结果不正确：将 c++ 数组映射到特征矩阵

Question

我之前使用映射功能将现有内存映射到特征矩阵，但是当尝试映射 fftw c++ 数组时，我得到错误的结果，几乎如果数组的一部分（切片？）被映射到特征矩阵而不是整个内存块。 这是我正在使用的代码：

static const int nx = 10;
static const int ny = 10; 
static const int nyk = ny/2 + 1;
static const int nxk = nx/2 + 1;
static const int ncomp = 2;

fftw_complex *uhk; // this is pert Te
uhk= (fftw_complex*) fftw_malloc((((nx)*(ny+1))*nyk)* sizeof(fftw_complex)); 
memset(uhk, 42, (((nx))*nyk)* sizeof(fftw_complex));


    for (int i = 0; i < nx; i++){
        for (int j = 0; j < nyk; j++){
            for (int k = 0; k < ncomp; k++){
                uhk[i + nyk*j][k] = //taking fft of some expression
        }

    }
    

Eigen::Map<Eigen::MatrixXcd, Eigen::Unaligned> uhOut(reinterpret_cast<std::complex<double>*>(uhkOut),nyk,nx);

 std::cout << uhOut<< '\n';

我得到的结果是，

 (0.00,0.00)  (0.00,0.00)  (0.00,0.00)  (0.00,0.00)  (0.00,0.00)  (0.00,0.00)  (0.00,0.00)  (0.00,0.00)  (0.00,0.00)  (0.00,0.00)
 (0.00,0.00)  (0.00,0.00)  (0.00,0.00)  (0.00,0.00)  (0.00,0.00)  (0.00,0.00)  (0.00,0.00)  (0.00,0.00)  (0.00,0.00)  (0.00,0.00)
 (0.00,0.00)  (0.00,0.00)  (0.00,0.00)  (0.00,0.00)  (0.00,0.00)  (0.00,0.00)  (0.00,0.00)  (0.00,0.00)  (0.00,0.00)  (0.00,0.00)
 (0.00,0.00)  (0.00,0.00)  (0.00,0.00)  (0.00,0.00)  (0.00,0.00)  (0.00,0.00)  (0.00,0.00)  (0.00,0.00)  (0.00,0.00)  (0.00,0.00)
 (0.00,0.00) (0.58,-0.28) (0.95,-0.46) (0.95,-0.46) (0.58,-0.28) (0.00,-0.00)  (0.00,0.00)  (0.00,0.00)  (0.00,0.00)  (0.00,0.00)
 (0.00,0.00) (0.59,-0.09) (0.95,-0.14) (0.95,-0.14) (0.59,-0.09) (0.00,-0.00)  (0.00,0.00)  (0.00,0.00)  (0.00,0.00)  (0.00,0.00)

但是，正确的结果应该是：

     (0.00,0.00)     (5.14,0.00)     (8.32,0.00)     (8.32,0.00)     (5.14,0.00)     (0.00,0.00)    (-5.14,0.00)    (-8.32,0.00)    (-8.32,0.00)    (-5.14,0.00)
     (0.00,0.00)    (2.43,-2.35)    (3.93,-3.81)    (3.93,-3.81)    (2.43,-2.35)    (0.00,-0.00)    (-2.43,2.35)    (-3.93,3.81)    (-3.93,3.81)    (-2.43,2.35)
     (0.00,0.00)    (0.52,-1.04)    (0.85,-1.68)    (0.85,-1.68)    (0.52,-1.04)    (0.00,-0.00)    (-0.52,1.04)    (-0.85,1.68)    (-0.85,1.68)    (-0.52,1.04)
     (0.00,0.00)    (0.57,-0.55)    (0.93,-0.89)    (0.93,-0.89)    (0.57,-0.55)    (0.00,-0.00)    (-0.57,0.55)    (-0.93,0.89)    (-0.93,0.89)    (-0.57,0.55)
     (0.00,0.00)    (0.58,-0.28)    (0.95,-0.46)    (0.95,-0.46)    (0.58,-0.28)    (0.00,-0.00)    (-0.58,0.28)    (-0.95,0.46)    (-0.95,0.46)    (-0.58,0.28)
     (0.00,0.00)    (0.59,-0.09)    (0.95,-0.14)    (0.95,-0.14)    (0.59,-0.09)    (0.00,-0.00)    (-0.59,0.09)    (-0.95,0.14)    (-0.95,0.14)    (-0.59,0.09)

我在这里误解了map的意义吗？ 为什么我的结果会被这样分割？

Answer 1

你计算你的偏移量是错误的。 [i+nyk*j]不正确。 j的最大值是(nyk-1)。 我认为您的意思是[i*njk+j] 。

要诊断这类问题，我建议这样做

uhk[j + nyk*i][k] = (k==0)?i:j; // Note I swapped i/j in the offsets

这将创建一个矩阵，其中实部与列号匹配，虚部是行号。 像这样的简单模式很明显，这些值没有正确写入预期的内存位置。