SubexponentialTime Algorithms for Maximum Independent Set in \(P_t\)Free and BroomFree Graphs
 149 Downloads
Abstract

If every component of H is a path, then dScattered Set on Hfree graphs with n vertices and m edges can be solved in time \(2^{\mathcal {O}(V(H)\sqrt{n+m}\log (n+m))}\), even if d is part of the input.

Otherwise, assuming the ExponentialTime Hypothesis (ETH), there is no \(2^{o(n+m)}\)time algorithm for dScattered Set for any fixed \(d\ge 3\) on Hfree graphs with nvertices and medges.
Keywords
Independent set Subexponential algorithms Approximation Scattered set Hfree graphs1 Introduction
There are some problems in discrete optimization that can be considered fundamental. The Maximum Independent Set problem (MIS, for short) is one of them. It takes a graph G as input, and asks for the maximum number \(\alpha (G)\) of mutually nonadjacent (i.e., independent) vertices in G. On unrestricted input, it is not only NPhard (its decision version “Is \(\alpha (G)\ge k\)?” being NPcomplete), but APXhard as well, and, in fact, not even approximable within \(\mathcal {O}(n^{1\varepsilon })\) in polynomial time for any \(\varepsilon >0\) unless \(\hbox {P}=\hbox {NP}\), as proved by Zuckerman [30]. For this reason, those classes of graphs on which MIS becomes tractable are of definite interest. One direction of this area is to study the complexity of MIS on Hfree graphs, that is, on graphs not containing any induced subgraph isomorphic to a given graph H.
For the majority of the graphs H, we know a negative answer on the complexity question. It is easy to see that if \(G'\) is obtained from G by subdividing each edge with 2t new vertices, then \(\alpha (G')=\alpha (G)+tE(G)\) holds. This can be used to show that MIS is NPhard on Hfree graphs whenever H is not a forest, and also if H contains a tree component with at least two vertices of degree larger than 2 (first observed in [2], see, e.g., [20]). As MIS is known to be NPhard on graphs of maximum degree at most 3, the case when H contains a vertex of degree at least 4 is also NPhard.
The above observations do not cover the case when every component of H is either a path, or a tree with exactly one degree3 vertex c with three paths of arbitrary lengths starting from c. There are no further unsolved classes but even this collection means infinitely many cases. For decades, on these graphs H only partial results have been obtained, proving polynomialtime solvability in some cases. A classical algorithm of Minty [24] and its corrected form by Sbihi [27] solved the problem when H is a claw (3 paths of length 1 in the model above). This happened in 1980. Much later, in 2004, Alekseev [3] generalized this result by an algorithm for H isomorphic to a fork (2 paths of length 1 and one path of length 2); a weighted counterpart of this result has been proven by Lozin and Milanic [21].
The seemingly easy case of \(P_t\)free graphs is poorly understood (where \(P_t\) is the path on t vertices). MIS on \(P_t\)free graphs is not known to be NPhard for any t; for all we know, it could be polynomialtime solvable for every fixed \(t\ge 1\). \(P_4\)free graphs (also known as cographs) have a very simple structure, which can be used to solve MIS with a lineartime recursion, but this does not generalize to \(P_t\)free graphs for larger t. In 2010, it was a breakthrough when Randerath and Schiermeyer [25] stated that MIS on \(P_5\)free graphs was solvable in subexponential time, more precisely within \(\mathcal {O}(C^{n^{1\varepsilon }})\) for any constants \(C>1\) and \(\varepsilon <1/4\). Designing an algorithm based on deep results, Lokshtanov et al. [20] finally proved that MIS is polynomialtime solvable on \(P_5\)free graphs. More recently, a quasipolynomial (\(n^{\log ^{\mathcal {O}(1)} n}\)time) algorithm was found for \(P_6\)free graphs [19] and finally a polynomialtime algorithm for \(P_6\)free graphs was announced [14]. For \(P_7\)free graphs, a partial result is known: MWIS is polynomialtime solvable if we additionally exclude a triangle [8]. A related result of Lozin and Mosca [22] asserts that MWIS is polynomialtime solvable on \((K_2+\mathrm {claw})\)free graphs.
