Iterative deepening vs depth-first search - algorithm

I keep reading about iterative deepening, but I don't understand how it differs from depth-first search.
I understood that depth-first search keeps going deeper and deeper.
In iterative deepening you establish a value of a level, if there is no solution at that level, you increment that value, and start again from scratch (the root).
Wouldn't this be the same thing as depth-first search?
I mean you would keep incrementing and incrementing, going deeper until you find a solution. I see this as the same thing! I would be going down the same branch, because if I start again from scratch I would go down the same branch as before.

In a depth-first search, you begin at some node in the graph and continuously explore deeper and deeper into the graph while you can find new nodes that you haven't yet reached (or until you find the solution). Any time the DFS runs out of moves, it backtracks to the latest point where it could make a different choice, then explores out from there. This can be a serious problem if your graph is extremely large and there's only one solution, since you might end up exploring the entire graph along one DFS path only to find the solution after looking at each node. Worse, if the graph is infinite (perhaps your graph consists of all the numbers, for example), the search might not terminate. Moreover, once you find the node you're looking for, you might not have the optimal path to it (you could have looped all over the graph looking for the solution even though it was right next to the start node!)
One potential fix to this problem would be to limit the depth of any one path taken by the DFS. For example, we might do a DFS search, but stop the search if we ever take a path of length greater than 5. This ensures that we never explore any node that's of distance greater than five from the start node, meaning that we never explore out infinitely or (unless the graph is extremely dense) we don't search the entire graph. However, this does mean that we might not find the node we're looking for, since we don't necessarily explore the entire graph.
The idea behind iterative deepening is to use this second approach but to keep increasing the depth at each level. In other words, we might try exploring using all paths of length one, then all paths of length two, then length three, etc. until we end up finding the node in question. This means that we never end up exploring along infinite dead-end paths, since the length of each path is capped by some length at each step. It also means that we find the shortest possible path to the destination node, since if we didn't find the node at depth d but did find it at depth d + 1, there can't be a path of length d (or we would have taken it), so the path of length d + 1 is indeed optimal.
The reason that this is different from a DFS is that it never runs into the case where it takes an extremely long and circuitous path around the graph without ever terminating. The lengths of the paths are always capped, so we never end up exploring unnecessary branches.
The reason that this is different from BFS is that in a BFS, you have to hold all of the fringe nodes in memory at once. This takes memory O(bd), where b is the branching factor. Compare this to the O(d) memory usage from iterative deepening (to hold the state for each of the d nodes in the current path). Of course, BFS never explores the same path multiple times, while iterative deepening may explore any path several times as it increases the depth limit. However, asymptotically the two have the same runtime. BFS terminates in O(bd) steps after considering all O(bd) nodes at distance d. Iterative deepening uses O(bd) time per level, which sums up to O(bd) overall, but with a higher constant factor.
In short:
DFS is not guaranteed to find an optimal path; iterative deepening is.
DFS may explore the entire graph before finding the target node; iterative deepening only does this if the distance between the start and end node is the maximum in the graph.
BFS and iterative deepening both run in time O(bd), but iterative deepening likely has a higher constant factor.
BFS uses O(bd) memory, while iterative deepening uses only O(d).

There is a decent page on wikipedia about this.
The basic idea I think you missed is that iterative deepening is primarily a heuristic. When a solution is likely to be found close to the root iterative deepening is will find it relatively fast while straightfoward depth-first-search could make a "wrong" decision and spend a lot of time on a fruitless deep branch.
(This is particularly important when the search tree can be infinite. In this case they are even less equivalent since DFS can get stuck forever while BFS or iterative deepening are sure to find the answer one day if it exists)

Just adding to what's already here, but here are some videos from University of Denver's Moving AI Lab that show the differences.
http://movingai.com/dfid.html
You can see in their examples iterative deepening wins when the goal is shallow (solution depth 3, tree depth) and the solution is on the right, but DFS wins no matter what if the solution is in the last row.
I got into this reading about chess programming, next up for me was thinking about quiescence search check that out if you want to know more about search strategies for AI programming.

