n",
" | Creating a view on a structured array so it can be used in
calculations\n",
" | \n",
" | >>> x = np.array([(1, 2),(3,4)], dtype=[('a', np.int8), ('b',
np.int8)])\n",
" | >>> xv = x.view(dtype=np.int8).reshape(-1,2)\n",
" | >>> xv\n",
" | array([[1, 2],\n",
" | [3, 4]], dtype=int8)\n",
" | >>> xv.mean(0)\n",
" | array([2., 3.])\n",
" | \n",
" | Making changes to the view changes the underlying array\n",
" | \n",
" | >>> xv[0,1] = 20\n",
" | >>> x\n",
" | array([(1, 20), (3, 4)], dtype=[('a', 'i1'), ('b', 'i1')])\n",
" | \n",
" | Using a view to convert an array to a recarray:\n",
" | \n",
" | >>> z = x.view(np.recarray)\n",
" | >>> z.a\n",
" | array([1, 3], dtype=int8)\n",
" | \n",
" | Views share data:\n",
" | \n",
" | >>> x[0] = (9, 10)\n",
" | >>> z[0]\n",
" | (9, 10)\n",
" | \n",
" | Views that change the dtype size (bytes per entry) should normally
be\n",
" | avoided on arrays defined by slices, transposes, fortran-ordering,
etc.:\n",
" | \n",
" | >>> x = np.array([[1, 2, 3], [4, 5, 6]], dtype=np.int16)\n",
" | >>> y = x[:, ::2]\n",
" | >>> y\n",
" | array([[1, 3],\n",
" | [4, 6]], dtype=int16)\n",
" | >>> y.view(dtype=[('width', np.int16), ('length', np.int16)])\n",
" | ...\n",
" | ValueError: To change to a dtype of a different size, the last axis
must be contiguous\n",
" | >>> z = y.copy()\n",
" | >>> z.view(dtype=[('width', np.int16), ('length', np.int16)])\n",
" | array([[(1, 3)],\n",
" | [(4, 6)]], dtype=[('width', '<i2'), ('length', '<i2')])\n",
" | \n",
" | However, views that change dtype are totally fine for arrays with a\
n",
" | contiguous last axis, even if the rest of the axes are not C-
contiguous:\n",
" | \n",
" | >>> x = np.arange(2 * 3 * 4, dtype=np.int8).reshape(2, 3, 4)\n",
" | >>> x.transpose(1, 0, 2).view(np.int16)\n",
" | array([[[ 256, 770],\n",
" | [3340, 3854]],\n",
" | <BLANKLINE>\n",
" | [[1284, 1798],\n",
" | [4368, 4882]],\n",
" | <BLANKLINE>\n",
" | [[2312, 2826],\n",
" | [5396, 5910]]], dtype=int16)\n",
" | \n",
" | ----------------------------------------------------------------------\
n",
" | Class methods defined here:\n",
" | \n",
" | __class_getitem__(...) from builtins.type\n",
" | a.__class_getitem__(item, /)\n",
" | \n",
" | Return a parametrized wrapper around the `~numpy.ndarray` type.\n",
" | \n",
" | .. versionadded:: 1.22\n",
" | \n",
" | Returns\n",
" | -------\n",
" | alias : types.GenericAlias\n",
" | A parametrized `~numpy.ndarray` type.\n",
" | \n",
" | Examples\n",
" | --------\n",
" | >>> from typing import Any\n",
" | >>> import numpy as np\n",
" | \n",
" | >>> np.ndarray[Any, np.dtype[Any]]\n",
" | numpy.ndarray[typing.Any, numpy.dtype[typing.Any]]\n",
" | \n",
" | Notes\n",
" | -----\n",
" | This method is only available for python 3.9 and later.\n",
" | \n",
" | See Also\n",
" | --------\n",
" | :pep:`585` : Type hinting generics in standard collections.\n",
" | numpy.typing.NDArray : An ndarray alias :term:`generic <generic
type>`\n",
" | \n",
" | ----------------------------------------------------------------------\
n",
" | Static methods defined here:\n",
" | \n",
" | __new__(*args, **kwargs) from builtins.type\n",
" | Create and return a new object. See help(type) for accurate
signature.\n",
" | \n",
" | ----------------------------------------------------------------------\
n",
" | Data descriptors defined here:\n",
" | \n",
" | T\n",
" | View of the transposed array.\n",
" | \n",
" | Same as ``self.transpose()``.\n",
" | \n",
" | Examples\n",
" | --------\n",
" | >>> a = np.array([[1, 2], [3, 4]])\n",
" | >>> a\n",
" | array([[1, 2],\n",
" | [3, 4]])\n",
" | >>> a.T\n",
" | array([[1, 3],\n",
" | [2, 4]])\n",
" | \n",
" | >>> a = np.array([1, 2, 3, 4])\n",
" | >>> a\n",
" | array([1, 2, 3, 4])\n",
" | >>> a.T\n",
" | array([1, 2, 3, 4])\n",
" | \n",
" | See Also\n",
" | --------\n",
" | transpose\n",
" | \n",
" | __array_interface__\n",
" | Array protocol: Python side.\n",
" | \n",
" | __array_priority__\n",
" | Array priority.\n",
" | \n",
" | __array_struct__\n",
" | Array protocol: C-struct side.\n",
" | \n",
" | base\n",
" | Base object if memory is from some other object.\n",
" | \n",
" | Examples\n",
" | --------\n",
" | The base of an array that owns its memory is None:\n",
" | \n",
" | >>> x = np.array([1,2,3,4])\n",
" | >>> x.base is None\n",
" | True\n",
" | \n",
" | Slicing creates a view, whose memory is shared with x:\n",
" | \n",
" | >>> y = x[2:]\n",
" | >>> y.base is x\n",
" | True\n",
" | \n",
" | ctypes\n",
" | An object to simplify the interaction of the array with the ctypes\
n",
" | module.\n",
" | \n",
" | This attribute creates an object that makes it easier to use arrays\
n",
" | when calling shared libraries with the ctypes module. The returned\
n",
" | object has, among others, data, shape, and strides attributes (see\
n",
" | Notes below) which themselves return ctypes objects that can be
used\n",
" | as arguments to a shared library.\n",
" | \n",
" | Parameters\n",
" | ----------\n",
" | None\n",
" | \n",
" | Returns\n",
" | -------\n",
" | c : Python object\n",
