Lin–Kernighan heuristic

In combinatorial optimization, Lin–Kernighan is one of the best heuristics for solving the symmetric travelling salesman problem.[citation needed] It belongs to the class of local search algorithms, which take a tour (Hamiltonian cycle) as part of the input and attempt to improve it by searching in the neighbourhood of the given tour for one that is shorter, and upon finding one repeats the process from that new one, until encountering a local minimum. As in the case of the related 2-opt and 3-opt algorithms, the relevant measure of "distance" between two tours is the number of edges which are in one but not the other; new tours are built by reassembling pieces of the old tour in a different order, sometimes changing the direction in which a sub-tour is traversed. Lin–Kernighan is adaptive and has no fixed number of edges to replace at a step, but favours small numbers such as 2 or 3.

Derivation

edit

For a given instance   of the travelling salesman problem, tours are uniquely determined by their sets of edges, so we may as well encode them as such. In the main loop of the local search, we have a current tour   and are looking for new tour   such that the symmetric difference   is not too large and the length   of the new tour is less than the length   of the current tour. Since   is typically much smaller than   and  , it is convenient to consider the quantity

  — the gain of using   when switching from  

since  : how much longer the current tour   is than the new tour  . Naively  -opt can be regarded as examining all   with exactly   elements (  in   but not in  , and another   in   but not in  ) such that   is again a tour, looking for such a set which has  . It is however easier to do those tests in the opposite order: first search for plausible   with positive gain, and only second check if   is in fact a tour.

Define a trail in   to be alternating (with respect to  ) if its edges are alternatingly in   and not in  , respectively. Because the subgraphs   and   are  -regular, the subgraph   will have vertices of degree  ,  , and   only, and at each vertex there are as many incident edges from   as there are from  . Hence (essentially by Hierholzer's algorithm for finding Eulerian circuits) the graph   decomposes into closed alternating trails. Sets   that may satisfy   for some tour   may thus be found by enumerating closed alternating trails in  , even if not every closed alternating trail   makes   into a tour; it could alternatively turn out to be a disconnected  -regular subgraph.

Key idea

edit

Alternating trails (closed or open) are built by extending a shorter alternating trail, so when exploring the neighbourhood of the current tour  , one is exploring a search tree of alternating trails. The key idea of the Lin–Kernighan algorithm is to remove from this tree all alternating trails which have gain  . This does not prevent finding every closed trail with positive gain, thanks to the following lemma.

Lemma. If   are numbers such that  , then there is a cyclic permutation of these numbers such that all partial sums are positive as well, i.e., there is some   such that

  for all  .

For a closed alternating trail  , one may define   if   and   if  ; the sum   is then the gain  . Here the lemma implies that there for every closed alternating trail with positive gain exists at least one starting vertex   for which all the gains of the partial trails are positive as well, so   will be found when the search explores the branch of alternating trails starting at  . (Prior to that the search may have considered other subtrails of   starting at other vertices but backed out because some subtrail failed the positive gain constraint.) Reducing the number of branches to explore translates directly to a reduction in runtime, and the sooner a branch can be pruned, the better.

This yields the following algorithm for finding all closed, positive gain alternating trails in the graph.

State: a stack of triples  , where   is a vertex,   is the current number of edges in the trail, and   is the current trail gain.
  1. For all  , push   onto the stack.
  2. While the stack is nonempty:
    1. Pop   off the stack and let  . The current alternating trail is now  .
    2. If   is even then:
      For each   such that   (there are at most two of these), push   onto the stack.
    3. If instead   is odd then:
      1. If   then report   as a closed alternating trail with gain  .
      2. For each   such that   and   (there may be   of these, or there could be none), push   onto the stack.
  3. Stop

As an enumeration algorithm this is slightly flawed, because it may report the same trail multiple times, with different starting points, but Lin–Kernighan does not care because it mostly aborts the enumeration after finding the first hit. It should however be remarked that:

  • Lin–Kernighan is not satisfied with just having found a closed alternating trail   of positive gain, it additionally requires that   is a tour.
  • Lin–Kernighan also restricts the search in various ways, most obviously regarding the search depth (but not only in that way). The above unrestricted search still terminates because at   there is no longer any unpicked edge remaining in  , but that is far beyond what is practical to explore.
  • In most iterations one wishes to quickly find a better tour  , so rather than actually listing all siblings in the search tree before exploring the first of them, one may wish to generate these siblings lazily.

Basic Lin–Kernighan algorithm

edit

The basic form of the Lin–Kernighan algorithm not only does a local search counterpart of the above enumeration, but it also introduces two parameters that narrow the search.

  • The backtracking depth   is an upper bound on the length of the alternating trail after backtracking; beyond this depth, the algorithm explores at most one way of extending the alternating trail. Standard value is that  .
  • The infeasibility depth   is an alternating path length beyond which it begins to be required that closing the current trail (regardless of the gain of doing so) yields an exchange to a new tour. Standard value is that  .

