1. Input data is an array of complex double-precision floating-point data, but the FFT core being modeled requires values in single-precision (32-bit) format. The data must therefore be representable in the MATLAB software ‘single’ datatype, even if it is represented in the ‘double’ datatype. The value is explicitly cast to the C++ ‘float’ datatype inside the MEX function.
2. Output data has the correct precision to model the FFT. The double-precision output array contains single-precision values which are representable in the MATLAB software ‘single’ datatype without error.
3. overflow indicates if exponent overflow has occurred during the FFT processing. A value of 0 indicates that overflow did not occur; a value of 1 indicates that exponent overflow did occur.
4. If overflow occurred, the output sample that overflowed is set to ± Infinity, depending on the sign of the internal result.