Is there a simple C ++ example on how to use Intel MKL FFT?

Question

I need to perform FFT and Inverse-FFT conversions. The input will be a vector and double matrices. Ideally, the output should be an array of std :: complex, but I can live with double _Complex.

I did not find a simple example, all Intel examples do a lot of things at once without enough comments.

I just want a simple C ++ example to take a double input vector (or matrix) and output the result converted by FFT (ideally using std :: complex).

Answer 1

I ended up testing a few things, and I finally ended up with these three functions that do what I want and that I looked at simple examples.

I tested it against some materials and I had good results. However, I did not conduct rigorous testing.

//Note after each operation status should be 0 on success std::vector<std::complex<float>> fft_complex(std::vector<std::complex<float>>& in){ std::vector<std::complex<float>> out(in.size()); DFTI_DESCRIPTOR_HANDLE descriptor; MKL_LONG status; status = DftiCreateDescriptor(&descriptor, DFTI_SINGLE, DFTI_COMPLEX, 1, in.size()); //Specify size and precision status = DftiSetValue(descriptor, DFTI_PLACEMENT, DFTI_NOT_INPLACE); //Out of place FFT status = DftiCommitDescriptor(descriptor); //Finalize the descriptor status = DftiComputeForward(descriptor, in.data(), out.data()); //Compute the Forward FFT status = DftiFreeDescriptor(&descriptor); //Free the descriptor return out; } std::vector<std::complex<float>> fft_real(std::vector<float>& in_real){ std::vector<std::complex<float>> in(in_real.size()); std::copy(in_real.begin(), in_real.end(), in.begin()); return fft_complex(in); } std::vector<float> ifft(std::vector<std::complex<float>>& in){ std::vector<std::complex<float>> out(in.size()); DFTI_DESCRIPTOR_HANDLE descriptor; MKL_LONG status; status = DftiCreateDescriptor(&descriptor, DFTI_SINGLE, DFTI_COMPLEX, 1, in.size()); //Specify size and precision status = DftiSetValue(descriptor, DFTI_PLACEMENT, DFTI_NOT_INPLACE); //Out of place FFT status = DftiSetValue(descriptor, DFTI_BACKWARD_SCALE, 1.0f / in.size()); //Scale down the output status = DftiCommitDescriptor(descriptor); //Finalize the descriptor status = DftiComputeBackward(descriptor, in.data(), out.data()); //Compute the Forward FFT status = DftiFreeDescriptor(&descriptor); //Free the descriptor std::vector<float> output(out.size()); for(std::size_t i = 0; i < out.size(); ++i){ output[i] = out[i].real(); } return output; } 

Answer 2

Try this Developer Guide for Intel® Math Kernel Library - C from INTEL. There are examples of using FFT functions with Intel MKL along with OpenMP.