I am using A* in order to solve the Asymmetric Traveling Salesman problem.
My state representation has 4 variables:
1 - Visited cities (List)
2 - Unvisitied cities (List)
3 - Current City (Integer)
4 - Current Cost (Integer)
However, even tho I find many path-construction algorithms such as Nearest Neighbor, k-opt and so on, I can't find an heuristic suitable for A*, which is, a h(n) function that takes a state as input and returns an integer corresponding to that state's quality.
So my question is, are there such heuristics? Any recommendations?
Thanks in advance
The weight of the minimum spanning tree of the subgraph that contains all unvisited vertices and the current vertex is a lower bound for the cost to finish the current path. It can be used with the A* algorithm as it can't overestimate the remaining distance (otherwise, the weight of the remaining path is smaller than the weight of minimum spanning tree and it spans the given vertices, which is a contradiction).
I've never tried it though so I don't know how well it'll work in practice.
There always are: h(n) = 0 always works. It is useless, turning A* into Dijkstra, but it's definitely admissible.
An other obvious one: let h(n) be the shortest edge from the current city back to the beginning. Still a huge underestimation, but at least it's not necessarily zero. It's obviously valid, the loop has to be closed eventually and (given this partial route) there is no shorter way to do it.
You can be a bit more clever here, for example you could use linear programming (make two variables for each edge, one for each direction, then for every city make a constraint forcing the sum of entering edges to be 1 and a constraint forcing the sum of exiting edges to be one, weights are obviously the distances) to find an underestimation of the length from the current node back to the beginning while touching every city in the set of unvisited cities. Of course if you're doing that, you might as well drop A* and just use the usual integer linear programming tricks. A* doesn't seem like a good fit here (especially in the beginning, the branching factor is too high and the heuristics won't guide it enough yet), but I haven't tried this so who knows.
Also, given the solution from the LP, you can improve it a lot by using some simple tricks (and some advanced tricks that whole books have been written about, but let's not go there, read the books if you want to know). For example, one thing the LP likes to do is form lots of little triangles. This will satisfy the degree constraints everywhere locally and keeps everything nice and short. But it's not a tour, and forcing it be more like a tour will make the heuristic higher=better. To remove the sub-tours, you can detect them in the fractional solution and then force the number of entries to the subgraph to be at least 1 (it may have to become more than 1 at some point, so don't force it to be exactly 1) and force the number of exits to be at least 1, by adding the corresponding constraints and solving again. There are many more tricks, but this should already give a very reasonable heuristic, much closer to the actual cost than using any of the overestimating heuristics and dividing them by their worst case overestimation factor. The problem with those is that usually the heuristic is pretty good, much better than their worst case factor, and then dividing by the worst case factor really kills the quality of the heuristic.
Related
I have a problem similar to the basic TSP but not quite the same.
I have a starting position for a player character, and he has to pick up n objects in the shortest time possible. He doesn't need to return to the original position and the order in which he picks up the objects does not matter.
In other words, the problem is to find the minimum-weight (distance) Hamiltonian path with a given (fixed) start vertex.
What I have currently, is an algorithm like this:
best_total_weight_so_far = Inf
foreach possible end vertex:
add a vertex with 0-weight edges to the start and end vertices
current_solution = solve TSP for this graph
remove the 0 vertex
total_weight = Weight (current_solution)
if total_weight < best_total_weight_so_far
best_solution = current_solution
best_total_weight_so_far = total_weight
However this algorithm seems to be somewhat time-consuming, since it has to solve the TSP n-1 times. Is there a better approach to solving the original problem?
It is a rather minor variation of TSP and clearly NP-hard. Any heuristic algorithm (and you really shouldn't try to do anything better than heuristic for a game IMHO) for TSP should be easily modifiable for your situation. Even the nearest neighbor probably wouldn't be bad -- in fact for your situation it would probably be better that when used in TSP since in Nearest Neighbor the return edge is often the worst. Perhaps you can use NN + 2-Opt to eliminate edge crossings.
On edit: Your problem can easily be reduced to the TSP problem for directed graphs. Double all of the existing edges so that each is replaced by a pair of arrows. The costs for all arrows is simply the existing cost for the corresponding edges except for the arrows that go into the start node. Make those edges cost 0 (no cost in returning at the end of the day). If you have code that solves the TSP for directed graphs you could thus use it in your case as well.
