Reputation: 159
My Data Structure:
class Cell
{
public:
struct CellLink
{
Cell *cell;
int weight;
};
public:
int row;
int column;
vector<CellLink> neighbors;
State state;
int totalCost = 0;
};
The primary function:
void AI::IterativeDeepeningSearch(Cell* cell)
{
Cell* temp;
int bound = 0;
while (true)
{
naturalFailure = false;
temp = IDShelper(cell, bound);
if (IsExit(temp))
{
break;
}
bound++;
}
}
The Helper:
Cell* AI::IDShelper(Cell* cell, int bound)
{
Cell* temp = cell;
SetEnvironment(cell, State::visited);
PrintEnvironment();
if (bound > 0)
{
for (int i = 0; i < cell->neighbors.size(); i++)
{
temp = IDShelper(cell->neighbors[i].cell, bound - 1);
if (IsExit(temp))
{
naturalFailure = true;
return temp;
}
}
}
else if (IsExit(cell))
{
return cell;
}
return temp;
}
I have made an Iterative Deepening Search for a maze. The problem is that it is taking literally hours to complete the search on a 21x21 maze while other algorithms take a couple of seconds.
I know that IDS is supposed to be slow but is it supposed to be that slow?
Upvotes: 1
Views: 1114
Reputation: 5187
I think I can see why this is slow.
In your helper, you're visiting the neighbors like so:
if (bound > 0)
{
for (int i = 0; i < cell->neighbors.size(); i++)
{
temp = IDShelper(cell->neighbors[i].cell, bound - 1);
if (IsExit(temp))
{
naturalFailure = true;
return temp;
}
}
}
but you're never using past results. You mark something as visited, but never check whether it is already visited.
Upvotes: 1