|, > and < in numpy datatype

Question

This could be a very silly question but I tried to google keywords like less and greater signs in data type of numpy and found no reference.

In the doc of numpy,

x = np.array([(1.0, 2), (3.0, 4)], dtype=[('x', float), ('y', int)])

outputs

array([(1.0, 2), (3.0, 4)],
      dtype=[('x', '<f8'), ('y', '<i4')])

But on my PC, the output is

array([(1.0, 2), (3.0, 4)],
      dtype=[('x', '>f8'), ('y', '>i4')])

What do < and > in the dtype mean and why there is the difference?

Answer 1

The keywords < and > stand for byte ordering, aka endianness. It is the order in which bytes from numbers are stored (when numbers are compossed of more than 1 byte, e.g. int16, int32, float32...). This page from the reference gives you all the information you need about it in numpy, but as a summary:

• | : it doesn't have a byte order because is redundant (on single byte numbers or strings)

• < : little-endian

• > : big-endian

As @tobias_k and @RobertKern pointed out, the default endianess, if not specified, is system dependant.