Related

What is the point of IDA* vs A* algorithm

I don't understand how IDA* saves memory space.
From how I understand IDA* is A* with iterative deepening.
What's the difference between the amount of memory A* uses vs IDA*.
Wouldn't the last iteration of IDA* behave exactly like A* and use the same amount of memory. When I trace IDA* I realize that it also has to remember a priority queue of the nodes that are below the f(n) threshold.
I understand that ID-Depth first search helps depth first search by allowing it to do a breadth first like search while not having to remember every every node. But I thought A* already behaves like depth first as in it ignores some sub-trees along the way. How does Iteratively deepening make it use less memory?
Another question is Depth first search with iterative deepening allows you to find the shortest path by making it behave breadth first like. But A* already returns optimal shortest path (given that heuristic is admissible). How does iterative deepening help it. I feel like IDA*'s last iteration is identical to A*.
In IDA*, unlike A*, you don't need to keep a set of tentative nodes which you intend to visit, therefore, your memory consumption is dedicated only to the local variables of the recursive function.
Although this algorithm is lower on memory consumption, it has its own flaws:
Unlike A*, IDA* doesn't utilize dynamic programming and therefore often ends up exploring the same nodes many times. (IDA* In Wiki)
The heuristic function still needs to be specified for your case in order to not scan the whole graph, yet the scan's memory required in every moment is only the path you are currently scanning without its surrounding nodes.
Here is a demo of the memory required in each algorithm:
In the A* algorithm all of the nodes and their surrounding nodes needs to be included in the "need to visit" list while in the IDA* you get the next nodes "lazily" when you reach its previews node so you don't need to include it in an extra set.
As mentioned in the comments, IDA* is basically just IDDFS with heuristics:

Why is Depth-First Search said to suffer from infinite loops?

I have read about DFS and BFS many times but I have this doubt lingering my mind since long. In a lot of articles it is mentioned that DFS can get stuck in infinite loops.
As far as I know, this limitation can easily be removed by keeping track of the visited nodes. In fact, in all the books that I have read, this little check is a part of DFS.
So why are 'infinite loops' mentioned as a disadvantage of DFS? Is it just because the original DFS algorithm did not have this check for visited nodes? Please explain.
(1) In graph search algorithms [used frequently on AI], DFS's main advantage is space efficiency. It is its main advantage on BFS. However, if you keep track of visited nodes, you lose this advantage, since you need to store all visited nodes in memory. Don't forget the size of visited nodes increases drastically over time, and for very large/infinite graphs - might not fit in memory.
(2) Sometimes DFS can be in an infinite branch [in infinite graphs]. An infinite branch is a branch that does not end [always has "more sons"], and also does not get you to your target node, so for DFS, you might keep expanding this branch inifinitely, and 'miss' the good branch, that leads to the target node.
Bonus:
You can overcome this flaw in DFS, while maintaining relatively small memory size by using a combination of DFS and BFS: Iterative Deepening DFS
a conventional DFS algorithm does track down nodes. A local search algorithm does not track down states and behaves with amnesia. So I think the loop mainly refers to the one an infinite branch(a branch with infinite possible states). In that case, DFS simply goes down and become too focused on one branch.
If you do not check for cycles, then DFS can get stuck in one and never find its target whereas BFS will always expand out to all nodes at the next depth and therefore will eventually find its target, even if cycles exist.
Put simply:
If your graph can have cycles and you're using DFS, then you must account for cycles. On the other hand, BFS provides the option to ignore cycles at the expense of efficiency, which is often acceptable when searching a small number of nodes.

