I'm passing a float as dynamic PyObject from Python to C# and this kind of assignment is automagically accepted:

double csharpfloat = PythonFunctionReturningFloat();

But when using this PyObject in other numerical operations, then explicit casting is required:

double csharpfloat = 1/(double)PythonFunctionReturningFloat();
double csharpfloat += (double)PythonFunctionReturningFloat();

Any reason for such inconsistent behavior?