At the risk of it getting slow (20 points should be fine), you can use the good old exact TSP algorithms in the way John describes. 20 points is really easy for TSP - instances with thousands of points are routinely solved and instances with tens of thousands of points have been solved.
For example, use linear programming and branch & bound.
Make an LP problem with one variable per edge (there are more edges now because it's directed), the variables will be between 0 and 1 where 0 means "don't take this edge in the solution", 1 means "take it" and fractional values sort of mean "take it .. a bit" (whatever that means).
The costs are obviously the distances, except for returning to the start. See John's answer.
Then you need constraints, namely that for each node the sum of its incoming edges is 1, and the sum of its outgoing edges is one. Also the sum of a pair of edges that was previously one edge must be smaller or equal to one. The solution now will consist of disconnected triangles, which is the smallest way to connect the nodes such that they all have both an incoming edge and an outgoing edge, and those edges are not both "the same edge". So the sub-tours must be eliminated. The simplest way to do that (probably strong enough for 20 points) is to decompose the solution into connected components, and then for each connected component say that the sum of incoming edges to it must be at least 1 (it can be more than 1), same thing with the outgoing edges. Solve the LP problem again and repeat this until there is only one component. There are more cuts you can do, such as the obvious Gomory cuts, but also fancy special TSP cuts (comb cuts, blossom cuts, crown cuts.. there are whole books about this), but you won't need any of this for 20 points.
What this gives you is, sometimes, directly the solution. Usually to begin with it will contain fractional edges. In that case it still gives you a good underestimation of how long the tour will be, and you can use that in the framework of branch & bound to determine the actual best tour. The idea there is to pick an edge that was fractional in the result, and pick it either 0 or 1 (this often turns edges that were previously 0/1 fractional, so you have to keep all "chosen edges" fixed in the whole sub-tree in order to guarantee termination). Now you have two sub-problems, solve each recursively. Whenever the estimation from the LP solution becomes longer than the best path you have found so far, you can prune the sub-tree (since it's an underestimation, all integral solutions in this part of the tree can only be even worse). You can initialize the "best so far solution" with a heuristic solution but for 20 points it doesn't really matter, the techniques I described here are already enough to solve 100-point problems.
I have a graph that represents a city. I know the location of places of interest (nodes, which have a Importance value), the location of the hotel I'm staying in, how the nodes are connected, the traversal time between them and have acess to latitude and longitude. There are no issues converting from time to distance and vice-versa.
The objective is to tour the city, maximizing the importance per day but limiting one day of travel to 10 hours. A day begins and ends at the hotel. I have a working A* algorithm that chooses the lowest value but with no heuristic yet, which I guess makes it a BB for now. With that in mind:
Since I have access to Lat/Long, my first stab at an heuristic, while
only dealing with times, would be the distance as the crow flies
between a node and the hotel. Would this be an admissible heuristic?
It gives me the shortest possible distance and time, so it wouldn't
overestimate.
Now let's say the Importance of a node is between 1-4. In order to factor it in, one idea could be g(neighbor) = g(current) + (edge_cost / Importance^2). Assuming this would be valid (if not, why?):
But now the heuristic values would be in a different unit. Could a solution to this simply be give the Hotel Importance = 1? If the value is the same, will it still be admissible? EDIT: I think this will end up giving me problems because of the difference in scale.
I still have to restrict the total amount of time. Should each node keep track of the total time spent, in order to compare to the limit, plus the g() and h() values, because of the different units?
And finally:
Since I have to start and end in the same node, what comes to mind is to explore a node and should I find the hotel see if I still have time to explore the neighbors instead of going back. However, if I still have time to expand to one more node, but time runs out and I can't get to the hotel from there, I'm assuming I'll have to backtrack to the parent.
I can't help but see similarities to the knapsack problem. Even though I have to use A*, is there any lesson I can take from it?
Must my heuristic be consistent in this case? If so, why?
By the way, the purpose here is pathfinding first, optimizations second.
This actually looks like a combination of the travelling salesman problem (TSP) and knapsack problem (KP). It's KP in this respect: the knapsack capacity is 10 (for total hours available in a day) and the locations are the items. The item value equals the location value. The item weight is equal to the time it takes to travel to the location (plus the location's portion of the trip back to the hotel). The challenge arises from the fact that an item's weight is unknown until you solve the optimal tour through the selected locations--enter the TSP and Pathfinding.
