Initialize and accessing 2D matrix pointer C++

I'm new in C++ and I try to get the "pointer mindset" to code with efficiency.

I'm trying to code el famoso Game of Life. So, I need a 2D matrix. Since it will be modify at each epoch (regarding the GoL rules), I would like to declare this matrix as a pointer to be able to point on it each time a cell evolves (and prevents rewriting).

I understood that I need to declare a vector of vector<int>. I proceed like this :

int i_size(50);
int j_size(50);
vector<int> arr(j_size, 0);
vector<vector<int>> *matrix;
matrix = new vector<vector<int>>;
    
for (int i(0); i<i_size; i++)
{
    matrix->push_back(arr);
}

Here, I hope to get a pointer matrix which point to a vector of vector well allocated and initialized at 0.

At this point, I'm stuck.

I know how to print the memory address of my matrix (cout << matrix).
I know how to print the memory address of each vector contained in matrix (cout << &matrix[i])

And that's all.

Questions

• Is my matrix declaration/initialization correct ?
• How can I access to my matrix ?
• How can I modify one cell of my matrix ?

Thanx all for your help. Again, I'm beginner so any advice will be helpful !

Your code is correct, but unecessarily complicated. You gain nothing by dynamically allocating the vector, because std::vector does already manage its elements dynamically.

Once you are settled with using a nested vector (*), the simple version of your code is this:

int i_size(50);
int j_size(50);
std::vector<std::vector<int>> matrix{ std::vector<int>(j_size),i_size};

Using more pointers does not imply more performance, but most of the time it implies more bugs and more complicated code. There is no "pointer mindset" that would lead to efficiency.

(*) What probably would gain you performance would be to use a flat 1D vector with index transformations. The big plus of a vector is its memory locality, but as the vector dynamically allocates an array to store the elements, the inner elements are not contigous in a nested vector. Alternatively consider to use std::array (which has no issue with nesting). When you do care about performance you must profile and measure.

First of all, your implementation is correct, as a matrix is created with the right size and you can access the value of the matrix with this syntax

(*matrix)[i][j]

In the same way you can modify the value of each element of the matrix by

(*matrix)[i][j]=value;

However, the use of a pointer doesn't increase the efficiency of your program in this case. Vector already take care of the dynamic memory allocation and deallocation.
By the way there is a more efficient way to declare the matrix with vector which would be

vector<vector<int>> matrix(i_size, vector<int>(j_size, 0));

I believe pointers are better use for arrays, so for exemple if you were to write your code like this.

int** matrix = new int*[i_size];
for (int i = 0; i < i_size; i++) {
    matrix[i] = new int[j_size]{};
}

Finally, if your goal is to code the game of life , I would like to point out that there is only two state for each sqaure so try to use bool instead of int