search

SciSharp / NumSharp

High Performance Computation for N-D Tensors in .NET, similar API to NumPy.
https://github.com/SciSharp
Apache License 2.0
1.34k stars 188 forks source link

arctan2() returning incorrect value #426

Open RoseberryPi opened 3 years ago

RoseberryPi commented 3 years ago

So I'm using numsharp np.arctan2(np.array(-0.0012562886517319706), np.array(-0.7499033624114052))

I get the wrong value, approximately, -4.33e-5

using numpy.net and C#'s Math library np.arctan2(np.array(-0.0012562886517319706), np.array(-0.7499033624114052)) Math.Atan2(-0.0012562886517319706, -0.7499033624114052)

I get -3.1399, which is the correct value.. Am I just being dumb and doing something wrong or is NumSharp not actually calculating the correct value?

furthmore, np.arctan2(1,1) is 90deg according to numsharp. Should be 45. np.arctan2(1,-1) is also 90deg....

I'm using version v0.20.5

ghost commented 3 years ago

Hello @RoseberryPi,

You are right, looking at code I don't understand the type casting done to (byte) instead of (double). There must be an explanation.

case NPTypeCode.Double:
{
   var out_addr = (double*)out_y.Address;
   var out_addr_x = (byte*)out_x.Address;
   Parallel.For(0, len, i => *(out_addr)  = Converts.ToDouble(Math.Atan2(*(out_addr) + i, *(out_addr_x) + i)));
   return out_y;
}

I tried casting to (double*) and it returns the expected value, 3.13991738776297