One approach might be to use a pathfinding algorithm (e.g. A*, Bellman–Ford, or Dijkstra's algorithm) primarily to compute a distance matrix between each node. The distance matrix can then be leveraged while solving the TSP portion of the problem: finding a tour through the locations and using the total time as the weight.
The next step is up to you. If you are looking for an approximate solution, many heuristics exist for both TSP and KP: See Christofides TSP Heuristic, or the Minimum TSP and Maximum Knapsack entries at the Compendium of NP Optimization problems.
If on the other hand you seek an optimal solution, you may be out of luck. Still I recommend you find a copy of Graph Theory. An Algorithmic Approach by Nicos Christofides (ISBN-13: 978-0121743505). It provides heuristics for early backtracking in a Depth-First-Search that expedite the search for optimal solutions to several NP-Complete problems.
NOTE: Due to the fact that the trip does not end at the same place it started and also the fact that every point can be visited more than once as long as I still visit all of them, this is not really a TSP variant, but I put it due to lack of a better definition of the problem.
So..
Suppose I am going on a hiking trip with n points of interest. These points are all connected by hiking trails. I have a map showing all trails with their distances, giving me a directed graph.
My problem is how to approximate a tour that starts at a point A and visits all n points of interest, while ending the tour anywhere but the point where I started and I want the tour to be as short as possible.
Due to the nature of hiking, I figured this would sadly not be a symmetric problem (or can I convert my asymmetric graph to a symmetric one?), since going from high to low altitude is obviously easier than the other way around.
Also I believe it has to be an algorithm that works for non-metric graphs, where the triangle inequality is not satisfied, since going from a to b to c might be faster than taking a really long and weird road that goes from a to c directly. I did consider if triangle inequality still holds, since there are no restrictions regarding how many times I visit each point, as long as I visit all of them, meaning I would always choose the shortest of two distinct paths from a to c and thus never takr the long and weird road.
I believe my problem is easier than TSP, so those algorithms do not fit this problem. I thought about using a minimum spanning tree, but I have a hard time convincing myself that they can be applied to a non-metric asymmetric directed graph.
What I really want are some pointers as to how I can come up with an approximation algorithm that will find a near optimal tour through all n points
To reduce your problem to asymmetric TSP, introduce a new node u and make arcs of length L from u to A and from all nodes but A to u, where L is very large (large enough that no optimal solution revisits u). Delete u from the tour to obtain a path from A to some other node via all others. Unfortunately this reduction preserves the objective only additively, which make the approximation guarantees worse by a constant factor.
The target of the reduction Evgeny pointed out is non-metric symmetric TSP, so that reduction is not useful to you, because the approximations known all require metric instances. Assuming that the collection of trails forms a planar graph (or is close to it), there is a constant-factor approximation due to Gharan and Saberi, which may unfortunately be rather difficult to implement, and may not give reasonable results in practice. Frieze, Galbiati, and Maffioli give a simple log-factor approximation for general graphs.
If there are a reasonable number of trails, branch and bound might be able to give you an optimal solution. Both G&S and branch and bound require solving the Held-Karp linear program for ATSP, which may be useful in itself for evaluating other approaches. For many symmetric TSP instances that arise in practice, it gives a lower bound on the cost of an optimal solution within 10% of the true value.
You can simplify this problem to a normal TSP problem with n+1 vertexes. To do this, take node 'A' and all the points of interest and compute a shortest path between each pair of these points. You can use the all-pairs shortest path algorithm on the original graph. Or, if n is significantly smaller than the original graph size, use single-source shortest path algorithm for these n+1 vertexes. Also you can set length of all the paths, ending at 'A', to some constant, larger than any other path, which allows to end the trip anywhere (this may be needed only for TSP algorithms, finding a round-trip path).
As a result, you get a complete graph, which is metric, but still asymmetric. All you need now is to solve a normal TSP problem on this graph. If you want to convert this asymmetric graph to a symmetric one, Wikipedia explains how to do it.
In a tower defense game, you have an NxM grid with a start, a finish, and a number of walls.
