npDataObject objects
Returns True if all elements evaluate to True.
Refer to numpy.all for full documentation.
See also
Returns True if any of the elements of a evaluate to True.
Refer to numpy.any for full documentation.
See also
Return indices of the maximum values along the given axis.
Refer to numpy.argmax for full documentation.
See also
Return indices of the minimum values along the given axis of a.
Refer to numpy.argmin for detailed documentation.
See also
Returns the indices that would sort this array.
Refer to numpy.argsort for full documentation.
See also
Copy of the array, cast to a specified type.
Parameters :  dtype : str or dtype
order : {‘C’, ‘F’, ‘A’, or ‘K’}, optional
casting : {‘no’, ‘equiv’, ‘safe’, ‘same_kind’, ‘unsafe’}, optional
subok : bool, optional
copy : bool, optional


Raises :  ComplexWarning : :

Examples
>>> x = np.array([1, 2, 2.5])
>>> x
array([ 1. , 2. , 2.5])
>>> x.astype(int)
array([1, 2, 2])
Swap the bytes of the array elements
Toggle between lowendian and bigendian data representation by returning a byteswapped array, optionally swapped inplace.
Parameters :  inplace: bool, optional :


Returns :  out: ndarray :

Examples
>>> A = np.array([1, 256, 8755], dtype=np.int16)
>>> map(hex, A)
['0x1', '0x100', '0x2233']
>>> A.byteswap(True)
array([ 256, 1, 13090], dtype=int16)
>>> map(hex, A)
['0x100', '0x1', '0x3322']
Arrays of strings are not swapped
>>> A = np.array(['ceg', 'fac'])
>>> A.byteswap()
array(['ceg', 'fac'],
dtype='S3')
Use an index array to construct a new array from a set of choices.
Refer to numpy.choose for full documentation.
See also
Return an array whose values are limited to [a_min, a_max].
Refer to numpy.clip for full documentation.
See also
Return selected slices of this array along given axis.
Refer to numpy.compress for full documentation.
See also
Complexconjugate all elements.
Refer to numpy.conjugate for full documentation.
See also
Return the complex conjugate, elementwise.
Refer to numpy.conjugate for full documentation.
See also
Return a copy of the array.
Parameters :  order : {‘C’, ‘F’, ‘A’, ‘K’}, optional


See also
numpy.copy, numpy.copyto
Examples
>>> x = np.array([[1,2,3],[4,5,6]], order='F')
>>> y = x.copy()
>>> x.fill(0)
>>> x
array([[0, 0, 0],
[0, 0, 0]])
>>> y
array([[1, 2, 3],
[4, 5, 6]])
>>> y.flags['C_CONTIGUOUS']
True
Return the cumulative product of the elements along the given axis.
Refer to numpy.cumprod for full documentation.
See also
Return the cumulative sum of the elements along the given axis.
Refer to numpy.cumsum for full documentation.
See also
Return specified diagonals.
Refer to numpy.diagonal() for full documentation.
See also
Dot product of two arrays.
Refer to numpy.dot for full documentation.
See also
Examples
>>> a = np.eye(2)
>>> b = np.ones((2, 2)) * 2
>>> a.dot(b)
array([[ 2., 2.],
[ 2., 2.]])
This array method can be conveniently chained:
>>> a.dot(b).dot(b)
array([[ 8., 8.],
[ 8., 8.]])
Dump a pickle of the array to the specified file. The array can be read back with pickle.load or numpy.load.
Parameters :  file : str


Returns the pickle of the array as a string. pickle.loads or numpy.loads will convert the string back to an array.
Parameters :  None : 

Fill the array with a scalar value.
Parameters :  value : scalar


Examples
>>> a = np.array([1, 2])
>>> a.fill(0)
>>> a
array([0, 0])
>>> a = np.empty(2)
>>> a.fill(1)
>>> a
array([ 1., 1.])
Return a copy of the array collapsed into one dimension.
Parameters :  order : {‘C’, ‘F’, ‘A’}, optional