" | Possessing attributes data, shape, strides, etc.\n",
" | \n",
" | See Also\n",
" | --------\n",
" | numpy.ctypeslib\n",
" | \n",
" | Notes\n",
" | -----\n",
" | Below are the public attributes of this object which were
documented\n",
" | in \"Guide to NumPy\" (we have omitted undocumented public
attributes,\n",
" | as well as documented private attributes):\n",
" | \n",
" | .. autoattribute:: numpy.core._internal._ctypes.data\n",
" | :noindex:\n",
" | \n",
" | .. autoattribute:: numpy.core._internal._ctypes.shape\n",
" | :noindex:\n",
" | \n",
" | .. autoattribute:: numpy.core._internal._ctypes.strides\n",
" | :noindex:\n",
" | \n",
" | .. automethod:: numpy.core._internal._ctypes.data_as\n",
" | :noindex:\n",
" | \n",
" | .. automethod:: numpy.core._internal._ctypes.shape_as\n",
" | :noindex:\n",
" | \n",
" | .. automethod:: numpy.core._internal._ctypes.strides_as\n",
" | :noindex:\n",
" | \n",
" | If the ctypes module is not available, then the ctypes attribute\n",
" | of array objects still returns something useful, but ctypes objects\
n",
" | are not returned and errors may be raised instead. In particular,\
n",
" | the object will still have the `às_parameter`` attribute which
will\n",
" | return an integer equal to the data attribute.\n",
" | \n",
" | Examples\n",
" | --------\n",
" | >>> import ctypes\n",
" | >>> x = np.array([[0, 1], [2, 3]], dtype=np.int32)\n",
" | >>> x\n",
" | array([[0, 1],\n",
" | [2, 3]], dtype=int32)\n",
" | >>> x.ctypes.data\n",
" | 31962608 # may vary\n",
" | >>> x.ctypes.data_as(ctypes.POINTER(ctypes.c_uint32))\n",
" | <__main__.LP_c_uint object at 0x7ff2fc1fc200> # may vary\n",
" | c_uint(0)\n",
" | c_ulong(4294967296)\n",
" | >>> x.ctypes.shape\n",
" | <numpy.core._internal.c_long_Array_2 object at 0x7ff2fc1fce60> # may
vary\n",
" | >>> x.ctypes.strides\n",
" | <numpy.core._internal.c_long_Array_2 object at 0x7ff2fc1ff320> # may
vary\n",
" | \n",
" | data\n",
" | Python buffer object pointing to the start of the array's data.\n",
" | \n",
" | dtype\n",
" | Data-type of the array's elements.\n",
" | \n",
" | \n",
" | Setting `àrr.dtype`` is discouraged and may be deprecated in
the\n",
" | future. Setting will replace the ``dtype`` without modifying
the\n",
" | memory (see also `ndarray.view` and `ndarray.astype`).\n",
" | \n",
" | Parameters\n",
" | ----------\n",
" | None\n",
" | \n",
" | Returns\n",
" | -------\n",
" | d : numpy dtype object\n",
" | \n",
" | See Also\n",
" | --------\n",
" | ndarray.astype : Cast the values contained in the array to a new
data-type.\n",
" | ndarray.view : Create a view of the same data but a different data-
type.\n",
" | numpy.dtype\n",
" | \n",
" | Examples\n",
" | --------\n",
" | >>> x\n",
" | array([[0, 1],\n",
" | [2, 3]])\n",
" | >>> x.dtype\n",
" | dtype('int32')\n",
" | >>> type(x.dtype)\n",
" | <type 'numpy.dtype'>\n",
" | \n",
" | flags\n",
" | Information about the memory layout of the array.\n",
" | \n",
" | Attributes\n",
" | ----------\n",
" | C_CONTIGUOUS (C)\n",
" | The data is in a single, C-style contiguous segment.\n",
" | F_CONTIGUOUS (F)\n",
" | The data is in a single, Fortran-style contiguous segment.\n",
" | OWNDATA (O)\n",
" | The array owns the memory it uses or borrows it from another
object.\n",
" | WRITEABLE (W)\n",
" | The data area can be written to. Setting this to False locks\
n",
" | the data, making it read-only. A view (slice, etc.) inherits
WRITEABLE\n",
" | from its base array at creation time, but a view of a writeable\
n",
" | array may be subsequently locked while the base array remains
writeable.\n",
" | (The opposite is not true, in that a view of a locked array may
not\n",
" | be made writeable. However, currently, locking a base object
does not\n",
" | lock any views that already reference it, so under that
circumstance it\n",
" | is possible to alter the contents of a locked array via a
previously\n",
" | created writeable view onto it.) Attempting to change a non-
writeable\n",
" | array raises a RuntimeError exception.\n",
" | ALIGNED (A)\n",
" | The data and all elements are aligned appropriately for the
hardware.\n",
" | WRITEBACKIFCOPY (X)\n",
" | This array is a copy of some other array. The C-API function\n",
" | PyArray_ResolveWritebackIfCopy must be called before
deallocating\n",
" | to the base array will be updated with the contents of this
array.\n",
" | FNC\n",
" | F_CONTIGUOUS and not C_CONTIGUOUS.\n",
" | FORC\n",
" | F_CONTIGUOUS or C_CONTIGUOUS (one-segment test).\n",
" | BEHAVED (B)\n",
" | ALIGNED and WRITEABLE.\n",
" | CARRAY (CA)\n",
" | BEHAVED and C_CONTIGUOUS.\n",
" | FARRAY (FA)\n",
" | BEHAVED and F_CONTIGUOUS and not C_CONTIGUOUS.\n",
" | \n",
" | Notes\n",
" | -----\n",
" | The `flags` object can be accessed dictionary-like (as in
`à.flags['WRITEABLE']``),\n",
" | or by using lowercased attribute names (as in
`à.flags.writeable``). Short flag\n",
" | names are only supported in dictionary access.\n",
" | \n",
" | Only the WRITEBACKIFCOPY, WRITEABLE, and ALIGNED flags can be\n",
" | changed by the user, via direct assignment to the attribute or
dictionary\n",