Enemies take the shortest path from start to finish without passing through any walls (they aren't usually constrained to the grid, but for simplicity's sake let's say they are. In either case, they can't move through diagonal "holes")
The problem (for this question at least) is to place up to K additional walls to maximize the path the enemies have to take. For example, for K=14
My intuition tells me this problem is NP-hard if (as I'm hoping to do) we generalize this to include waypoints that must be visited before moving to the finish, and possibly also without waypoints.
But, are there any decent heuristics out there for near-optimal solutions?
[Edit] I have posted a related question here.
I present a greedy approach and it's maybe close to the optimal (but I couldn't find approximation factor). Idea is simple, we should block the cells which are in critical places of the Maze. These places can help to measure the connectivity of maze. We can consider the vertex connectivity and we find minimum vertex cut which disconnects the start and final: (s,f). After that we remove some critical cells.
To turn it to the graph, take dual of maze. Find minimum (s,f) vertex cut on this graph. Then we examine each vertex in this cut. We remove a vertex its deletion increases the length of all s,f paths or if it is in the minimum length path from s to f. After eliminating a vertex, recursively repeat the above process for k time.
But there is an issue with this, this is when we remove a vertex which cuts any path from s to f. To prevent this we can weight cutting node as high as possible, means first compute minimum (s,f) cut, if cut result is just one node, make it weighted and set a high weight like n^3 to that vertex, now again compute the minimum s,f cut, single cutting vertex in previous calculation doesn't belong to new cut because of waiting.
But if there is just one path between s,f (after some iterations) we can't improve it. In this case we can use normal greedy algorithms like removing node from a one of a shortest path from s to f which doesn't belong to any cut. after that we can deal with minimum vertex cut.
The algorithm running time in each step is:
min-cut + path finding for all nodes in min-cut
O(min cut) + O(n^2)*O(number of nodes in min-cut)
And because number of nodes in min cut can not be greater than O(n^2) in very pessimistic situation the algorithm is O(kn^4), but normally it shouldn't take more than O(kn^3), because normally min-cut algorithm dominates path finding, also normally path finding doesn't takes O(n^2).
I guess the greedy choice is a good start point for simulated annealing type algorithms.
P.S: minimum vertex cut is similar to minimum edge cut, and similar approach like max-flow/min-cut can be applied on minimum vertex cut, just assume each vertex as two vertex, one Vi, one Vo, means input and outputs, also converting undirected graph to directed one is not hard.
it can be easily shown (proof let as an exercise to the reader) that it is enough to search for the solution so that every one of the K blockades is put on the current minimum-length route. Note that if there are multiple minimal-length routes then all of them have to be considered. The reason is that if you don't put any of the remaining blockades on the current minimum-length route then it does not change; hence you can put the first available blockade on it immediately during search. This speeds up even a brute-force search.
But there are more optimizations. You can also always decide that you put the next blockade so that it becomes the FIRST blockade on the current minimum-length route, i.e. you work so that if you place the blockade on the 10th square on the route, then you mark the squares 1..9 as "permanently open" until you backtrack. This saves again an exponential number of squares to search for during backtracking search.
You can then apply heuristics to cut down the search space or to reorder it, e.g. first try those blockade placements that increase the length of the current minimum-length route the most. You can then run the backtracking algorithm for a limited amount of real-time and pick the best solution found thus far.
I believe we can reduce the contained maximum manifold problem to boolean satisifiability and show NP-completeness through any dependency on this subproblem. Because of this, the algorithms spinning_plate provided are reasonable as heuristics, precomputing and machine learning is reasonable, and the trick becomes finding the best heuristic solution if we wish to blunder forward here.
Consider a board like the following:
..S........
#.#..#..###
...........
...........
..........F
This has many of the problems that cause greedy and gate-bound solutions to fail. If we look at that second row:
#.#..#..###
Our logic gates are, in 0-based 2D array ordered as [row][column]:
[1][4], [1][5], [1][6], [1][7], [1][8]
We can re-render this as an equation to satisfy the block:
if ([1][9] AND ([1][10] AND [1][11]) AND ([1][12] AND [1][13]):
traversal_cost = INFINITY; longest = False # Infinity does not qualify
Excepting infinity as an unsatisfiable case, we backtrack and rerender this as:
if ([1][14] AND ([1][15] AND [1][16]) AND [1][17]:
traversal_cost = 6; longest = True
And our hidden boolean relationship falls amongst all of these gates. You can also show that geometric proofs can't fractalize recursively, because we can always create a wall that's exactly N-1 width or height long, and this represents a critical part of the solution in all cases (therefore, divide and conquer won't help you).