Returns :  y : ndarray

Examples
>>> a = np.array([[1,2], [3,4]])
>>> a.flatten()
array([1, 2, 3, 4])
>>> a.flatten('F')
array([1, 3, 2, 4])
Returns a field of the given array as a certain type.
A field is a view of the array data with a given datatype. The values in the view are determined by the given type and the offset into the current array in bytes. The offset needs to be such that the view dtype fits in the array dtype; for example an array of dtype complex128 has 16byte elements. If taking a view with a 32bit integer (4 bytes), the offset needs to be between 0 and 12 bytes.
Parameters :  dtype : str or dtype
offset : int


Examples
>>> x = np.diag([1.+1.j]*2)
>>> x[1, 1] = 2 + 4.j
>>> x
array([[ 1.+1.j, 0.+0.j],
[ 0.+0.j, 2.+4.j]])
>>> x.getfield(np.float64)
array([[ 1., 0.],
[ 0., 2.]])
By choosing an offset of 8 bytes we can select the complex part of the array for our view:
>>> x.getfield(np.float64, offset=8)
array([[ 1., 0.],
[ 0., 4.]])
Copy an element of an array to a standard Python scalar and return it.
Parameters :  *args : Arguments (variable number and type)


Returns :  z : Standard Python scalar object

Notes
When the data type of a is longdouble or clongdouble, item() returns a scalar array object because there is no available Python scalar that would not lose information. Void arrays return a buffer object for item(), unless fields are defined, in which case a tuple is returned.
item is very similar to a[args], except, instead of an array scalar, a standard Python scalar is returned. This can be useful for speeding up access to elements of the array and doing arithmetic on elements of the array using Python’s optimized math.
Examples
>>> x = np.random.randint(9, size=(3, 3))
>>> x
array([[3, 1, 7],
[2, 8, 3],
[8, 5, 3]])
>>> x.item(3)
2
>>> x.item(7)
5
>>> x.item((0, 1))
1
>>> x.item((2, 2))
3
Insert scalar into an array (scalar is cast to array’s dtype, if possible)
There must be at least 1 argument, and define the last argument as item. Then, a.itemset(*args) is equivalent to but faster than a[args] = item. The item should be a scalar value and args must select a single item in the array a.
Parameters :  *args : Arguments


Notes
Compared to indexing syntax, itemset provides some speed increase for placing a scalar into a particular location in an ndarray, if you must do this. However, generally this is discouraged: among other problems, it complicates the appearance of the code. Also, when using itemset (and item) inside a loop, be sure to assign the methods to a local variable to avoid the attribute lookup at each loop iteration.
Examples
>>> x = np.random.randint(9, size=(3, 3))
>>> x
array([[3, 1, 7],
[2, 8, 3],
[8, 5, 3]])
>>> x.itemset(4, 0)
>>> x.itemset((2, 2), 9)
>>> x
array([[3, 1, 7],
[2, 0, 3],
[8, 5, 9]])
Return the maximum along a given axis.
Refer to numpy.amax for full documentation.
See also
Returns the average of the array elements along given axis.
Refer to numpy.mean for full documentation.
See also
Return the minimum along a given axis.
Refer to numpy.amin for full documentation.
See also
Return the array with the same data viewed with a different byte order.
Equivalent to:
arr.view(arr.dtype.newbytorder(new_order))
Changes are also made in all fields and subarrays of the array data type.
Parameters :  new_order : string, optional


