Why bitwise left shift in Python and C# has different values?
Python:
>>> 2466250752<<1
4932501504L
C#:
System.Console.Write((2466250752 << 1).ToString()); // output is 637534208
-
If you want the similar behavior in C# you'll need to use BigInteger (for any length) or Long (for this case)Martheen– Martheen2013-03-25 03:14:39 +00:00Commented Mar 25, 2013 at 3:14
-
1I need similar behavior in Python :)user2160982– user21609822013-03-25 03:17:24 +00:00Commented Mar 25, 2013 at 3:17
-
Oh, just AND the value right after shifting right with 2^32-1Martheen– Martheen2013-03-25 03:21:07 +00:00Commented Mar 25, 2013 at 3:21
Add a comment
|
2 Answers
You are overflowing the 32-bit (unsigned) integer in C#.
In python, all integers are arbitrarily-sized. That means the integer will expand to whatever size is required. Note that I added the underscores:
>>> a = 2466250752
>>>
>>> hex(a)
'0x9300_0000L'
>>>
>>> hex(a << 1)
'0x1_2600_0000L'
^-------------- Note the additional place
In C#, uint is only 32-bits. When you shift left, you are exceeding the size of the integer, causing overflow.
Before shifting:
After shifting:
Notice that a does not have the leading 1 that python showed.
To get around this limitation for this case, you can use a ulong which is 64 bits instead of 32 bits. This will work for values up to 264-1.
answered Mar 25, 2013 at 3:10
Jonathon Reinhart
138k38 gold badges266 silver badges347 bronze badges
1 Comment
Jonathon Reinhart
@SnakesandCoffee Ha, oops. Fixed.
Python makes sure your integers don't overflow, while C# allows for overflow (but throws an exception on overflow in a checked context). In practice this means you can treat Python integers as having infinite width, while a C# int or uint is always 4 bytes.
Notice in your Python example that the value "4932501504L" has a trailing L, which means long integer. Python automatically performs math in long (size-of-available-memory-width, unlike C#'s long, which is 8 bytes) integers when overflow would occur in int values. You can see the rationale behind this idea in PEP 237.
EDIT: To get the Python result in C#, you cannot use a plain int or long - those types have limited size. One type whose size is limited only by memory is BigInteger. It will be slower than int or long for arithmetic, so I wouldn't recommend using it on every application, but it can come in handy.
As an example, you can write almost the same code as in C#, with the same result as in Python:
Console.WriteLine(new BigInteger(2466250752) << 1);
// output is 4932501504
This works for arbitrary shift sizes. For instance, you can write
Console.WriteLine(new BigInteger(2466250752) << 1000);
// output is 26426089082476043843620786304598663584184261590451906619194221930186703343408641580508146166393907795104656740341094823575842096015719243506448572304002696283531880333455226335616426281383175835559603193956495848019208150304342043576665227249501603863012525070634185841272245152956518296810797380454760948170752
Of course, this would overflow a long.
3 Comments
user2160982
Thanks, but how I can get on Python the result as on C#?
Adam Mihalcin
@user2160982 Oops, I misread your comment. Martheen comments correctly on the question: you should take a bitwise and with 2^32 - 1.
Martheen
Aah, 'bitwise and', so that's the correct way to describe it. All this time I used AND like some screamer