Question about breadth-first completeness vs depth-first incompleteness

According to Norvig in AIMA (Artificial Intelligence: A modern approach), the Depth-first algorithm is not complete (will not always produce a solution) because there are cases when the subtree being descended will be infinite.
On the other hand, the Breadth-first approach is said to be complete if the branching factor is not infinite. But isn't that somewhat the same "thing" as in the case of the subtree being infinite in DFS?
Can't the DFS be said to be complete if the tree's depth is finite? How is then that the BFS is complete and the DFS is not, since the completeness of the BFS relies on the branching factor being finite!
A tree can be infinite without having an infinite branching factor. As an example, consider the state tree for Rubik's Cube. Given a configuration of the cube, there is a finite number of moves (18, I believe, since a move consists of picking one of the 9 "planes" and rotating it in one of the two possible directions). However, the tree is infinitely deep, since it is perfectly possible to e.g. only rotate the same plane alternatingly back and forth (never making any progress). In order to prevent a DFS from doing this, one normally caches all the visited states (effectively pruning the state tree) - as you probably know. However, if the state space is too large (or actually infinite), this won't help.
I have not studied AI extensively, but I assume that the rationale for saying that BFS is complete while DFS is not (completeness is, after all, just a term that is defined somewhere) is that infinitely deep trees occur more frequently than trees with infinite branching factors (since having an infinite branching factor means that you have an infinite number of choices, which I believe is not common - games and simulations are usually discrete). Even for finite trees, BFS will normally perform better because DFS is very likely to start out on a wrong path, exploring a large portion of the tree before reaching the goal. Still, as you point out, in a finite tree, DFS will eventually find the solution if it exists.
DFS can not stuck in cycles (if we have a list of opened and closed states). The algorithm is not complete since it does not find a solution in an infinite space, even though the solution is in depth d which is much lower than infinity.
Imagine a strangely defined state space where each node has same number of successors as following number in Fibonacci sequence. So, it's recursively defined and therefore infinite. We're looking for node 2 (marked green in the graph). If DFS starts with the right branch of tree, it will take infinite number of steps to verify that our node is not there. Therefore it's not complete (it won't finish in reasonable time). BFS would find the solution in 3rd iteration.
Rubik's cube state space is finite, it is huge, but finite (human stuck in cycles but DFS won't repeat the same move twice). DFS would find very inefficient way how to solve it, sometimes this kind of solution is infeasible. Usually we consider maximum depth infinite, but our resources (memory) are always finite.
The properties of depth-first search depend strongly on whether the graph-search or
tree-search version is used. The graph-search version, which avoids repeated states and redundant
paths, is complete in finite state spaces because it will eventually expand every node.
The tree-search version, on the other hand, is not complete—for example, in Figure 3.6 the
algorithm will follow the Arad–Sibiu–Arad–Sibiu loop forever
Source: AI: a modern approach

When is backward search better than forward?

I'm studying graph search algorithms (for this question sake, lets limit algorithms only on DFS, BreadthFS, ID).
All these algorithms can be implemented as either forward search (from start node to end node) or backward search (from end node to start node).
My question is, when will backward search perform better than forward? Is there a general rule for that?
With a breadth-first search or iterative deepening, I think the mathematical answer to your question involves the notion of a "ball" around a vertex. Define Ball(v, n) to be the set of nodes at distance at most n from node v, and let the distance from the start node s to the destination node t be d. Then in the worst case a forward search will perform better than a backward search if |Ball(s, d)| < |Ball(t, d)|. This is true because breadth-first search always (and ID in the worst case) expands out all nodes of some distance k from the start node before ever visiting any nodes of depth k + 1. Consequently, if there's a smaller number of nodes around the start than the target a forward search should be faster, whereas if there's a smaller number of nodes around the target than the start and backward search should be faster. Unfortunately, it's hard to know this number a priori; you usually either have to run the search to determine which is the case. You could potentially use the branching factor around the two nodes as a heuristic for this value, but it wouldn't necessarily guarantee one search would be faster.
One interesting algorithm you might want to consider exploring is bidirectional breadth-first search, which does a search simultaneously from the source and target nodes. It tends to be much faster than the standard breadth-first search (in particular, with a branching factor b and distance d between the nodes, BFS takes roughly O(bd) time while bidirectional BFS takes O(bd/2)). It's also not that hard to code up once you have a good BFS implementation.
As for depth-first search, I actually don't know of a good way to determine which will be faster because in the worst-case both searches could explore the entire graph before finding a path. If someone has a good explanation about how to determine which will be better, it would be great if they could post it.