Returns :  new_arr : array

Return the indices of the elements that are nonzero.
Refer to numpy.nonzero for full documentation.
See also
Return the product of the array elements over the given axis
Refer to numpy.prod for full documentation.
See also
Peak to peak (maximum  minimum) value along a given axis.
Refer to numpy.ptp for full documentation.
See also
Set a.flat[n] = values[n] for all n in indices.
Refer to numpy.put for full documentation.
See also
Return a flattened array.
Refer to numpy.ravel for full documentation.
See also
Repeat elements of an array.
Refer to numpy.repeat for full documentation.
See also
Returns an array containing the same data with a new shape.
Refer to numpy.reshape for full documentation.
See also
Change shape and size of array inplace.
Parameters :  new_shape : tuple of ints, or n ints
refcheck : bool, optional


Returns :  None : 
Raises :  ValueError :
SystemError :

See also
Notes
This reallocates space for the data area if necessary.
Only contiguous arrays (data elements consecutive in memory) can be resized.
The purpose of the reference count check is to make sure you do not use this array as a buffer for another Python object and then reallocate the memory. However, reference counts can increase in other ways so if you are sure that you have not shared the memory for this array with another Python object, then you may safely set refcheck to False.
Examples
Shrinking an array: array is flattened (in the order that the data are stored in memory), resized, and reshaped:
>>> a = np.array([[0, 1], [2, 3]], order='C')
>>> a.resize((2, 1))
>>> a
array([[0],
[1]])
>>> a = np.array([[0, 1], [2, 3]], order='F')
>>> a.resize((2, 1))
>>> a
array([[0],
[2]])
Enlarging an array: as above, but missing entries are filled with zeros:
>>> b = np.array([[0, 1], [2, 3]])
>>> b.resize(2, 3) # new_shape parameter doesn't have to be a tuple
>>> b
array([[0, 1, 2],
[3, 0, 0]])
Referencing an array prevents resizing...
>>> c = a
>>> a.resize((1, 1))
Traceback (most recent call last):
...
ValueError: cannot resize an array that has been referenced ...
Unless refcheck is False:
>>> a.resize((1, 1), refcheck=False)
>>> a
array([[0]])
>>> c
array([[0]])
Return a with each element rounded to the given number of decimals.
Refer to numpy.around for full documentation.
See also
Find indices where elements of v should be inserted in a to maintain order.
For full documentation, see numpy.searchsorted
See also
Put a value into a specified place in a field defined by a datatype.
Place val into a‘s field defined by dtype and beginning offset bytes into the field.
Parameters :  val : object
dtype : dtype object
offset : int, optional


Returns :  None : 
See also
Examples
>>> x = np.eye(3)
>>> x.getfield(np.float64)
array([[ 1., 0., 0.],
[ 0., 1., 0.],
[ 0., 0., 1.]])
>>> x.setfield(3, np.int32)
>>> x.getfield(np.int32)
array([[3, 3, 3],
[3, 3, 3],
[3, 3, 3]])
>>> x
array([[ 1.00000000e+000, 1.48219694e323, 1.48219694e323],
[ 1.48219694e323, 1.00000000e+000, 1.48219694e323],
[ 1.48219694e323, 1.48219694e323, 1.00000000e+000]])
>>> x.setfield(np.eye(3), np.int32)
>>> x
array([[ 1., 0., 0.],
[ 0., 1., 0.],
[ 0., 0., 1.]])
Set array flags WRITEABLE, ALIGNED, and UPDATEIFCOPY, respectively.
These Booleanvalued flags affect how numpy interprets the memory area used by a (see Notes below). The ALIGNED flag can only be set to True if the data is actually aligned according to the type. The UPDATEIFCOPY flag can never be set to True. The flag WRITEABLE can only be set to True if the array owns its own memory, or the ultimate owner of the memory exposes a writeable buffer interface, or is a string. (The exception for string is made so that unpickling can be done without copying memory.)
Parameters :  write : bool, optional
align : bool, optional
uic : bool, optional


