MATLAB: Atan2(0, -55) is different from atan2(-1*0, -55)

atan2negative zeroprecision

Just found that
    >> atan2(0, -55) 
    ans =
        3.1416
  >> atan2(-1*0, -55)
ans =
     -3.1416
I don't understand how is this possible? -1 *0 doesn't equal to 0?? How can I convert such behaviour to C++?

Best Answer

Numerically they are equal.
z = 0;
mz = -0;
z == mz
However, they are not identical.
format hex
z
mz
Note that the sign bits are different. You can see this:
format
plusInf = 1/z
minusInf = 1/mz
As for how you can translate this to C++ ... why not just call std::atan2?
 If the implementation supports IEEE floating-point arithmetic (IEC 60559),
 ...
    If y is ±0 and x is negative or -0, ±π is returned
This is specifically called out in section 9.2.1 of IEEE 754-2008.
Related Question