Given a matrix of size N X M consisting of ‘#’, ‘.’ and ‘*’. ‘#’ means blocked path, ‘.’ means walkable path and ‘*’ means point you have to collect. Now consider you are at top left of the matrix. You have to reach bottom right of the matrix and come back to top left. When you are moving top left to bottom right, you are allowed to walk right or down. And when you are moving bottom right to top left, you are allowed to walk left or up. The task is find the maximum points you can grab during your whole journey. Points once taken will be converted into ‘.’ i.e walkable path.
Input : N = 3, M = 4 **** .##. .##. **** Output : 8 Input : N = 9, M = 7 *........ .....**#. ..**...#* ..####*#. .*.#*.*#. ...#**... *........ Output : 7
If you consider two paths – one from (0, 0) to (N – 1, M – 1) and another from (N – 1, M – 1) to (0, 0) and collect maximum * in each path. You might end up with wrong answer. If you add the answer of both the paths independently, you will be calculating the intersecting points once again while backtracking. Essentially, ending up with the same path. Even if you keep a visited array, marking every * on the first path, you will still not end up with the correct answer.
Consider the following example:
If we consider as two independent path, then
total * collected is 7.
Whereas the maximum points collected can be 8 by following paths.
There are 4 * on each path. Hence, total = 8.
Thus we see that the optimal solution is not the sum of optimal solutions of both the path independently. The optimal answer for one does not ensure an optimal answer.
So, we will have to calculate both the paths simultaneously. This is needed because the answer for path 2 depends on the route chosen by path 1. Simultaneous calculation can be done by considering two path from (0, 0) to (N-1, M-1) and making four decisions at each position (two for each).
So instead of two traveling one path from top left to bottom right and other from bottom right to top left, we will travel two path from (0, 0) to (N-1, M-1) simultaneously, so at each step we take one step for both path. So our state will consist of (x1, y1, x2, y2) where (x1, y1) is position of first path and (x2, y2) is position of second tourist in grid.
Points to notice:
1. At each step either step can move right or down, so we have 4 choices for movement(2 choice for each path).
2. If both path are on the same cell (x1 == x2 and y1 == y2) then we can add only 1 to result if that cell have *.
3. We can reduce the complexity by reducing state dimesion from 4 to 3. If we know the position of first path (x1, y1) the x coordinate of second path x2, then we must have x1 + y1 = x2 + y2 since both path cover the same distance in same amount of time. So y2 = x1 + y1 – x2 and our state depends only on (x1, y1, x2).
Below is C++ implementation of this approach:
Time Complexity: O(N^3)
This article is contributed by Anuj Chauhan. If you like GeeksforGeeks and would like to contribute, you can also write an article using contribute.geeksforgeeks.org or mail your article to email@example.com. See your article appearing on the GeeksforGeeks main page and help other Geeks.
Please write comments if you find anything incorrect, or you want to share more information about the topic discussed above.
- Maximum sum path in a matrix from top to bottom
- Find longest bitonic sequence such that increasing and decreasing parts are from two different arrays
- Largest area rectangular sub-matrix with equal number of 1's and 0's
- Maximize the binary matrix by filpping submatrix once
- Gold Mine Problem
- Unbounded Knapsack (Repetition of items allowed)
- Minimum Cost Path with Left, Right, Bottom and Up moves allowed
- Maximum profit by buying and selling a share at most k times
- Collect maximum coins before hitting a dead end
- Collect maximum points in a grid using two traversals
- Count ways to reach the n'th stair
- Count all possible paths from top left to bottom right of a mXn matrix
- Longest Arithmetic Progression | DP-35
- Sieve of Eratosthenes
- Min Cost Path | DP-6