Notes
Array flags provide information about how the memory area used for the array is to be interpreted. There are 6 Boolean flags in use, only three of which can be changed by the user: UPDATEIFCOPY, WRITEABLE, and ALIGNED.
WRITEABLE (W) the data area can be written to;
ALIGNED (A) the data and strides are aligned appropriately for the hardware (as determined by the compiler);
UPDATEIFCOPY (U) this array is a copy of some other array (referenced by .base). When this array is deallocated, the base array will be updated with the contents of this array.
All flags can be accessed using their first (upper case) letter as well as the full name.
Examples
>>> y
array([[3, 1, 7],
[2, 0, 0],
[8, 5, 9]])
>>> y.flags
C_CONTIGUOUS : True
F_CONTIGUOUS : False
OWNDATA : True
WRITEABLE : True
ALIGNED : True
UPDATEIFCOPY : False
>>> y.setflags(write=0, align=0)
>>> y.flags
C_CONTIGUOUS : True
F_CONTIGUOUS : False
OWNDATA : True
WRITEABLE : False
ALIGNED : False
UPDATEIFCOPY : False
>>> y.setflags(uic=1)
Traceback (most recent call last):
File "<stdin>", line 1, in <module>
ValueError: cannot set UPDATEIFCOPY flag to True
Sort an array, inplace.
Parameters :  axis : int, optional
kind : {‘quicksort’, ‘mergesort’, ‘heapsort’}, optional
order : list, optional


See also
Notes
See sort for notes on the different sorting algorithms.
Examples
>>> a = np.array([[1,4], [3,1]])
>>> a.sort(axis=1)
>>> a
array([[1, 4],
[1, 3]])
>>> a.sort(axis=0)
>>> a
array([[1, 3],
[1, 4]])
Use the order keyword to specify a field to use when sorting a structured array:
>>> a = np.array([('a', 2), ('c', 1)], dtype=[('x', 'S1'), ('y', int)])
>>> a.sort(order='y')
>>> a
array([('c', 1), ('a', 2)],
dtype=[('x', 'S1'), ('y', '<i4')])
Remove singledimensional entries from the shape of a.
Refer to numpy.squeeze for full documentation.
See also
Returns the standard deviation of the array elements along given axis.
Refer to numpy.std for full documentation.
See also
Return the sum of the array elements over the given axis.
Refer to numpy.sum for full documentation.
See also
Return a view of the array with axis1 and axis2 interchanged.
Refer to numpy.swapaxes for full documentation.
See also
Return an array formed from the elements of a at the given indices.
Refer to numpy.take for full documentation.
See also
Write array to a file as text or binary (default).
Data is always written in ‘C’ order, independent of the order of a. The data produced by this method can be recovered using the function fromfile().
Parameters :  fid : file or str
sep : str
format : str


Notes
This is a convenience function for quick storage of array data. Information on endianness and precision is lost, so this method is not a good choice for files intended to archive data or transport data between machines with different endianness. Some of these problems can be overcome by outputting the data as text files, at the expense of speed and file size.
Return the array as a (possibly nested) list.
Return a copy of the array data as a (nested) Python list. Data items are converted to the nearest compatible Python type.
Parameters :  none : 

Returns :  y : list

Notes
The array may be recreated, a = np.array(a.tolist()).
Examples
>>> a = np.array([1, 2])
>>> a.tolist()
[1, 2]
>>> a = np.array([[1, 2], [3, 4]])
>>> list(a)
[array([1, 2]), array([3, 4])]
>>> a.tolist()
[[1, 2], [3, 4]]
Construct a Python string containing the raw data bytes in the array.
Constructs a Python string showing a copy of the raw contents of data memory. The string can be produced in either ‘C’ or ‘Fortran’, or ‘Any’ order (the default is ‘C’order). ‘Any’ order means Corder unless the F_CONTIGUOUS flag in the array is set, in which case it means ‘Fortran’ order.
Parameters :  order : {‘C’, ‘F’, None}, optional