We explore MIS and some variants on Hfree graphs from the viewpoint of subexponentialtime algorithms in this work. That is, instead of aiming for algorithms with running time \(n^{\mathcal {O}(1)}\) on nvertex graphs, we ask if \(2^{o(n)}\) algorithms are possible. Very recently, Brause [9] and independently the conference version of this paper [4] observed that the subexponential algorithm of Randerath and Schiermeyer [25] can be generalized to arbitrary fixed \(t\ge 5\) with running time roughly \(2^{\mathcal {O}(n^{11/t})}\). Our first result shows a significantly improved subexponentialtime algorithm for every t.
Theorem 1
For every fixed \(t\ge 5\), MIS on nvertex \(P_t\)free graphs can be solved in subexponential time, namely, it can be solved by a \(2^{\mathcal {O}(\sqrt{n\log n})}\)time algorithm.
The algorithm is based on the combination of two ideas. First, we generalize the observation of Randerath and Schiermeyer [25] stating that in a large connected \(P_5\)free graph there exists a highdegree vertex. Namely, we prove that such a vertex always exists in a large connected \(P_t\)free graph for general \(t\ge 5\) and it can be used for efficient branching. Next we prove the combinatorial result that a \(P_t\)free graph of maximum degree \(\varDelta \) has treewidth \(\mathcal {O}(t\varDelta )\); the proof is inspired by Gyárfás’ proof of the \(\chi \)boundedness of \(P_t\)free graphs [15]. Thus if the maximum degree drops below a certain threshold during the branching procedure, then we can use standard algorithmic techniques exploiting bounded treewidth.
While our algorithm works for \(P_t\)free graphs with arbitrary large t, it does not seem to be extendable to Hfree graphs where H is the subdivision of a \(K_{1,3}\). Hence, the existence of subexponentialtime algorithms on such graphs remains an open question. However, we are able to give a subexponentialtime constantfactor approximation algorithm for the case when H is a (d, t)broom. A (d, t)broom\(B_{d,t}\) is a graph consisting of a path \(P_t\) and d additional vertices of degree one, all adjacent to one of the endpoints of the path. In other words, \(B_{d,t}\) is a star \(K_{1,d+1}\) with one of the edges subdivided to make it a path with t vertices. For \(d=2\), we obtain the generalized forks and \(t=3\), \(d=2\) yields the traditional fork. We prove the following theorem; here d and t are considered constants, hidden in the big\(\mathcal {O}\) notation.
Theorem 2
Let \(d,t \ge 2\) be fixed integers. One can find a dapproximation to Maximum Independent Set on an nvertex \(B_{d,t}\)free graph G in time \(2^{\mathcal {O}(n^{3/4} \log n)}\).
Let us remark that on \(K_{1,d+1}\)free graphs, a folklore lineartime (and very simple) dapproximation algorithm exists for Maximum Independent Set ; better d / 2approximation algorithms also exist [5, 6, 16, 29]. On forkfree graphs, Independent Set can be solved in polynomial time [3]. For general graphs, we do not expect that a constantfactor approximation can be obtained in subexponential time for the problem. Strong evidence for this was given by Chalermsook et al. [10], who showed that the existence of such an algorithm would violate the ExponentialTime Hypothesis (ETH) of Impagliazzo, Paturi, and Zane, which can be informally stated as nvariable 3SAT cannot be solved in \(2^{o(n)}\) time (see [11, 17, 18]).
Scattered Set (also known under other names such as dispersion or distanced independent set [1, 7, 12, 23, 26, 28]) is the natural generalization of MIS where the vertices of the solution are required to be at distance at least d from each other; the size of the largest such set will be denoted by \(\alpha _d(G)\). We can consider with d being part of the input, or assume that \(d\ge 2\) is a fixed constant, in which case we call the problem dScattered Set. Clearly, MIS is exactly the same as 2Scattered Set. Despite its similarity to MIS, the branching algorithm of Theorem 1 cannot be generalized: we give evidence that there is no subexponentialtime algorithm for 3Scattered Set on \(P_5\)free graphs.
Theorem 3
Assuming the ETH, there is no \(2^{o(n)}\)time algorithm for dScattered Set with \(d=3\) on \(P_5\)free graphs with n vertices.