What are the practical factors to consider when choosing between Depth-First Search (DFS) and Breadth-First Search (BFS)? [closed]

Closed. This question is opinion-based. It is not currently accepting answers.
Want to improve this question? Update the question so it can be answered with facts and citations by editing this post.
Closed 2 years ago.
This post was edited and submitted for review 6 months ago and failed to reopen the post:
Original close reason(s) were not resolved
Improve this question
I understand the differences between DFS and BFS, but I'm interested to know what factors to consider when choosing DFS vs BFS.
Things like avoiding DFS for very deep trees, etc.
That heavily depends on the structure of the search tree and the number and location of solutions (aka searched-for items).
If you know a solution is not far from the root of the tree, a
breadth first search (BFS) might be better.
If the tree is very deep and solutions are rare, depth first search
(DFS) might take an extremely long time, but BFS could be faster.
If the tree is very wide, a BFS might need too much memory, so it
might be completely impractical.
If solutions are frequent but located deep in the tree, BFS could be
impractical.
If the search tree is very deep you will need to restrict the search
depth for depth first search (DFS), anyway (for example with
iterative deepening).
But these are just rules of thumb; you'll probably need to experiment.
I think in practice you'll usually not use these algorithms in their pure form anyway. There could be heuristics that help to explore promising parts of the search space first, or you might want to modify your search algorithm to be able to parallelize it efficiently.
Depth-first Search
Depth-first searches are often used in simulations of games (and game-like situations in the real world). In a typical game you can choose one of several possible actions. Each choice leads to further choices, each of which leads to further choices, and so on into an ever-expanding tree-shaped graph of possibilities.
For example in games like Chess, tic-tac-toe when you are deciding what move to make, you can mentally imagine a move, then your opponent’s possible responses, then your responses, and so on. You can decide what to do by seeing which move leads to the best outcome.
Only some paths in a game tree lead to your win. Some lead to a win by your opponent, when you reach such an ending, you must back up, or backtrack, to a previous node and try a different path. In this way you explore the tree until you find a path with a successful conclusion. Then you make the first move along this path.
Breadth-first search
The breadth-first search has an interesting property: It first finds all the vertices that are one edge away from the starting point, then all the vertices that are two edges away, and so on. This is useful if you’re trying to find the shortest path from the starting vertex to a given vertex. You start a BFS, and when you find the specified vertex, you know the path you’ve traced so far is the shortest path to the node. If there were a shorter path, the BFS would have found it already.
Breadth-first search can be used for finding the neighbour nodes in peer to peer networks like BitTorrent, GPS systems to find nearby locations, social networking sites to find people in the specified distance and things like that.
Nice Explanation from
http://www.programmerinterview.com/index.php/data-structures/dfs-vs-bfs/
An example of BFS
Here’s an example of what a BFS would look like. This is something like Level Order Tree Traversal where we will use QUEUE with ITERATIVE approach (Mostly RECURSION will end up with DFS). The numbers represent the order in which the nodes are accessed in a BFS:
In a depth first search, you start at the root, and follow one of the branches of the tree as far as possible until either the node you are looking for is found or you hit a leaf node ( a node with no children). If you hit a leaf node, then you continue the search at the nearest ancestor with unexplored children.
An example of DFS
Here’s an example of what a DFS would look like. I think post order traversal in binary tree will start work from the Leaf level first. The numbers represent the order in which the nodes are accessed in a DFS:
Differences between DFS and BFS
Comparing BFS and DFS, the big advantage of DFS is that it has much lower memory requirements than BFS, because it’s not necessary to store all of the child pointers at each level. Depending on the data and what you are looking for, either DFS or BFS could be advantageous.
