Data Structures

This section describes the data structures of the KML_FFT library functions.

Structures of complex numbers:

• Double-precision complex numbers:

  typedef struct { 
      double r; 
      double i; 
  } kml_fft_complex;

• Single-precision complex numbers:

  typedef struct { 
      float r; 
      float i; 
  } kml_fftf_complex;

kml_fft_plan_s structures and definition of kml_fft_plan:

• Double precision:

  struct kml_fft_plan_s { 
      problem_s problem; 
      solver_s solver; 
      KML_UINT num_threads; 
  }; 
  typedef struct kml_fft_plan_s *kml_fft_plan;

• Single precision:

  struct kml_fftf_plan_s { 
      problem_s problem; 
      solver_s solver; 
      KML_UINT num_threads; 
  }; 
  typedef struct kml_fftf_plan_s *kml_fftf_plan;

• problem_s is the structure of the FFT problem, including the transform type and input/output parameters.

• solver_s is the parameter of selecting a solver for FFT, including the functions used for the solving and the parameters such as kml_fft_plan_s for converting the problem into a subproblem.

• num_threads indicates the number of threads.

Half-complex form:

If n0xn1 array A meets , where A(n0, j) is A(0, j) and A(i, n1) is A(i, 0), a half-complex n0x[(n1/2)+1] array B is used to store array A.