Because there are   alternating trails of length  , and the final round of the algorithm may have to check all of them before concluding that the current tour is locally optimal, we get   (standard value  ) as a lower bound on the exponent of the algorithm complexity. Lin & Kernighan report   as an empirical exponent of   in the average overall running time for their algorithm, but other implementors have had trouble reproducing that result.[1] It appears unlikely that the worst-case running time is polynomial.[2]

In terms of a stack as above, the algorithm is:

Input: an instance   of the travelling salesman problem, and a tour  
Output: a locally optimal tour
Variables:
a stack of triples  , where   is a vertex,   is the current number of edges in the trail, and   is the current trail gain,
the sequence   of vertices in the current alternating trail,
the best set   of exchange edges found for current tour, and its corresponding gain  .
Initialise the stack to being empty.
Repeat
Set   and  .
For all  , push   onto the stack.
While the stack is nonempty:
Pop   off the stack and let  .
If   is even then
for each   such that  ,
push   onto the stack if:  , or   and   is a tour (Hamiltonicity check)
else (  is odd):
If  ,  , and   is a tour (Hamiltonicity check) then let   and  .
For each   such that   and  , push   onto the stack.
End if.
Let   be the top element on the stack (peek, not pop). If   then
if   then
set   (update current tour) and clear the stack.
else if   then
pop all elements   off the stack that have  
end if
end if
end while
until  .
Return  

The length of the alternating trails considered are thus not explicitly bounded, but beyond the backtracking depth   no more than one way of extending the current trail is considered, which in principle stops those explorations from raising the exponent in the runtime complexity.

Limitations

edit

The closed alternating trails found by the above method are all connected, but the symmetric difference   of two tours need not be, so in general this method of alternating trails cannot explore the full neighbourhood of a trail  . The literature on the Lin–Kernighan heuristic uses the term sequential exchanges for those that are described by a single alternating trail. The smallest non-sequential exchange would however replace 4 edges and consist of two cycles of 4 edges each (2 edges added, 2 removed), so it is long compared to the typical Lin–Kernighan exchange, and there are few of these compared to the full set of 4-edge exchanges.

In at least one implementation by Lin & Kernighan there was an extra final step considering such non-sequential exchanges of 4 edges before declaring a tour locally optimal, which would mean the tours produced are 4-opt unless one introduces further constraints on the search (which Lin and Kernighan in fact did). The literature is vague on exactly what is included in the Lin–Kernighan heuristic proper, and what constitutes further refinements.

For the asymmetric TSP, the idea of using positive gain alternating trails to find favourable exchanges is less useful, because there are fewer ways in which pieces of a tour can be rearranged to yield new tours when one may not reverse the orientation of a piece. Two pieces can only be patched together to reproduce the original tour. Three pieces can be patched together to form a different tour in one way only, and the corresponding alternating trail does not extend to a closed trail for rearranging four pieces into a new tour. To rearrange four pieces, one needs a non-sequential exchange.

Checking Hamiltonicity

edit

The Lin–Kernighan heuristic checks the validity of tour candidates   at two points: obviously when deciding whether a better tour has been found, but also as a constraint to descending in the search tree, as controlled via the infeasibility depth  . Concretely, at larger depths in the search a vertex   is only appended to the alternating trail if   is a tour. By design that set of edges constitutes a 2-factor in  , so what needs to be determined is whether that 2-factor consists of a single Hamiltonian cycle, or instead is made up of several cycles.

If naively posing this subproblem as giving a subroutine the set of   edges as input, one ends up with   as the time complexity for this check, since it is necessary to walk around the full tour before being able to determine that it is in fact a Hamiltonian cycle. That is too slow for the second usage of this test, which gets carried out for every alternating trail with more than   edges from  . If keeping track of more information, the test can instead be carried out in constant time.

A useful degree of freedom here is that one may choose the order in which step 2.3.2 iterates over all vertices; in particular, one may follow the known tour  . After picking   edges from  , the remaining subgraph   consists of   paths. The outcome of the Hamiltonicity test done when considering the  th edge   depends only on in which of these paths that   resides and whether   is before or after  . Hence it would be sufficient to examine   different cases as part of performing step 2.3.2 for  ; as far as   is concerned, the outcome of this test can be inherited information rather than something that has to be computed fresh.

References

edit
  1. ^ Melamed, I. I.; Sergeev, S. I.; Sigal, I. Kh. (1989). "The traveling salesman problem. Approximate algorithms". Avtomatika i Telemekhanika. 11: 3–26.
  2. ^ Papadimitriou, C. H. (1992). "The complexity of the Lin–Kernighan heuristic for the travelling salesman problem". SIAM Journal on Computing. 21 (3): 450–465. doi:10.1137/0221030.
edit