" | entry, or by calling `ndarray.setflags`.\n",
" | \n",
" | The array flags cannot be set arbitrarily:\n",
" | \n",
" | - WRITEBACKIFCOPY can only be set ``False``.\n",
" | - ALIGNED can only be set ``True`` if the data is truly aligned.\n",
" | - WRITEABLE can only be set ``True`` if the array owns its own
memory\n",
" | or the ultimate owner of the memory exposes a writeable buffer\n",
" | interface or is a string.\n",
" | \n",
" | Arrays can be both C-style and Fortran-style contiguous
simultaneously.\n",
" | This is clear for 1-dimensional arrays, but can also be true for
higher\n",
" | dimensional arrays.\n",
" | \n",
" | Even for contiguous arrays a stride for a given dimension\n",
" | `àrr.strides[dim]`` may be *arbitrary* if `àrr.shape[dim] == 1``\
n",
" | or the array has no elements.\n",
" | It does *not* generally hold that ``self.strides[-1] ==
self.itemsize``\n",
" | for C-style contiguous arrays or ``self.strides[0] ==
self.itemsize`` for\n",
" | Fortran-style contiguous arrays is true.\n",
" | \n",
" | flat\n",
" | A 1-D iterator over the array.\n",
" | \n",
" | This is a `numpy.flatiter` instance, which acts similarly to, but is
not\n",
" | a subclass of, Python's built-in iterator object.\n",
" | \n",
" | See Also\n",
" | --------\n",
" | flatten : Return a copy of the array collapsed into one dimension.\
n",
" | \n",
" | flatiter\n",
" | \n",
" | Examples\n",
" | --------\n",
" | >>> x = np.arange(1, 7).reshape(2, 3)\n",
" | >>> x\n",
" | array([[1, 2, 3],\n",
" | [4, 5, 6]])\n",
" | >>> x.flat[3]\n",
" | 4\n",
" | >>> x.T\n",
" | array([[1, 4],\n",
" | [2, 5],\n",
" | [3, 6]])\n",
" | >>> x.T.flat[3]\n",
" | 5\n",
" | >>> type(x.flat)\n",
" | <class 'numpy.flatiter'>\n",
" | \n",
" | An assignment example:\n",
" | \n",
" | >>> x.flat = 3; x\n",
" | array([[3, 3, 3],\n",
" | [3, 3, 3]])\n",
" | >>> x.flat[[1,4]] = 1; x\n",
" | array([[3, 1, 3],\n",
" | [3, 1, 3]])\n",
" | \n",
" | imag\n",
" | The imaginary part of the array.\n",
" | \n",
" | Examples\n",
" | --------\n",
" | >>> x = np.sqrt([1+0j, 0+1j])\n",
" | >>> x.imag\n",
" | array([ 0. , 0.70710678])\n",
" | >>> x.imag.dtype\n",
" | \n",
" | itemsize\n",
" | Length of one array element in bytes.\n",
" | \n",
" | Examples\n",
" | --------\n",
" | >>> x = np.array([1,2,3], dtype=np.float64)\n",
" | 8\n",
" | >>> x = np.array([1,2,3], dtype=np.complex128)\n",
" | 16\n",
" | \n",
" | nbytes\n",
" | Total bytes consumed by the elements of the array.\n",
" | \n",
" | Notes\n",
" | -----\n",
" | Does not include memory consumed by non-element attributes of the\
n",
" | array object.\n",
" | \n",
" | Examples\n",
" | --------\n",
" | >>> x = np.zeros((3,5,2), dtype=np.complex128)\n",
" | >>> x.nbytes\n",
" | 480\n",
" | >>> np.prod(x.shape) * x.itemsize\n",
" | 480\n",
" | \n",
" | ndim\n",
" | Number of array dimensions.\n",
" | \n",
" | Examples\n",
" | --------\n",
" | >>> x = np.array([1, 2, 3])\n",
" | >>> x.ndim\n",
" | 1\n",
" | >>> y = np.zeros((2, 3, 4))\n",
" | >>> y.ndim\n",
" | 3\n",
" | \n",
" | real\n",
" | The real part of the array.\n",
" | \n",
" | Examples\n",
" | --------\n",
" | >>> x = np.sqrt([1+0j, 0+1j])\n",
" | >>> x.real\n",
" | array([ 1. , 0.70710678])\n",
" | >>> x.real.dtype\n",
" | \n",
" | See Also\n",
" | --------\n",
" | numpy.real : equivalent function\n",
" | \n",
" | shape\n",
" | Tuple of array dimensions.\n",
" | \n",
" | The shape property is usually used to get the current shape of an
array,\n",
" | but may also be used to reshape the array in-place by assigning a
tuple of\n",
" | array dimensions to it. As with `numpy.reshape`, one of the new
shape\n",
" | dimensions can be -1, in which case its value is inferred from the
size of\n",
" | the array and the remaining dimensions. Reshaping an array in-place
will\n",
" | fail if a copy is required.\n",
" | \n",
" | \n",
" | Setting `àrr.shape`` is discouraged and may be deprecated in
the\n",
" | future. Using `ndarray.reshape` is the preferred approach.\n",
" | \n",
" | Examples\n",
" | --------\n",
" | >>> x = np.array([1, 2, 3, 4])\n",
" | >>> x.shape\n",
" | (4,)\n",
" | >>> y = np.zeros((2, 3, 4))\n",
" | >>> y.shape\n",
" | (2, 3, 4)\n",
" | >>> y.shape = (3, 8)\n",
" | >>> y\n",
" | array([[ 0., 0., 0., 0., 0., 0., 0., 0.],\n",
" | [ 0., 0., 0., 0., 0., 0., 0., 0.],\n",
" | [ 0., 0., 0., 0., 0., 0., 0., 0.]])\n",
" | >>> y.shape = (3, 6)\n",
" | ValueError: total size of new array must be unchanged\n",
" | >>> np.zeros((4,2))[::2].shape = (-1,)\n",
" | AttributeError: Incompatible shape for in-place modification. Use\
n",
" | `.reshape()` to make a copy with the desired shape.\n",
" | \n",
" | See Also\n",
" | --------\n",
" | numpy.shape : Equivalent getter function.\n",
" | numpy.reshape : Function similar to setting ``shape``.\n",
" | ndarray.reshape : Method similar to setting ``shape``.\n",
" | \n",
" | size\n",
" | \n",
" | Equal to ``np.prod(a.shape)``, i.e., the product of the array's\n",
" | dimensions.\n",
" | \n",
" | Notes\n",
" | -----\n",
" | à.size` returns a standard arbitrary precision Python integer.
