递归到迭代 DFS python

Question

我正在尝试将递归代码转换为迭代代码。 任务是在网格中找到最大的区域（由单元组成的连接单元）。

代码从这里引用： https://www.geeksforgeeks.org/find-length-largest-region-boolean-matrix/我尝试使用堆栈和循环来替换递归，但它不起作用

这是我尝试过的代码，它不会重现与递归方法相同的结果。

我已经测试过

M = np.array([[0, 0, 1, 1, 0], [1, 0, 1, 1, 0], [0, 1, 0, 0, 0], [0, 0, 0, 0, 2],[0, 0, 0, 1, 0],[0, 0, 3, 0, 0]]) 

def DFS(M, row, col, visited): 
    rowNbr = [-1, -1, -1, 0, 0, 1, 1, 1]  
    colNbr = [-1, 0, 1, -1, 1, -1, 0, 1]  

    # Mark this cell as visited  
    visited[row][col] = True
    stack = [] 

    for k in range(8): 
        if (isSafe(M, row + rowNbr[k],  
                   col + colNbr[k], visited)): 

            stack.push(M[row][col])
            row = row + rowNbr[k]
            col = col + colNbr[k]
            for k in range(8): 
                if (isSafe(M, row + rowNbr[k],  
                    col + colNbr[k], visited)):
                    stack.push(M[row][col])

Answer 1

使用堆栈的 DFS 的一般结构是

stack = [] # initialize Stack
visited = set() # initialize hash table for looking at visited nodes
stack.append(startNode) # put in the start node

while len(stack) != 0: # check whether there is anything in the To-Do list
   newNode = stack.pop() # get next node to visit
   if newNode not in visited: # update visited if this node has not been visited
      visited.add(newNode) 
   for neighbor in newNode.neighbors: # iterate over neighbors
      if neighbor not in visited: # check whether neighbors were visited
         stack.append(neighbor) # this node was not seen before, add it to To-Do list

在您的情况下，您似乎没有使迭代次数取决于堆栈中是否还有一个元素，并且您没有从堆栈中获取下一个要访问的元素，因为您没有从中取出任何元素。

您可以尝试以下方法：

def DFS(M, row, col, visited):
    rowNbr = [-1, -1, -1, 0, 0, 1, 1, 1]
    colNbr = [-1, 0, 1, -1, 1, -1, 0, 1]
    # initialize stack
    stack = [] 
    stack.append((row, col))
    while len(stack) != 0:
        row, col = stack.pop()
        if not visited[row, col]:
            visited[row, col] = 1
        # iterate over neighbors
        for k in range(8):
            if (isSafe(M, row + rowNbr[k], col + colNbr[k], visited)):
                row = row + rowNbr[k]
                col = col + colNbr[k]
                if not visited[row, col]:
                    stack.append((row, col))

这使用(row, col)的元组来标记位置并将它们存储在堆栈中。