ifft2.cpp

c++的数学物理方程数值算法源程序。这是"Numerical Methods for Physics"第二版的源程序。

#include "NumMeth.h"

void fft( Matrix& RealA, Matrix& ImagA);

void ifft2( Matrix& RealA, Matrix& ImagA) {
// Routine to compute inverse two dimensional Fourier transform 
// using FFT algorithm
// Inputs
//    RealA, ImagA         Real and imaginary parts of transform array
// Outputs
//    RealA, ImagA         Real and imaginary parts of data array

  int i, j, N = RealA.nRow();
  Matrix RealT(N), ImagT(N);  // Temporary work vector

  //* Loop over the columns of the matrix
  for( j=1; j<=N; j++ ) {
	//* Copy out a column into a vector and take its complex conjugate
    for( i=1; i<=N; i++ ) {
      RealT(i) = RealA(i,j);      
      ImagT(i) = -1.0*ImagA(i,j); 
    }
	//* Take FFT of the vector
    fft(RealT,ImagT);         
	//* Copy the transformed vector back into the column
    for( i=1; i<=N; i++ ) {
      RealA(i,j) = RealT(i);      
      ImagA(i,j) = ImagT(i);
    }
  }

  //* Loop over the rows of the matrix
  double invN2 = 1.0/(N*N);
  for( i=1; i<=N; i++ ) {
	//* Copy out a row into a vector and take its complex conjugate
    for( j=1; j<=N; j++ ) {
      RealT(j) = RealA(i,j);  
      ImagT(j) = ImagA(i,j);
    }
	//* Take FFT of the vector
    fft(RealT,ImagT);         
	//* Copy the transformed vector back, taking its complex conjugate
	//  and applying the 1/N normalization
    for( j=1; j<=N; j++ ) {
      RealA(i,j) = RealT(j)*invN2;      
      ImagA(i,j) = -1.0*ImagT(j)*invN2; 
    }                                   
  }
}