This\n",
" | may not be the case with other methods of obtaining the same value\
n",
" | (like the suggested ``np.prod(a.shape)``, which returns an instance\
n",
" | of ``np.int_``), and may be relevant if the value is used further
in\n",
" | calculations that may overflow a fixed size integer type.\n",
" | \n",
" | Examples\n",
" | --------\n",
" | >>> x = np.zeros((3, 5, 2), dtype=np.complex128)\n",
" | >>> x.size\n",
" | 30\n",
" | >>> np.prod(x.shape)\n",
" | 30\n",
" | \n",
" | strides\n",
" | Tuple of bytes to step in each dimension when traversing an array.\
n",
" | \n",
" | The byte offset of element ``(i[0], i[1], ..., i[n])`` in an array
à`\n",
" | is::\n",
" | \n",
" | offset = sum(np.array(i) * a.strides)\n",
" | \n",
" | A more detailed explanation of strides can be found in the\n",
" | \"ndarray.rst\" file in the NumPy reference guide.\n",
" | \n",
" | \n",
" | Setting `àrr.strides`` is discouraged and may be deprecated in
the\n",
" | future. `numpy.lib.stride_tricks.as_strided` should be
preferred\n",
" | to create a new view of the same data in a safer way.\n",
" | \n",
" | Notes\n",
" | -----\n",
" | Imagine an array of 32-bit integers (each 4 bytes)::\n",
" | \n",
" | x = np.array([[0, 1, 2, 3, 4],\n",
" | [5, 6, 7, 8, 9]], dtype=np.int32)\n",
" | \n",
" | This array is stored in memory as 40 bytes, one after the other\n",
" | (known as a contiguous block of memory). The strides of an array
tell\n",
" | us how many bytes we have to skip in memory to move to the next
position\n",
" | along a certain axis. For example, we have to skip 4 bytes (1
value) to\n",
" | move to the next column, but 20 bytes (5 values) to get to the same\
n",
" | position in the next row. As such, the strides for the array `x`
will be\n",
" | ``(20, 4)``.\n",
" | \n",
" | See Also\n",
" | --------\n",
" | numpy.lib.stride_tricks.as_strided\n",
" | \n",
" | Examples\n",
" | --------\n",
" | >>> y = np.reshape(np.arange(2*3*4), (2,3,4))\n",
" | >>> y\n",
" | array([[[ 0, 1, 2, 3],\n",
" | [ 4, 5, 6, 7],\n",
" | [ 8, 9, 10, 11]],\n",
" | [[12, 13, 14, 15],\n",
" | [16, 17, 18, 19],\n",
" | [20, 21, 22, 23]]])\n",
" | >>> y.strides\n",
" | (48, 16, 4)\n",
" | >>> y[1,1,1]\n",
" | 17\n",
" | >>> offset=sum(y.strides * np.array((1,1,1)))\n",
" | >>> offset/y.itemsize\n",
" | 17\n",
" | \n",
" | >>> x = np.reshape(np.arange(5*6*7*8),
(5,6,7,8)).transpose(2,3,1,0)\n",
" | >>> x.strides\n",
" | (32, 4, 224, 1344)\n",
" | >>> i = np.array([3,5,2,2])\n",
" | >>> offset = sum(i * x.strides)\n",
" | >>> x[3,5,2,2]\n",
" | 813\n",
" | >>> offset / x.itemsize\n",
" | 813\n",
" | \n",
" | ----------------------------------------------------------------------\
n",
" | Data and other attributes defined here:\n",
" | \n",
" | __hash__ = None\n",
"\n"
]
}
],
"source": [
"help(a.flatten())"
]
},
{
"metadata": {},
"source": [
"# Copying Arrays"
]
},
{
"metadata": {},
"source": [
"simply using \"=\" does not make a copy, but much like with lists, you will
just have multiple names pointing to the same ndarray object\n",
"\n",
"Therefore, we need to understand if two arrays, À` and `B` point to:\n",
"* the same array, including shape and data/memory space\n",
"* the same data/memory space, but perhaps different shapes (a _view_)\n",
"* a separate cpy of the data (i.e. stored completely separately in memory)\n",
"\n",
"All of these are possible:\n",
"* `B = A`\n",
" \n",
" this is _assignment_. No copy is made. À` and `B` point to the same data
in memory and share the same shape, etc. They are just two different labels for
the same object in memory\n",
" \n",
"\n",
"* `B = A[:]`\n",
"\n",
" this is a _view_ or _shallow copy_. The shape info for A and B are stored
independently, but both point to the same memory location for data\n",
" \n",
" \n",
"* `B = A.copy()`\n",
"\n",
" this is a _deep_ copy. A completely separate object will be created in
memory, with a completely separate location in memory.\n",
" \n",
"Let's look at examples"
]
},
{
"metadata": {},
"outputs": [
{
"name": "stdout",
"text": [
"[0 1 2 3 4 5 6 7 8 9]\n"
]
}
],
"source": [
"print(a)\n"
]
},
{
"metadata": {},
"source": [
"Here is assignment -- we can just use the ìs` operator to test for equality"
]
},
{
"metadata": {},
"outputs": [
{
"data": {
"text/plain": [
"True"
]
},
"metadata": {},
}
],
"source": [
"b = a\n",
"b is a"
]
},
{
"metadata": {},
"source": [
"Since `b` and à` are the same, changes to the shape of one are reflected in
the other -- no copy is made."