Returns :  s : str

Examples
>>> x = np.array([[0, 1], [2, 3]])
>>> x.tostring()
'\x00\x00\x00\x00\x01\x00\x00\x00\x02\x00\x00\x00\x03\x00\x00\x00'
>>> x.tostring('C') == x.tostring()
True
>>> x.tostring('F')
'\x00\x00\x00\x00\x02\x00\x00\x00\x01\x00\x00\x00\x03\x00\x00\x00'
Return the sum along diagonals of the array.
Refer to numpy.trace for full documentation.
See also
Returns a view of the array with axes transposed.
For a 1D array, this has no effect. (To change between column and row vectors, first cast the 1D array into a matrix object.) For a 2D array, this is the usual matrix transpose. For an nD array, if axes are given, their order indicates how the axes are permuted (see Examples). If axes are not provided and a.shape = (i[0], i[1], ... i[n2], i[n1]), then a.transpose().shape = (i[n1], i[n2], ... i[1], i[0]).
Parameters :  axes : None, tuple of ints, or n ints


Returns :  out : ndarray

See also
Examples
>>> a = np.array([[1, 2], [3, 4]])
>>> a
array([[1, 2],
[3, 4]])
>>> a.transpose()
array([[1, 3],
[2, 4]])
>>> a.transpose((1, 0))
array([[1, 3],
[2, 4]])
>>> a.transpose(1, 0)
array([[1, 3],
[2, 4]])
Returns the variance of the array elements, along given axis.
Refer to numpy.var for full documentation.
See also
New view of array with the same data.
Parameters :  dtype : datatype, optional
type : Python type, optional


Notes
a.view() is used two different ways:
a.view(some_dtype) or a.view(dtype=some_dtype) constructs a view of the array’s memory with a different datatype. This can cause a reinterpretation of the bytes of memory.
a.view(ndarray_subclass) or a.view(type=ndarray_subclass) just returns an instance of ndarray_subclass that looks at the same array (same shape, dtype, etc.) This does not cause a reinterpretation of the memory.
Examples
>>> x = np.array([(1, 2)], dtype=[('a', np.int8), ('b', np.int8)])
Viewing array data using a different type and dtype:
>>> y = x.view(dtype=np.int16, type=np.matrix)
>>> y
matrix([[513]], dtype=int16)
>>> print type(y)
<class 'numpy.matrixlib.defmatrix.matrix'>
Creating a view on a structured array so it can be used in calculations
>>> x = np.array([(1, 2),(3,4)], dtype=[('a', np.int8), ('b', np.int8)])
>>> xv = x.view(dtype=np.int8).reshape(1,2)
>>> xv
array([[1, 2],
[3, 4]], dtype=int8)
>>> xv.mean(0)
array([ 2., 3.])
Making changes to the view changes the underlying array
>>> xv[0,1] = 20
>>> print x
[(1, 20) (3, 4)]
Using a view to convert an array to a record array:
>>> z = x.view(np.recarray)
>>> z.a
array([1], dtype=int8)
Views share data:
>>> x[0] = (9, 10)
>>> z[0]
(9, 10)
Same as self.transpose(), except that self is returned if self.ndim < 2.
Examples
>>> x = np.array([[1.,2.],[3.,4.]])
>>> x
array([[ 1., 2.],
[ 3., 4.]])
>>> x.T
array([[ 1., 3.],
[ 2., 4.]])
>>> x = np.array([1.,2.,3.,4.])
>>> x
array([ 1., 2., 3., 4.])
>>> x.T
array([ 1., 2., 3., 4.])
list with descriptions for each axis.
list with offset values for each axis [px].
list with scale values for each axis [unit/px].
list containing the units for each axis.
Base object if memory is from some other object.
Examples
The base of an array that owns its memory is None:
>>> x = np.array([1,2,3,4])
>>> x.base is None
True
Slicing creates a view, whose memory is shared with x:
>>> y = x[2:]
>>> y.base is x
True
An object to simplify the interaction of the array with the ctypes module.
This attribute creates an object that makes it easier to use arrays when calling shared libraries with the ctypes module. The returned object has, among others, data, shape, and strides attributes (see Notes below) which themselves return ctypes objects that can be used as arguments to a shared library.
Parameters :  None : 

