Memory layout
This lesson explains the memory layout using NumPy.
We'll cover the following...
The NumPy documentation defines the ndarray class very clearly:
An instance of class ndarray consists of a contiguous one-dimensional segment of computer memory (owned by the array, or by some other object), combined with an indexing scheme that maps N integers into the location of an item in the block.
Said differently, an array is mostly a contiguous block of memory whose parts can be accessed using an indexing scheme. Such indexing scheme is in turn defined by a shape and a data type and this is precisely what is needed when you define a new array:
import numpy as npZ = np.arange(9).reshape(3,3).astype(np.int16)print(Z)print(Z.itemsize)# returns size of Z in bytesprint(Z.shape)# returns the x dimension and y dimension of Zprint(Z.ndim)# dimension in Z i.e (2 in this case) since the array is 2D
Here, we know that itemsize is 2 bytes (int16), the shape is (3,3) and
the number of dimensions is 2.
To calculate the dimension we can also use
len(Z.shape).
Furthermore, we can deduce the strides of the array that define the number of bytes to step in each dimension when traversing the array.
import numpy as npZ = np.arange(9).reshape(3,3).astype(np.int16)stride = Z.shape[1]*Z.itemsize, Z.itemsize # store stride of Zprint("Stride(as np.int16):",stride)print("Z.stride(np.16):",Z.strides)Z = np.arange(9).reshape(3,3).astype(np.int32)stride= Z.shape[1]*Z.itemsize, Z.itemsize #stores stride of Zprint("Stride(as np.int32):",stride)print("Z.stride(np.32):",Z.strides)
Here in this example, we have to skip 2 bytes (1 value) to move to the next column, but 6 bytes (3 values) to get to the same position in the next row. As such, the strides for the array Z will be (6, 2).
With all this information, we know how to access a specific item (designed by an index tuple) and more precisely, how to compute the start and end ...