]
},
{
"metadata": {},
"outputs": [
{
"name": "stdout",
"text": [
"[[0 1 2 3 4]\n",
" [5 6 7 8 9]]\n"
]
},
{
"data": {
"text/plain": [
"(2, 5)"
]
},
"metadata": {},
}
],
"source": [
"b.shape = (2,5)\n",
"print(b)\n",
"a.shape"
]
},
{
"metadata": {},
"outputs": [
{
"data": {
"text/plain": [
"True"
]
},
"metadata": {},
}
],
"source": [
"b is a"
]
},
{
"metadata": {},
"outputs": [
{
"name": "stdout",
"text": [
"[[0 1 2 3 4]\n",
" [5 6 7 8 9]]\n"
]
}
],
"source": [
"print(a)"
]
},
{
"metadata": {},
"source": [
"a shallow copy creates a new *view* into the array -- the data is the same,
but the array properties can be different"
]
},
{
"metadata": {},
"outputs": [
{
"name": "stdout",
"text": [
"[[ 0 1 2 3]\n",
" [ 4 5 6 7]\n",
" [ 8 9 10 11]]\n",
"[ 0 1 2 3 4 5 6 7 8 9 10 11]\n"
]
}
],
"source": [
"c = a[:]\n",
"a.shape = (3,4)\n",
"print(a)\n",
"print(c) \n"
]
},
{
"metadata": {},
"source": [
"since the underlying data is the same memory, changing an element of one is
reflected in the other"
]
},
{
"metadata": {},
"outputs": [
{
"name": "stdout",
"text": [
"[[ 0 -1 2 3]\n",
" [ 4 5 6 7]\n",
" [ 8 9 10 11]]\n",
"[ 0 -1 2 3 4 5 6 7 8 9 10 11]\n"
]
}
],
"source": [
"c[1] = -1\n",
"print(a)\n",
"print(c)\n"
]
},
{
"metadata": {},
"source": [
"Even slices into an array are just views, still pointing to the same memory"
]
},
{
"metadata": {},
"outputs": [
{
"name": "stdout",
"text": [
"[3 4 5 6 7]\n"
]
}
],
"source": [
"d = c[3:8]\n",
"print(d)"
]
},
{
"metadata": {},
"outputs": [],
"source": [
"d[:] = 0"
]
},
{
"metadata": {},
"outputs": [
{
"name": "stdout",
"text": [
"[[ 0 -1 2 0]\n",
" [ 0 0 0 0]\n",
" [ 8 9 10 11]]\n",
"[ 0 -1 2 0 0 0 0 0 8 9 10 11]\n",
"[0 0 0 0 0]\n"
]
}
],
"source": [
"print(a)\n",
"print(c)\n",
"print(d)"
]
},
{
"metadata": {
"collapsed": true
},
"source": [
"There are lots of ways to inquire if two arrays are the same, views, own their
own data, etc"
]
},
{
"metadata": {},
"outputs": [
{
"name": "stdout",
"text": [
"False\n",
"True\n",
"False\n",
"True\n"
]
}
],
"source": [
"print(c is a)\n",
"print(c.base is a)\n",
"print(c.flags.owndata)\n",
"print(a.flags.owndata)\n"
]
},
{
"metadata": {},
"source": [
"to make a copy of the data of the array that you can deal with independently
of the original, you need a deep copy"
]
},
{
"metadata": {},
"outputs": [
{
"name": "stdout",
"text": [
"[[ 0 -1 2 0]\n",
" [ 0 0 0 0]\n",
" [ 8 9 10 11]]\n",
"[[0 0 0 0]\n",
" [0 0 0 0]\n",
" [0 0 0 0]]\n"
]
}
],
"source": [
"d = a.copy()\n",
"d[:,:] = 0.0\n",
"\n",
"print(a)\n",
"print(d)"
]
},
{
"metadata": {},
"source": [
"# Boolean Indexing"
]
},
{
"metadata": {},
"source": [
"There are lots of fun ways to index arrays to access only those elements that
meet a certain condition"
]
},
{
"metadata": {},
"outputs": [
{
"data": {
"text/plain": [
"array([[ 0, 1, 2, 3],\n",
" [ 4, 5, 6, 7],\n",
" [ 8, 9, 10, 11]])"
]
},
"metadata": {},
}
],
"source": [
"a"
]
},
{
"metadata": {},
"source": [
"Here we set all the elements in the array that are > 4 to zero"
]
},
{
"metadata": {},
"outputs": [
{
"data": {
"text/plain": [
"array([[0, 1, 2, 3],\n",
" [4, 0, 0, 0],\n",
" [0, 0, 0, 0]])"
]
},
"metadata": {},
}
],
"source": [
"a[a > 4] =0\n",
"a"
]
},
{
"metadata": {},
"source": [
"and now, all the zeros to -1"
]
},
{
"metadata": {},
"outputs": [
{
"data": {
"text/plain": [
"array([[-1, 1, 2, 3],\n",
" [ 4, -1, -1, -1],\n",
" [-1, -1, -1, -1]])"
]
},
"metadata": {},
}
],
"source": [
"a[a == 0] = -1\n",
"a"
]
},
{
"metadata": {},
"outputs": [
{
"data": {
"text/plain": [
"array([[ True, False, False, False],\n",
" [False, True, True, True],\n",
" [ True, True, True, True]])"
]
},
"metadata": {},
}
],
"source": [
"a == -1"
]
},
{
"metadata": {},
"source": [
"if we have 2 tests, we need to use `logical_and()` or `logical_or()`"
]
},
{
"metadata": {},
"outputs": [
{
"data": {
"text/plain": [
"array([[ 0, 1, 2, 3],\n",
" [ 0, 0, 0, 0],\n",
" [ 0, 0, 10, 11]])"
]
},
"metadata": {},
}
],
"source": [
"a[np.logical_and(a > 3, a<=9)] = 0.0\n",
"a"
]
},
{