Furthermore, because perturbations across different rows are significant:
..S........
#.#........
...#..#....
.......#..#
..........F
We can show that, without a complete set of computable geometric identities, the complete search space reduces itself to N-SAT.
By extension, we can also show that this is trivial to verify and non-polynomial to solve as the number of gates approaches infinity. Unsurprisingly, this is why tower defense games remain so fun for humans to play. Obviously, a more rigorous proof is desirable, but this is a skeletal start.
Do note that you can significantly reduce the n term in your n-choose-k relation. Because we can recursively show that each perturbation must lie on the critical path, and because the critical path is always computable in O(V+E) time (with a few optimizations to speed things up for each perturbation), you can significantly reduce your search space at a cost of a breadth-first search for each additional tower added to the board.
Because we may tolerably assume O(n^k) for a deterministic solution, a heuristical approach is reasonable. My advice thus falls somewhere between spinning_plate's answer and Soravux's, with an eye towards machine learning techniques applicable to the problem.
The 0th solution: Use a tolerable but suboptimal AI, in which spinning_plate provided two usable algorithms. Indeed, these approximate how many naive players approach the game, and this should be sufficient for simple play, albeit with a high degree of exploitability.
The 1st-order solution: Use a database. Given the problem formulation, you haven't quite demonstrated the need to compute the optimal solution on the fly. Therefore, if we relax the constraint of approaching a random board with no information, we can simply precompute the optimum for all K tolerable for each board. Obviously, this only works for a small number of boards: with V! potential board states for each configuration, we cannot tolerably precompute all optimums as V becomes very large.
The 2nd-order solution: Use a machine-learning step. Promote each step as you close a gap that results in a very high traversal cost, running until your algorithm converges or no more optimal solution can be found than greedy. A plethora of algorithms are applicable here, so I recommend chasing the classics and the literature for selecting the correct one that works within the constraints of your program.
The best heuristic may be a simple heat map generated by a locally state-aware, recursive depth-first traversal, sorting the results by most to least commonly traversed after the O(V^2) traversal. Proceeding through this output greedily identifies all bottlenecks, and doing so without making pathing impossible is entirely possible (checking this is O(V+E)).
Putting it all together, I'd try an intersection of these approaches, combining the heat map and critical path identities. I'd assume there's enough here to come up with a good, functional geometric proof that satisfies all of the constraints of the problem.
At the risk of stating the obvious, here's one algorithm
1) Find the shortest path
2) Test blocking everything node on that path and see which one results in the longest path
3) Repeat K times
Naively, this will take O(K*(V+ E log E)^2) but you could with some little work improve 2 by only recalculating partial paths.
As you mention, simply trying to break the path is difficult because if most breaks simply add a length of 1 (or 2), its hard to find the choke points that lead to big gains.
If you take the minimum vertex cut between the start and the end, you will find the choke points for the entire graph. One possible algorithm is this
1) Find the shortest path
2) Find the min-cut of the whole graph
3) Find the maximal contiguous node set that intersects one point on the path, block those.
4) Wash, rinse, repeat
3) is the big part and why this algorithm may perform badly, too. You could also try
the smallest node set that connects with other existing blocks.
finding all groupings of contiguous verticies in the vertex cut, testing each of them for the longest path a la the first algorithm
The last one is what might be most promising
If you find a min vertex cut on the whole graph, you're going to find the choke points for the whole graph.
Here is a thought. In your grid, group adjacent walls into islands and treat every island as a graph node. Distance between nodes is the minimal number of walls that is needed to connect them (to block the enemy).
In that case you can start maximizing the path length by blocking the most cheap arcs.
I have no idea if this would work, because you could make new islands using your points. but it could help work out where to put walls.
I suggest using a modified breadth first search with a K-length priority queue tracking the best K paths between each island.
i would, for every island of connected walls, pretend that it is a light. (a special light that can only send out horizontal and vertical rays of light)
Use ray-tracing to see which other islands the light can hit
say Island1 (i1) hits i2,i3,i4,i5 but doesn't hit i6,i7..
then you would have line(i1,i2), line(i1,i3), line(i1,i4) and line(i1,i5)
Mark the distance of all grid points to be infinity. Set the start point as 0.