Returns :  c : Python object

See also
numpy.ctypeslib
Notes
Below are the public attributes of this object which were documented in “Guide to NumPy” (we have omitted undocumented public attributes, as well as documented private attributes):
Be careful using the ctypes attribute  especially on temporary arrays or arrays constructed on the fly. For example, calling (a+b).ctypes.data_as(ctypes.c_void_p) returns a pointer to memory that is invalid because the array created as (a+b) is deallocated before the next Python statement. You can avoid this problem using either c=a+b or ct=(a+b).ctypes. In the latter case, ct will hold a reference to the array until ct is deleted or reassigned.
If the ctypes module is not available, then the ctypes attribute of array objects still returns something useful, but ctypes objects are not returned and errors may be raised instead. In particular, the object will still have the as parameter attribute which will return an integer equal to the data attribute.
Examples
>>> import ctypes
>>> x
array([[0, 1],
[2, 3]])
>>> x.ctypes.data
30439712
>>> x.ctypes.data_as(ctypes.POINTER(ctypes.c_long))
<ctypes.LP_c_long object at 0x01F01300>
>>> x.ctypes.data_as(ctypes.POINTER(ctypes.c_long)).contents
c_long(0)
>>> x.ctypes.data_as(ctypes.POINTER(ctypes.c_longlong)).contents
c_longlong(4294967296L)
>>> x.ctypes.shape
<numpy.core._internal.c_long_Array_2 object at 0x01FFD580>
>>> x.ctypes.shape_as(ctypes.c_long)
<numpy.core._internal.c_long_Array_2 object at 0x01FCE620>
>>> x.ctypes.strides
<numpy.core._internal.c_long_Array_2 object at 0x01FCE620>
>>> x.ctypes.strides_as(ctypes.c_longlong)
<numpy.core._internal.c_longlong_Array_2 object at 0x01F01300>
Python buffer object pointing to the start of the array’s data.
Datatype of the array’s elements.
Parameters :  None : 

Returns :  d : numpy dtype object 
See also
numpy.dtype
Examples
>>> x
array([[0, 1],
[2, 3]])
>>> x.dtype
dtype('int32')
>>> type(x.dtype)
<type 'numpy.dtype'>
Information about the memory layout of the array.
Attributes :  C_CONTIGUOUS (C) :
F_CONTIGUOUS (F) :
OWNDATA (O) :
WRITEABLE (W) :
ALIGNED (A) :
UPDATEIFCOPY (U) :
FNC :
FORC :
BEHAVED (B) :
CARRAY (CA) :
FARRAY (FA) :