"metadata": {},
"source": [
"Our test that we index the array with returns a boolean array of the same
shape:"
]
},
{
"metadata": {},
"outputs": [
{
"data": {
"text/plain": [
"array([[False, False, False, False],\n",
" [False, False, False, False],\n",
" [False, False, True, True]])"
]
},
"metadata": {},
}
],
"source": [
"a >4"
]
},
{
"metadata": {},
"source": [
"# Avoiding Loops"
]
},
{
"metadata": {},
"source": [
"Python's default implementation (known as CPython) does some operations very
slowly.\n",
"This is in part due to the dynamic, interpreted nature of the language: the
fact that types are flexible, so that sequences of operations cannot be compiled
down to efficient machine code as in languages like C and Fortran.\n",
"Recently there have been various attempts to address this weakness: well-known
examples are the [PyPy](http://pypy.org/) project, a just-in-time compiled
implementation of Python; the [Cython](http://cython.org) project, which converts
Python code to compilable C code; and the [Numba](http://numba.pydata.org/)
project, which converts snippets of Python code to fast LLVM bytecode.\n",
"Each of these has its strengths and weaknesses, but it is safe to say that
none of the three approaches has yet surpassed the reach and popularity of the
standard CPython engine.\n",
"\n",
"The relative sluggishness of Python generally manifests itself in situations
where many small operations are being repeated – for instance looping over arrays
to operate on each element."
]
},
{
"metadata": {},
"source": [
"In general, you want to avoid loops over elements on an array.\n",
"\n",
"Here, let's create 1-d x and y coordinates and then try to fill some larger
array"
]
},
{
"metadata": {},
"outputs": [
{
"name": "stdout",
"text": [
"(32,)\n",
"(64,)\n"
]
},
{
"data": {
"text/plain": [
"array([0. , 0.015625, 0.03125 , 0.046875, 0.0625 , 0.078125,\n",
" 0.09375 , 0.109375, 0.125 , 0.140625, 0.15625 , 0.171875,\n",
" 0.1875 , 0.203125, 0.21875 , 0.234375, 0.25 , 0.265625,\n",
" 0.28125 , 0.296875, 0.3125 , 0.328125, 0.34375 , 0.359375,\n",
" 0.375 , 0.390625, 0.40625 , 0.421875, 0.4375 , 0.453125,\n",
" 0.46875 , 0.484375, 0.5 , 0.515625, 0.53125 , 0.546875,\n",
" 0.5625 , 0.578125, 0.59375 , 0.609375, 0.625 , 0.640625,\n",
" 0.65625 , 0.671875, 0.6875 , 0.703125, 0.71875 , 0.734375,\n",
" 0.75 , 0.765625, 0.78125 , 0.796875, 0.8125 , 0.828125,\n",
" 0.84375 , 0.859375, 0.875 , 0.890625, 0.90625 , 0.921875,\n",
" 0.9375 , 0.953125, 0.96875 , 0.984375])"
]
},
"metadata": {},
}
],
"source": [
"M = 32\n",
"N = 64\n",
"xmin = ymin = 0.0\n",
"xmax = ymax = 1.0\n",
"\n",
"x = np.linspace(xmin, xmax, M, endpoint=False)\n",
"y = np.linspace(ymin, ymax, N, endpoint=False)\n",
"\n",
"print(x.shape)\n",
"print(y.shape)\n",
"y"
]
},
{
"metadata": {},
"source": [
"we'll time out code"
]
},
{
"metadata": {},
"outputs": [],
"source": [
"import time\n",
"import numpy as np\n",
"M = 32\n",
"N = 64\n",
"\n",
"x = np.linspace(0.0, 1.0, M, endpoint=False)\n",
"y = np.linspace(0.0, 1.0, N, endpoint=False)\n",
"\n"
]
},
{
"metadata": {},
"outputs": [
{
"name": "stdout",
"text": [
"[[ 0. 0. 0. ... 0. 0.\n",
" 0. ]\n",
" [ 0. 0.00306796 0.00613588 ... 0.18605515 0.18906866\n",
" 0.1920804 ]\n",
" [ 0. 0.00613588 0.01227154 ... 0.365613 0.37131719\n",
" 0.37700741]\n",
" ...\n",
" [ 0. 0.08885355 0.17700422 ... -0.75520138 -0.69397146\n",
" -0.62725182]\n",
" [ 0. 0.09190896 0.18303989 ... -0.62005721 -0.54532499\n",
" -0.4659765 ]\n",
" [ 0. 0.0949635 0.18906866 ... -0.46325978 -0.37700741\n",
" -0.28734746]]\n",
"time elapsed: 0.005005598068237305 s\n"
]
}
],
"source": [
"t0 = time.time()\n",
"\n",
"g = np.zeros((M,N))\n",
"\n",
"for i in range(M):\n",
" for j in range(N):\n",
" g[i,j]=np.sin(2.0*np.pi*x[i]*y[j])\n",
"print(g)\n",
"print(\"time elapsed: {} s\" . format(t1-t0))\n",
" "
]
},
{
"metadata": {},
"source": [
"Now let's instead do this using all array syntax. First will extend our 1-d
coordinate arrays to be 2-d"
]
},
{
"metadata": {},
"outputs": [
{