Now use breadth first search from the start. Every grid point, mark the distance of that grid point to be the minimum distance of its neighbors.
But.. here is the catch..
every time you get to a grid-point that is on a line() between two islands, Instead of recording the distance as the minimum of its neighbors, you need to make it a priority queue of length K. And record the K shortest paths to that line() from any of the other line()s
This priority queque then stays the same until you get to the next line(), where it aggregates all priority ques going into that point.
You haven't showed the need for this algorithm to be realtime, but I may be wrong about this premice. You could then precalculate the block positions.
If you can do this beforehand and then simply make the AI build the maze rock by rock as if it was a kind of tree, you could use genetic algorithms to ease up your need for heuristics. You would need to load any kind of genetic algorithm framework, start with a population of non-movable blocks (your map) and randomly-placed movable blocks (blocks that the AI would place). Then, you evolve the population by making crossovers and transmutations over movable blocks and then evaluate the individuals by giving more reward to the longest path calculated. You would then simply have to write a resource efficient path-calculator without the need of having heuristics in your code. In your last generation of your evolution, you would take the highest-ranking individual, which would be your solution, thus your desired block pattern for this map.
Genetic algorithms are proven to take you, under ideal situation, to a local maxima (or minima) in reasonable time, which may be impossible to reach with analytic solutions on a sufficiently large data set (ie. big enough map in your situation).
You haven't stated the language in which you are going to develop this algorithm, so I can't propose frameworks that may perfectly suit your needs.
Note that if your map is dynamic, meaning that the map may change over tower defense iterations, you may want to avoid this technique since it may be too intensive to re-evolve an entire new population every wave.
I'm not at all an algorithms expert, but looking at the grid makes me wonder if Conway's game of life might somehow be useful for this. With a reasonable initial seed and well-chosen rules about birth and death of towers, you could try many seeds and subsequent generations thereof in a short period of time.
You already have a measure of fitness in the length of the creeps' path, so you could pick the best one accordingly. I don't know how well (if at all) it would approximate the best path, but it would be an interesting thing to use in a solution.
A-star is used to find the shortest path between a startnode and an endnode in a graph. What algorithm is used to solve something were the target state isn't specifically known and we instead only have a criteria for the target state?
For example, can a sudoku puzzle be solved with an Astar-like algorithm? We dont know how the endstate will look like (which number is where) but we do know the rules of sudoku, a criteria for a winning state. Therefore I have a startnode and just a criteria for the endnode, which algorithm to use?
A* requires a graph, a cost function for traversal of that graph, a heuristic as to whether a node in the graph is closer to the goal than another, and a test whether the goal is reached.
Searching a Sudoku solution space doesn't really have a traversal cost to minimize, only a global cost ( the number of unsolved squares ), so all traversals would be equal cost, so A* doesn't really help - any cell you could assign costs one move and moves you one closer to the goal, so A* would be no better than choosing the next step at random.
It might be possible to apply an A* search based on the estimated/measured cost of applying the different techniques at each point, which would then try to find a path through the solution space with the least computational cost. In that case the graph would not just be the solution states of the puzzle, but you'd be choosing between the techniques to apply at that point - you'd know an estimate of the cost of a transition, but not where that transition 'goes', except that if successful, it's one step closer to the goal.
Yes, A* can be used when a specific goal state cannot be identified. (Pete Kirkham's answer implies this, but doesn't emphasise it much.)
When a specific goal state can't be identified, it's sometimes harder to come up with a useful heuristic lower bound on the remaining cost needed to complete a partial solution -- and the efficiency of A* depends on choosing an effective heuristic. But it doesn't mean it can't be applied. Any problem that can be solved on a computer can be solved using a breadth-first search, plus an array of flags indicating whether a state has been seen before; which is the same as A* with a heuristic lower bound that is always zero. (Of course, this is not the most efficient algorithm for solving many problems.)
You dont have to know the exact target endstate. It all comes down to the heuristic function, when it returns 0 you could assume to have found (at least) one of the valid endstates.
So during the a*, instead of checking if current_node == target_node, check if current_node.h() returns 0. If so, it should be infinitely close and/or overlapping the goal/endstate.