I have a binary matrix which I create by NumPy. The matrix has 5 rows and 32 columns.
array([[1, 1, ..., 1, 1],
[0, 1, ..., 0, 1],
[1, 1, ..., 0, 1],
[0, 0, ..., 1, 0],
[1, 1, ..., 0, 1]])
I convert a matrix rows into a string, and next to integer.
str = ''.join(map(str,array[0])).replace(' ','')
int(str, base=2)
How can I convert the string into the float (float32 - IEEE-754 single)?
Using struct.pack, struct.unpack:
>>> a = [0, 0, 1, 1, 1, 1, 1, 1, 1, 0, 0, 0, 1, 1, 0, 0,
... 1, 1, 0, 0, 1, 1, 0, 0, 1, 1, 0, 0, 1, 1, 0, 1]
>>> i = int(''.join(map(str, a)), 2)
>>> import struct
>>> struct.unpack('f', struct.pack('I', i))[0]
1.100000023841858
import struct
matrix = array(...)
st_i = struct.Struct('I')
st_f = struct.Struct('f')
float_values = [
st_f.unpack(st_i.pack(int(''.join(map(str, a)), 2)))
for a in matrix
]
NOTE: According to the byteorder of the array, you need to prepend <, > before the structure format.
BTW, overwriting str is not a good idea. You cannot use str function/type after the assignment.
It is kind of convoluted, but you can get the same result as your original code in a single liner as:
In [61]: a = np.random.randint(2, size=(5, 32))
In [62]: for x in a:
....: x_t = ''.join(map(str, x))
....: print x_t, int(x_t, 2)
....:
11111111100000111010110110101100 4286819756
01001000110000111001000100011110 1220776222
10101111100100010000111010100111 2945519271
01101111101100011111101001100110 1873934950
11001000110101000111010100000011 3369366787
In [63]: np.packbits(a.reshape(-1, 8)).reshape(-1, 4)[:, ::-1].copy().view(np.uint32)
Out[63]:
array([[4286819756],
[1220776222],
[2945519271],
[1873934950],
[3369366787]], dtype=uint32)
