I need to convert B to a 1-D array I have a matrix which I am trying to turn to a 1-D array. However I am getting an array with 1 cell which holds the 1-D array. How do I fix it ?
Python CODE:
import numpy as np
def computecoreset():
a = np.matrix('1 2 19 22; 3 4 28 11')
B = a.ravel()
cal = np.random.choice(B, 3)
return cal
print(computecoreset())
However:
B = [[ 1 2 19 22 3 4 28 11]]
Which is not the same as
[ 1 2 19 22 3 4 28 11]
Thanks in advance
a.ravel() being a method of NumPy matrix would still keep it as a matrix and NumPy matrices can't be 1D. So, instead we can use NumPy's ravel() method to convert into a flattened 1D NumPy array, like so -
np.ravel(a)
Sample run -
In [40]: a
Out[40]:
matrix([[ 1, 2, 19, 22],
[ 3, 4, 28, 11]])
In [41]: np.ravel(a)
Out[41]: array([ 1, 2, 19, 22, 3, 4, 28, 11])
Use np.array to define your 2D arrays.
The extra convenience of specifying data as strings, as in your example, and having a few methods proper to 2D arrays attached to a matrix object are, at least in my opinion, outbalanced by the minor flexibility of the latter data structure, in particular most expressions that involve a matrix lead to a matrix result, even if this is not what you're looking for.
E.g., the double matrix product, i.e., a scalar, when it involves matrix objects is a 2D matrix and you will see the following recurring idiom
x = np.matrix('1;2;3')
K = np.matrix('1 2 3; 2 4 5; 3 5 6')
V = 0.5 * (x.T*K*x)[0,0]
when you want the strain energy of an elastic system represented as free-standing floating point value.
Note also that, using arrays, you have no more to write
V = 0.5*np.dot(x.T, np.dot(K, x))
but you can use the new syntax
V = 0.5 * x.T@K@x
That said, if you still want a matrix, the answer by Divakar is exactly what you want and you're right in accepting it.