Notes
The flags object can be accessed dictionarylike (as in a.flags['WRITEABLE']), or by using lowercased attribute names (as in a.flags.writeable). Short flag names are only supported in dictionary access.
Only the UPDATEIFCOPY, WRITEABLE, and ALIGNED flags can be changed by the user, via direct assignment to the attribute or dictionary entry, or by calling ndarray.setflags.
The array flags cannot be set arbitrarily:
A 1D iterator over the array.
This is a numpy.flatiter instance, which acts similarly to, but is not a subclass of, Python’s builtin iterator object.
Examples
>>> x = np.arange(1, 7).reshape(2, 3)
>>> x
array([[1, 2, 3],
[4, 5, 6]])
>>> x.flat[3]
4
>>> x.T
array([[1, 4],
[2, 5],
[3, 6]])
>>> x.T.flat[3]
5
>>> type(x.flat)
<type 'numpy.flatiter'>
An assignment example:
>>> x.flat = 3; x
array([[3, 3, 3],
[3, 3, 3]])
>>> x.flat[[1,4]] = 1; x
array([[3, 1, 3],
[3, 1, 3]])
The imaginary part of the array.
Examples
>>> x = np.sqrt([1+0j, 0+1j])
>>> x.imag
array([ 0. , 0.70710678])
>>> x.imag.dtype
dtype('float64')
Length of one array element in bytes.
Examples
>>> x = np.array([1,2,3], dtype=np.float64)
>>> x.itemsize
8
>>> x = np.array([1,2,3], dtype=np.complex128)
>>> x.itemsize
16
returns new dictionary with all tags inside
Total bytes consumed by the elements of the array.
Notes
Does not include memory consumed by nonelement attributes of the array object.
Examples
>>> x = np.zeros((3,5,2), dtype=np.complex128)
>>> x.nbytes
480
>>> np.prod(x.shape) * x.itemsize
480
Number of array dimensions.
Examples
>>> x = np.array([1, 2, 3])
>>> x.ndim
1
>>> y = np.zeros((2, 3, 4))
>>> y.ndim
3
The real part of the array.
See also
Examples
>>> x = np.sqrt([1+0j, 0+1j])
>>> x.real
array([ 1. , 0.70710678])
>>> x.real.dtype
dtype('float64')
Tuple of array dimensions.
Notes
May be used to “reshape” the array, as long as this would not require a change in the total number of elements
Examples
>>> x = np.array([1, 2, 3, 4])
>>> x.shape
(4,)
>>> y = np.zeros((2, 3, 4))
>>> y.shape
(2, 3, 4)
>>> y.shape = (3, 8)
>>> y
array([[ 0., 0., 0., 0., 0., 0., 0., 0.],
[ 0., 0., 0., 0., 0., 0., 0., 0.],
[ 0., 0., 0., 0., 0., 0., 0., 0.]])
>>> y.shape = (3, 6)
Traceback (most recent call last):
File "<stdin>", line 1, in <module>
ValueError: total size of new array must be unchanged
Number of elements in the array.
Equivalent to np.prod(a.shape), i.e., the product of the array’s dimensions.
Examples
>>> x = np.zeros((3, 5, 2), dtype=np.complex128)
>>> x.size
30
>>> np.prod(x.shape)
30
Tuple of bytes to step in each dimension when traversing an array.
The byte offset of element (i[0], i[1], ..., i[n]) in an array a is:
offset = sum(np.array(i) * a.strides)
A more detailed explanation of strides can be found in the “ndarray.rst” file in the NumPy reference guide.
See also
numpy.lib.stride_tricks.as_strided
Notes
Imagine an array of 32bit integers (each 4 bytes):
x = np.array([[0, 1, 2, 3, 4],
[5, 6, 7, 8, 9]], dtype=np.int32)
This array is stored in memory as 40 bytes, one after the other (known as a contiguous block of memory). The strides of an array tell us how many bytes we have to skip in memory to move to the next position along a certain axis. For example, we have to skip 4 bytes (1 value) to move to the next column, but 20 bytes (5 values) to get to the same position in the next row. As such, the strides for the array x will be (20, 4).
Examples
>>> y = np.reshape(np.arange(2*3*4), (2,3,4))
>>> y
array([[[ 0, 1, 2, 3],
[ 4, 5, 6, 7],
[ 8, 9, 10, 11]],
[[12, 13, 14, 15],
[16, 17, 18, 19],
[20, 21, 22, 23]]])
>>> y.strides
(48, 16, 4)
>>> y[1,1,1]
17
>>> offset=sum(y.strides * np.array((1,1,1)))
>>> offset/y.itemsize
17
>>> x = np.reshape(np.arange(5*6*7*8), (5,6,7,8)).transpose(2,3,1,0)
>>> x.strides
(32, 4, 224, 1344)
>>> i = np.array([3,5,2,2])
>>> offset = sum(i * x.strides)
>>> x[3,5,2,2]
813
>>> offset / x.itemsize
813
tag dictionary for this data object.
value unit description.
value offset [default: 0.0].
value scale [default: 1.0].
value unit string.