"name": "stdout",
"text": [
"[0. 0.03125 0.0625 0.09375 0.125 0.15625 0.1875 0.21875 0.25\n",
" 0.28125 0.3125 0.34375 0.375 0.40625 0.4375 0.46875 0.5 0.53125\n",
" 0.5625 0.59375 0.625 0.65625 0.6875 0.71875 0.75 0.78125 0.8125\n",
" 0.84375 0.875 0.90625 0.9375 0.96875]\n",
"[0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.\n",
" 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.\n",
" 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
"[0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.\n",
" 0. 0. 0. 0. 0. 0. 0. 0.]\n",
"[0. 0.015625 0.03125 0.046875 0.0625 0.078125 0.09375 0.109375\n",
" 0.125 0.140625 0.15625 0.171875 0.1875 0.203125 0.21875 0.234375\n",
" 0.25 0.265625 0.28125 0.296875 0.3125 0.328125 0.34375 0.359375\n",
" 0.375 0.390625 0.40625 0.421875 0.4375 0.453125 0.46875 0.484375\n",
" 0.5 0.515625 0.53125 0.546875 0.5625 0.578125 0.59375 0.609375\n",
" 0.625 0.640625 0.65625 0.671875 0.6875 0.703125 0.71875 0.734375\n",
" 0.75 0.765625 0.78125 0.796875 0.8125 0.828125 0.84375 0.859375\n",
" 0.875 0.890625 0.90625 0.921875 0.9375 0.953125 0.96875 0.984375]\n"
]
}
],
"source": [
"x2d, y2d = np.meshgrid(x, y, indexing=\"ij\")\n",
"\n",
"print(x2d[:,0])\n",
"print(x2d[0,:])\n",
"\n",
"print(y2d[:,0])\n",
"print(y2d[0,:])\n"
]
},
{
"metadata": {},
"outputs": [
{
"name": "stdout",
"text": [
"time elapsed: 0.0 s\n"
]
}
],
"source": [
"g2 = np.sin(2.0*np.pi*x2d*y2d)\n",
"print(\"time elapsed: {} s\".format(t1-t0))"
]
},
{
"metadata": {},
"source": [
"## NumPy Standard Data Types\n",
"\n",
"NumPy arrays contain values of a single type, so it is important to have
detailed knowledge of those types and their limitations.\n",
"Because NumPy is built in C, the types will be familiar to users of C,
Fortran, and other related languages.\n",
"\n",
"The standard NumPy data types are listed in the following table.\n",
"Note that when constructing an array, they can be specified using a string:\
n",
"\n",
"```python\n",
"np.zeros(10, dtype='int16')\n",
"```\n",
"\n",
"Or using the associated NumPy object:\n",
"\n",
"```python\n",
"np.zeros(10, dtype=np.int16)\n",
"```"
]
},
{
"metadata": {},
"source": [
"| Data type\t | Description |\n",
"|---------------|-------------|\n",
"| ``bool_`` | Boolean (True or False) stored as a byte |\n",
"| `ìnt_`` | Default integer type (same as C ``long``; normally either
`ìnt64`` or `ìnt32``)| \n",
"| `ìntc`` | Identical to C `ìnt`` (normally `ìnt32`` or `ìnt64``)| \
n",
"| `ìntp`` | Integer used for indexing (same as C ``ssize_t``; normally
either `ìnt32`` or `ìnt64``)| \n",
"| `ìnt8`` | Byte (-128 to 127)| \n",
"| `ìnt16`` | Integer (-32768 to 32767)|\n",
"| `ìnt32`` | Integer (-2147483648 to 2147483647)|\n",
"| `ìnt64`` | Integer (-9223372036854775808 to 9223372036854775807)| \n",
"| `ùint8`` | Unsigned integer (0 to 255)| \n",
"| ``float_`` | Shorthand for ``float64``.| \n",
"| ``float16`` | Half precision float: sign bit, 5 bits exponent, 10 bits
mantissa| \n",
"| ``float32`` | Single precision float: sign bit, 8 bits exponent, 23 bits
mantissa| \n",
"| ``float64`` | Double precision float: sign bit, 11 bits exponent, 52 bits
mantissa| \n",
"| ``complex_`` | Shorthand for ``complex128``.| \n",
"| ``complex64`` | Complex number, represented by two 32-bit floats| \n",
"| ``complex128``| Complex number, represented by two 64-bit floats| "
]
},
{
"metadata": {},
"source": [
"More advanced type specification is possible, such as specifying big or little
endian numbers; for more information, refer to the [NumPy
documentation](http://numpy.org/).\n",
"NumPy also supports compound data types, which will be covered in [Structured
Data: NumPy's Structured Arrays](02.09-Structured-Data-NumPy.ipynb)."
]
},
{
"metadata": {},
"source": [
"## Array Concatenation and Splitting\n",
"\n",
"All of the preceding routines worked on single arrays. It's also possible to
combine multiple arrays into one, and to conversely split a single array into
multiple arrays. We'll take a look at those operations here."