For example, given a family tree if one were looking for someone on the tree who’s still alive, then it would be safe to assume that person would be on the bottom of the tree. This means that a BFS would take a very long time to reach that last level. A DFS, however, would find the goal faster. But, if one were looking for a family member who died a very long time ago, then that person would be closer to the top of the tree. Then, a BFS would usually be faster than a DFS. So, the advantages of either vary depending on the data and what you’re looking for.
One more example is Facebook; Suggestion on Friends of Friends. We need immediate friends for suggestion where we can use BFS. May be finding the shortest path or detecting the cycle (using recursion) we can use DFS.
Breadth First Search is generally the best approach when the depth of the tree can vary, and you only need to search part of the tree for a solution. For example, finding the shortest path from a starting value to a final value is a good place to use BFS.
Depth First Search is commonly used when you need to search the entire tree. It's easier to implement (using recursion) than BFS, and requires less state: While BFS requires you store the entire 'frontier', DFS only requires you store the list of parent nodes of the current element.
DFS is more space-efficient than BFS, but may go to unnecessary depths.
Their names are revealing: if there's a big breadth (i.e. big branching factor), but very limited depth (e.g. limited number of "moves"), then DFS can be more preferrable to BFS.
On IDDFS
It should be mentioned that there's a less-known variant that combines the space efficiency of DFS, but (cummulatively) the level-order visitation of BFS, is the iterative deepening depth-first search. This algorithm revisits some nodes, but it only contributes a constant factor of asymptotic difference.
When you approach this question as a programmer, one factor stands out: if you're using recursion, then depth-first search is simpler to implement, because you don't need to maintain an additional data structure containing the nodes yet to explore.
Here's depth-first search for a non-oriented graph if you're storing “already visited” information in the nodes:
def dfs(origin): # DFS from origin:
origin.visited = True # Mark the origin as visited
for neighbor in origin.neighbors: # Loop over the neighbors
if not neighbor.visited: dfs(neighbor) # Visit each neighbor if not already visited
If storing “already visited” information in a separate data structure:
def dfs(node, visited): # DFS from origin, with already-visited set:
visited.add(node) # Mark the origin as visited
for neighbor in node.neighbors: # Loop over the neighbors
if not neighbor in visited: # If the neighbor hasn't been visited yet,
dfs(neighbor, visited) # then visit the neighbor
dfs(origin, set())
Contrast this with breadth-first search where you need to maintain a separate data structure for the list of nodes yet to visit, no matter what.
One important advantage of BFS would be that it can be used to find the shortest path between any two nodes in an unweighted graph.
Whereas, we cannot use DFS for the same.
The following is a comprehensive answer to what you are asking.
In simple terms:
Breadth First Search (BFS) algorithm, from its name "Breadth", discovers all the neighbours of a node through the out edges of the node then it discovers the unvisited neighbours of the previously mentioned neighbours through their out edges and so forth, till all the nodes reachable from the origional source are visited (we can continue and take another origional source if there are remaining unvisited nodes and so forth). That's why it can be used to find the shortest path (if there is any) from a node (origional source) to another node if the weights of the edges are uniform.
Depth First Search (DFS) algorithm, from its name "Depth", discovers the unvisited neighbours of the most recently discovered node x through its out edges. If there is no unvisited neighbour from the node x, the algorithm backtracks to discover the unvisited neighbours of the node (through its out edges) from which node x was discovered, and so forth, till all the nodes reachable from the origional source are visited (we can continue and take another origional source if there are remaining unvisited nodes and so forth).
Both BFS and DFS can be incomplete. For example if the branching factor of a node is infinite, or very big for the resources (memory) to support (e.g. when storing the nodes to be discovered next), then BFS is not complete even though the searched key can be at a distance of few edges from the origional source. This infinite branching factor can be because of infinite choices (neighbouring nodes) from a given node to discover.
