I ran the following code in a jupyter notebook cell

ndarrs=np.array(["1.2","1.5","1.6"], dtype=np.string_)


It returned |S3 as shown below.

Can someone help me understand the meaning of this symbol?



1 Answer 1


In Python 3, you shouldn't really specify the np.string_ dtype, as it is left there for backwards compatibility with Python 2. The S type you see using np.dtype is a map to the bytes_ type, a zero-terminated string buffer, which shouldn't be used.

The S just means string and the number gives the number of bytes.

In [1]: s = ""        # start with an empty string                                                                                                

In [2]: for i in range(5):    # make the string larger
            s += str(i) 
            a = np.array([s], dtype=np.string_) 
    [b'0']  |S1
    [b'01'] |S2
    [b'012']    |S3
    [b'0123']   |S4
    [b'01234']  |S5
    [b'012345'] |S6

For python 3 you should instead use np.unicode:

In [1]: a = np.array(["hello", "world"], dtype=np.unicode)                                                             

In [2]: type(a)                                                                                                       
Out[2]: numpy.ndarray

In [3]: b.dtype                                                                                                       
Out[3]: dtype('<U5')
  • < means little-endian
  • U means a unicode string
  • 5 relates to the number of bytes used to hold the string. If you had really long strings, that number would then increase.

Have a look at this documentation, which states:

Note on string types:

For backward compatibility with Python 2 the S and a typestrings remain zero-terminated bytes and np.string_ continues to map to np.bytes_. To use actual strings in Python 3 use U or np.unicode_.

  • $\begingroup$ Thanks that was really useful !!! $\endgroup$
    – user582621
    Apr 16, 2020 at 12:12

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.