]
},
{
"metadata": {},
"source": [
"### Concatenation of arrays\n",
"\n",
"Concatenation, or joining of two arrays in NumPy, is primarily accomplished
using the routines ``np.concatenate``, ``np.vstack``, and ``np.hstack``.\n",
"``np.concatenate`` takes a tuple or list of arrays as its first argument, as
we can see here:"
]
},
{
"metadata": {},
"outputs": [
{
"data": {
"text/plain": [
"array([4, 3, 2, 1, 1, 2, 3, 4])"
]
},
"metadata": {},
}
],
"source": [
"x = np.array([1, 2, 3, 4])\n",
"y = np.array([4, 3, 2, 1])\n",
"np.concatenate([y,x])"
]
},
{
"metadata": {},
"source": [
"You can also concatenate more than two arrays at once:"
]
},
{
"metadata": {},
"outputs": [
{
"name": "stdout",
"text": [
"[1 2 3 4 4 3 2 1 9 4 6]\n"
]
}
],
"source": [
"z =[9, 4, 6]\n",
"print(np.concatenate([x, y, z]))"
]
},
{
"metadata": {},
"source": [
"It can also be used for two-dimensional arrays:"
]
},
{
"metadata": {},
"outputs": [
{
"ename": "TypeError",
"evalue": "Field elements must be 2- or 3-tuples, got '40'",
"output_type": "error",
"traceback": [
"\
u001b[1;31m------------------------------------------------------------------------
---\u001b[0m",
"\u001b[1;31mTypeError\u001b[0m Traceback
(most recent call last)",
"Cell \u001b[1;32mIn[67], line 1\u001b[0m\n\u001b[1;32m----> 1\u001b[0m
grid \u001b[38;5;241m=\u001b[39m np\u001b[38;5;241m.\u001b[39marray([\
u001b[38;5;241m10\u001b[39m, \u001b[38;5;241m20\u001b[39m, \u001b[38;5;241m30\
u001b[39m],\n\u001b[0;32m 2\u001b[0m [\u001b[38;5;241m40\
u001b[39m, \u001b[38;5;241m50\u001b[39m, \u001b[38;5;241m60\u001b[39m])\n",
"\u001b[1;31mTypeError\u001b[0m: Field elements must be 2- or 3-tuples, got
'40'"
]
}
],
"source": [
"grid = np.array([10, 20, 30],\n",
" [40, 50, 60])"
]
},
{
"execution_count": null,
"metadata": {},
"outputs": [],
"source": [
"# concatenate along the first axis\n",
"np.concatenate([grid, grid])"
]
},
{
"metadata": {},
"outputs": [],
"source": [
"# concatenate along the second axis (zero-indexed)\n",
"np.concatenate([grid, grid], axis=1)"
]
},
{
"metadata": {},
"source": [
"### Splitting of arrays\n",
"\n",
"The opposite of concatenation is splitting, which is implemented by the
functions ``np.split``, ``np.hsplit``, and ``np.vsplit``. For each of these, we
can pass a list of indices giving the split points:"
]
},
{
"metadata": {},
"outputs": [],
"source": [
"x = [1, 2, 3, 4, 4, 3, 2, 1, 9, 6]\n",
"x1, x2, x3, x4 = np.split(x, [3,5,8])\n",
"print(x1, x2, x3, x4)"
]
},
{
"metadata": {},
"source": [
"Notice that *N* split-points, leads to *N + 1* subarrays.\n",
"The related functions ``np.hsplit`` and ``np.vsplit`` are similar:"
]
},
{
"metadata": {},
"outputs": [
{
"data": {
"text/plain": [
"array([[ 0, 1, 2, 3, 4],\n",
" [ 5, 6, 7, 8, 9],\n",
" [10, 11, 12, 13, 14],\n",
" [15, 16, 17, 18, 19],\n",
" [20, 21, 22, 23, 24]])"
]
},
"metadata": {},
}
],
"source": [
"grid = np.arange(25).reshape((5,5))\n",
"grid"
]
},
{
"metadata": {
"scrolled": true
},
"outputs": [
{
"name": "stdout",
"text": [
"[[ 0 1]\n",
" [ 5 6]\n",
" [10 11]\n",
" [15 16]\n",
" [20 21]]\n",
"[[ 2 3 4]\n",
" [ 7 8 9]\n",
" [12 13 14]\n",
" [17 18 19]\n",
" [22 23 24]]\n"
]
}
],
"source": [
"left, right = np.hsplit(grid, [2])\n",
"print(left)\n",
"print(right)\n"
]
},
{
"metadata": {},
"source": [
"### Aggregates\n",
"\n",
"For binary ufuncs, there are some interesting aggregates that can be computed
directly from the object.\n",
"For example, if we'd like to *reduce* an array with a particular operation, we
can use the ``reduce`` method of any ufunc.\n",
"A reduce repeatedly applies a given operation to the elements of an array
until only a single result remains.\n",
"\n",
"For example, calling ``reduce`` on the `àdd`` ufunc returns the sum of all
elements in the array:"
]
},
{
"metadata": {},
"outputs": [
{
"name": "stdout",
"text": [
"[1]\n",
"[4 3 2 1]\n"
]
}
],
"source": [
"x = np.arange(1,2)\n",
"print(x)\n",
"print(y)"
]
},
{
"metadata": {},
"source": [
"Similarly, calling ``reduce`` on the ``multiply`` ufunc results in the product
of all array elements:"
]
},
{
"metadata": {},
"outputs": [
{
"data": {
"text/plain": [
"1"
]
},
"metadata": {},
}
],
"source": [
"np.multiply.reduce(x)"
]
},
{
"metadata": {},
"source": [
"If we'd like to store all the intermediate results of the computation, we can
instead use `àccumulate``:"
]
},
{
"metadata": {},
"outputs": [
{
"data": {
"text/plain": [
"array([1])"
]
},
"metadata": {},
}
],
"source": [
"np.add.accumulate(x)"
]
},
{
"metadata": {},
"outputs": [
{
"data": {
"text/plain": [
"array([1])"
]
},
"metadata": {},
}
],
"source": [
"np.multiply.accumulate(x)"
]
},
{
"metadata": {},
"outputs": [],
"source": []
}
],
"metadata": {
"kernelspec": {
"display_name": "Python 3 (ipykernel)",
"language": "python",
"name": "python3"
},
"language_info": {
"codemirror_mode": {
"name": "ipython",
"version": 3
},
"file_extension": ".py",
"mimetype": "text/x-python",
"name": "python",
"nbconvert_exporter": "python",
"pygments_lexer": "ipython3",
"version": "3.11.4"
}
},
"nbformat": 4,
"nbformat_minor": 2
}

Numpy Ipynb

Uploaded by

Numpy Ipynb

Uploaded by

{

You might also like