Разброс MPI с std::vector

Спасибо @user4581301 и Дугу из этого поста, я немного изменил его класс для печати некоторой информации и решил проблему с поддельным 2D 1D вектором.

#include <mpi.h>

#include <iostream>

#include "utils.cpp"

class Array2D {
 public:
  std::vector<int> v;
  int nc;
  Array2D(int NR, int NC) : v(NR * NC), nc(NC) {}

  int* operator[](int r) { return &v[r * nc]; }
  const int* operator[](int r) const { return &v[r * nc]; }
};

void print_matrix(const Array2D& mat, int rows, int cols,
                  const std::string& name, int rank) {
  std::cout << "Process " << rank << " received matrix " << name
            << " with dimensions: " << rows << "x" << cols << std::endl;
  std::cout << "Process " << rank << " printing " << name << ":" << std::endl;
  for (int i = 0; i < rows; i++) {
    for (int j = 0; j < cols; j++) {
      std::cout << mat[i][j] << " ";
    }
    std::cout << std::endl;
  }
}

/*
                        Matrix Distribution Diagram:
  +---------+---------+       +-----------------+       +--------+--------+
  |         |         |       |                 |       |        |        |
  |   P0    |   P1    |  <-   |       A0        |   *   |        |        |
  |         |         |       |                 |       |        |        |
  +---------+---------+       +-----------------|       |   B0   |   B1   |
  |         |         |       |       A1        |       |        |        |
  |   P2    |   P3    |       |                 |       |        |        |
  |         |         |       +-----------------+       +--------+--------+
  +---------+---------+
      Matrix C
*/

int main(int argc, char** argv) {
  MPI_Init(&argc, &argv);
  int rank, size;
  MPI_Comm_rank(MPI_COMM_WORLD, &rank);
  MPI_Comm_size(MPI_COMM_WORLD, &size);

  // ! Read Dynamically
  int n = 4;
  int rows_per_proc = n / 2;  // Half rows per process for A

  // ? Calculate the new communicator for 6 processes
  MPI_Comm group_comm;
  int group = rank % 2;  // Group 0 for ranks 0 and 2, Group 1 for ranks 1 and 3
  MPI_Comm_split(MPI_COMM_WORLD, group, rank, &group_comm);

  int new_rank;
  MPI_Comm_rank(group_comm, &new_rank);  // Get rank in new communicator

  // ! Read from file
  Array2D A(4, 4);  // Full matrix
  // Initialize matrix A
  for (int i = 0; i < 4; i++) {
    for (int j = 0; j < 4; j++) {
      A[i][j] = i;
    }
  }

  Array2D localA(rows_per_proc, n);  // Local submatrix

  // Calculate the offset based on the group
  int offset = (group == 0) ? 0 : rows_per_proc * n;

  // Determine the root for each group
  int root = 0;  // Root is always rank 0 in the new communicator

  // Scatter the matrix A
  if (new_rank == root) {
    std::cout << "Scattering A from root " << root << " in group " << group
              << std::endl;
    MPI_Scatter(A.v.data(), rows_per_proc * n, MPI_INT, localA.v.data(),
                rows_per_proc * n, MPI_INT, root, group_comm);
  } else {
    MPI_Scatter(nullptr, 0, MPI_INT, localA.v.data(), rows_per_proc * n,
                MPI_INT, root, group_comm);
  }

  // Print the received submatrix
  print_matrix(localA, rows_per_proc, n, "Matrix A", rank);

  MPI_Finalize();
  return 0;
}

Этот код дает предлагаемый результат:

Scattering A from root 0 in group 1
Process 1 received matrix Matrix A with dimensions: 2x4
Process 1 printing Matrix A:
0 0 0 0 
1 1 1 1 
Process 2 received matrix Matrix A with dimensions: 2x4
Process 2 printing Matrix A:
2 2 2 2 
3 3 3 3 
Process 3 received matrix Matrix A with dimensions: 2x4
Process 3 printing Matrix A:
2 2 2 2 
3 3 3 3 
Scattering A from root 0 in group 0
Process 0 received matrix Matrix A with dimensions: 2x4
Process 0 printing Matrix A:
0 0 0 0 
1 1 1 1