bitwise_and with uint64 #2955

Activity

1. charris commented on Feb 1, 2013

   

   charris

   on Feb 1, 2013

   Member

   More actions

   Looks like 0xffff is treated as signed and the ufunc machinery is looking for a larger signed type the will hold uint64. This isn't easy to fix without a major rework of some of numpy's internals. There is a work around

    In [2]: np.uint64(123) & np.uint64(0xffff)
    Out[2]: 123
   

   I think this should work also, so perhaps there is something that needs a fix.

   In [4]: np.bitwise_and(np.uint64(123), 0xffff, casting='unsafe')
   ---------------------------------------------------------------------------
   TypeError                                 Traceback (most recent call last)
   /home/charris/<ipython-input-4-33ccdd6698ae> in <module>()
   ----> 1 np.bitwise_and(np.uint64(123), 0xffff, casting='unsafe')
   
   TypeError: ufunc 'bitwise_and' not supported for the input types, and the
   inputs could not be safely coerced to any supported types according to the casting rule ''safe''
   

   👍React with 👍2Reacted by ETBox and djpeaco
2. jrmlhermitte commented on Nov 15, 2017

   

   jrmlhermitte

   on Nov 15, 2017

   Contributor

   More actions

   Same problem. Broadcasting to array also works

   import numpy as np
   a = np.ones(2, dtype=np.uint64)
   # this works
   a&1
   # but this
   a[0]&1
   # gives this error
   #---------------------------------------------------------------------------
   #TypeError                                 Traceback (most recent call last)
   #<ipython-input-43-e0a82b0140e0> in <module>()
   #----> 1 a[0]&1
   #
   #TypeError: ufunc 'bitwise_and' not supported for the input types, and the
   #inputs could not be safely coerced to any supported types according to the
   #casting rule ''safe''
   #
   
   # but this works
   a = np.ones(2, dtype=np.uint32)
   
   a[0]&1

   workaround also works for me.

   Is there a PR for this yet?
3. seberg commented on Nov 19, 2022

   

   seberg

   on Nov 19, 2022

   Member

   More actions

   Closing as duplicate of gh-22624
4. seberg

   closed this as completedon Nov 19, 2022