If the depth is infinite, or very big for the resources (memory) to support (e.g. when storing the nodes to be discovered next), then DFS is not complete even though the searched key can be the third neighbor of the origional source. This infinite depth can be because of a situation where there is, for every node the algorithm discovers, at least a new choice (neighbouring node) that is unvisited before.
Therefore, we can conclude when to use the BFS and DFS. Suppose we are dealing with a manageable limited branching factor and a manageable limited depth. If the searched node is shallow i.e. reachable after some edges from the origional source, then it is better to use BFS. On the other hand, if the searched node is deep i.e. reachable after a lot of edges from the origional source, then it is better to use DFS.
For example, in a social network if we want to search for people who have similar interests of a specific person, we can apply BFS from this person as an origional source, because mostly these people will be his direct friends or friends of friends i.e. one or two edges far.
On the other hand, if we want to search for people who have completely different interests of a specific person, we can apply DFS from this person as an origional source, because mostly these people will be very far from him i.e. friend of friend of friend.... i.e. too many edges far.
Applications of BFS and DFS can vary also because of the mechanism of searching in each one. For example, we can use either BFS (assuming the branching factor is manageable) or DFS (assuming the depth is manageable) when we just want to check the reachability from one node to another having no information where that node can be. Also both of them can solve same tasks like topological sorting of a graph (if it has).
BFS can be used to find the shortest path, with unit weight edges, from a node (origional source) to another. Whereas, DFS can be used to exhaust all the choices because of its nature of going in depth, like discovering the longest path between two nodes in an acyclic graph. Also DFS, can be used for cycle detection in a graph.
In the end if we have infinite depth and infinite branching factor, we can use Iterative Deepening Search (IDS).
I think it depends on what problems you are facing.
shortest path on simple graph -> bfs
all possible results -> dfs
search on graph(treat tree, martix as a graph too) -> dfs
....
Some algorithms depend on particular properties of DFS (or BFS) to work. For example the Hopcroft and Tarjan algorithm for finding 2-connected components takes advantage of the fact that each already visited node encountered by DFS is on the path from root to the currently explored node.
For BFS, we can consider Facebook example. We receive suggestion to add friends from the FB profile from other other friends profile. Suppose A->B, while B->E and B->F, so A will get suggestion for E And F. They must be using BFS to read till second level.
DFS is more based on scenarios where we want to forecast something based on data we have from source to destination. As mentioned already about chess or sudoku.
Once thing I have different here is, I believe DFS should be used for shortest path because DFS will cover the whole path first then we can decide the best. But as BFS will use greedy's approach so might be it looks like its the shortest path, but the final result might differ.
Let me know whether my understanding is wrong.
According to the properties of DFS and BFS.
For example,when we want to find the shortest path.
we usually use bfs,it can guarantee the 'shortest'.
but dfs only can guarantee that we can come from this point can achieve that point ,can not guarantee the 'shortest'.
Because Depth-First Searches use a stack as the nodes are processed, backtracking is provided with DFS. Because Breadth-First Searches use a queue, not a stack, to keep track of what nodes are processed, backtracking is not provided with BFS.
When tree width is very large and depth is low use DFS as recursion stack will not overflow.Use BFS when width is low and depth is very large to traverse the tree.
This is a good example to demonstrate that BFS is better than DFS in certain case. https://leetcode.com/problems/01-matrix/
When correctly implemented, both solutions should visit cells that have farther distance than the current cell +1.
But DFS is inefficient and repeatedly visited the same cell resulting O(n*n) complexity.
For example,
1,1,1,1,1,1,1,1,
1,1,1,1,1,1,1,1,
1,1,1,1,1,1,1,1,
0,0,0,0,0,0,0,0,

Resources