Writing a DFS with iterative deepening without recursion

Question

So currently i have a DFS with the following pseudocode

procedure DFS(Graph,source):
      create a stack S
      push source onto S
      mark source
      while S is not empty:
          pop an item from S into v
          for each edge e incident on v in Graph:
              let w be the other end of e
              if w is not marked:
                 mark w
                 push w onto S

How do I alter this function to accept a third argument that limits the depth of the search?

It's not fully correct DFS algorithm. It visits all successors of vertex first and then goes deeper. It should go deeper first, then backtrack to visit other child nodes. — MorioBoncz, Jun 18 '12 at 19:26

score 2 · Accepted Answer · answered Sep 21 '11 at 03:11

2

Let Node a structure for each node of the graph, add a field called level and then:

procedure DFS(Graph,source, depth):
  create a stack S
  source.level = 0
  push source onto S
  mark source
  while S is not empty:
      pop an item from S into v
      if v.level > depth
        continue

      for each edge e incident on v in Graph:
          let w be the other end of e
          if w is not marked:
             mark w
             w.level = v.level + 1
             push w onto S

answered Sep 21 '11 at 03:11

rendon

2,323
1
19
25

1

Pushing every neighbor into the stack would give a BFS because you end of looking at the neighbors first before the children. – Matilda Sep 27 '12 at 04:00
4

Nop. The key point here is the data structure, if you use a queue then you get a BFS, with a stack you get a DFS. – rendon Oct 01 '12 at 19:26
I think there are some problem in the above solution. You cannot pop an item from S into V immediately – spiralmoon Dec 01 '15 at 20:37
@ThinkRecursively I know it seems like Python or any other language, but it's pseudo code: `pop an item from S into v` can be translated into many instructions. – rendon Dec 02 '15 at 19:55
1

You pseudo-code is incorrect. You mark all `w` in the same iteration, so it __cannot__ be DFS; while @Matilda 's comment is also incorrect, he/she pointed out that your answer is probably incorrect, then a reasonable person will check one's answer. But you said "Nop. The key point here is data structure", – NeoZoom.lua Dec 31 '20 at 10:09
while DFS requires a stack, it doesn't mean that one use a stack in his code guarantee its code behave like DFS. – NeoZoom.lua Dec 31 '20 at 10:11

NeoZoom.lua · Answer 2 · 2020-12-31T10:43:47.837

The procedure will return success when an object is found.
When an object is found, the S will contain nodes in the path [root, source). (the source itself is not included.)

procedure DFS(Graph, source, depth):
    StackInit(S)
    if source is goal then
        return success
    markVisited(source)
    S.push(null, source)              // S is a close-list
    while S is not empty then do
        c, p := S.pop()
        r := next(c, p)               // return the next sibling of c
        if r is null then
            continue
        S.push(r, p)
        if r is marked visited then   // that already marked means it cannot be goal
            continue
        if r is goal then
            return success
        markVisited(r)
        if equal(depth(r), depth) then // here is what OP wants.
            continue
        S.push(null, r)

return fail

Writing a DFS with iterative deepening without recursion

2 Answers2

Linked