BFS using deque - c++

Question

Really having trouble figuring out how to fix my code. I know there are obviously errors since it isn't running, but I'm unsure of what exactly they are, or how to go about fixing them. Any help/insight would be extremely appreciated. Thank you!!
struct vertices
{
int value;
int parent;
int visited;
int distance;
};
int BFS(vertices *v, int **adj_matrix, int num_nodes)
{
int target;
int cur_v = 0;
bool found = false;
int steps = 0;
cin >> target >> num_nodes;
adj_matrix [num_nodes][num_nodes];
deque<int> q;
for(int i = 0; i < num_nodes; i++)
{
v[i].visited = 0;
v[i].distance = INFINITY;
v[i].parent = 0;
v[1].visited = 1;
v[i].distance = 0;
q.push_front(v[1].value);
while(!q.empty())
{
cur_v = q.front();
q.pop_back();
v[cur_v].visited = 1;
for(int n=0; n< num_nodes; n++)
{
if(adj_matrix[cur_v][i] == n)
{
if(v[n].visited == 0)
{
v[n].visited = 1;
v[n].distance = ((v[cur_v].distance)+1);
v[n].parent = cur_v;
q.push_front(v[n].value);
steps ++;
}
}
}
}
}
return steps;
}
int main()
{
int target;
int num_nodes;
cin >> target;
cin >> num_nodes;
vertices *v = new vertices[num_nodes];
int **adj_matrix [num_nodes][num_nodes];
for(int i=0; i < num_nodes; ++i)
{
int node;
int value;
cin >> node >> value;
int num_edges;
cin >> num_edges;
for(int j=0; j<num_edges;++j)
{
int other_node;
cin >> other_node;
**adj_matrix[node][other_node] = 1;
**adj_matrix[other_node][node] = 1;
}
}
}

Answer 1

The first obvious error is that you are using the wrong data
structure. When you are implementing known concepts like matrix,
BFS, you need to spend a good amount of time to think about how to
implement inputs , outputs, data structures from the algorithm.
Some people like to use std::vector<std::vector<int>> for matrices. I almost always uses std::vector<int> for matrix data.
Second error is that you are mutating the algorithm. That thing is not BFS. It is not obvious at first.
You call multiple instance of BFS inside you BFS implementation. Using the same input vertices, whose elements are being modified. At the next loop, you dont start with a clean state.
You are flattening the algorithm, removing abstractions. What if you now have to use a adjency list? You have to modify the whole thing.
Third obvious error is coding style. You didnt do such a bad job because the vertices structure is self explanatory
I had to highlight all instances of cur_vto realizes that it stands for current vertex.
Using n for a counter is not a great idea.
Taking input inside the algorithm implementation is a big nono.