In light of the negative result of Theorem 3, we slightly change our objective by aiming for an algorithm that is subexponential in the size of the input, that is, in the total number of vertices and edges of the graph G. As the number of edges of G can be up to quadratic in the number of vertices, this is a weaker goal: an algorithm that is subexponential in the number of edges is not necessarily subexponential in the number of vertices. We give a complete characterization when such algorithms are possible for Scattered Set.
Theorem 4
 1.
If every component of H is a path, then dScattered Set on Hfree graphs with n vertices and m edges can be solved in time \(2^{\mathcal {O}(V(H)\sqrt{n+m}\log (n+m))}\), even if d is part of the input.
 2.
Otherwise, assuming the ETH, there is no \(2^{o(n+m)}\)time algorithm for dScattered Set for any fixed \(d\ge 3\) on Hfree graphs with nvertices and medges.
The algorithmic side of Theorem 4 is based on the combinatorial observation that the treewidth of \(P_t\)free graphs is sublinear in the number of edges, which means that standard algorithms on boundedtreewidth graphs can be invoked to solve the problem in time subexponential in the number of edges. It has not escaped our notice that this approach is completely generic and could be used for many other problems (e.g., Hamiltonian Cycle, 3Coloring, and so on), where \(2^{\mathcal {O}(t)}\cdot n^{\mathcal {O}(1)}\) or perhaps \(2^{t\cdot \log ^{\mathcal {O}(1)} t}\cdot n^{\mathcal {O}(1)}\)time algorithms are known on graphs of treewidth t. For the lowerbound part of Theorem 4, we need to examine only two cases: clawfree graphs and \(C_t\)free graphs (where \(C_t\) is the cycle on t vertices); the other cases then follow immediately.
The paper is organized as follows. Section 2 introduces basic notation and contains some technical tools for bounding the running time of recursive algorithms. Section 3 contains the combinatorial results that allow us to bound the treewidth of \(P_t\)free graphs. The algorithmic results for Maximum Independent Set (Theorems 1 and 2) appear in Sect. 4. The upper and lower bounds for dScattered Set, which together prove Theorem 4, are proved in Sect. 5.
2 Preliminaries
Simple undirected graphs are investigated here throughout. The vertex set of graph G will be denoted by V(G), the edge set by E(G). The notation \(d_G(x,y)\) for distance, G[X] for the subgraph induced by the vertex set X, will have the usual meaning, similarly as \(N_G[X]\) and \(N_G(X)\) for the closed and open neighborhood respectively of vertex set X in G. \(\varDelta (G)\) is the maximum degree in G. For a vertex set X in G, \(GX\) means the induced subgraph \(H:=G[VX]\). \(P_t\) (\(C_t\)) is the chordless path (cycle) on t vertices. Finally, a graph is Hfree if it does not contain H as an induced subgraph.
A distanced (dscattered) set in a graph G is a vertex set \(S\subseteq V(G)\) such that for every pair of vertices in S, the distance between them is at least d in the graph. For \(d=2\), we obtain the traditional notion of independent set (stable set). For \(d>c\), a distanced set is a distancec set as well, for example, for \(d\ge 2\), any distanced set is an independent set.
The algorithmic problem Maximum Weight Independent Set is the problem of maximizing the sum of the weights in an independent set of a graph with nonnegative vertex weights w. The maximum is denoted by \(\alpha _w(G)\). For a weight w function that has value 1 everywhere, we obtain the usual problem Maximum Independent Set (MIS) with maximum \(\alpha (G)\).
An algorithm A is subexponential in parameter \(p>1\) if the number of steps executed by A is a subexponential function of the parameter p. We will use here this notion for graphs, mostly in the following cases: p is the number n of vertices, the number m of edges, or \(p=n+m\) (which is considered to be the size of the input generally). Several different definitions are used in the literature under the name subexponential function. Each of them means some condition: this function (with variable \(p>1\), called the parameter) may not be larger than some bound, depending on p. Here we use two versions, where the bound is of type exp(o(p)) and \(exp(p^{1\epsilon })\) respectively, with some \(\epsilon >0\). (Clearly, the second one is the more strict.) Throughout the paper, w we mean. A problem \(\varPi \) is subexponential if there exists some subexponential algorithm solving \(\varPi \).
2.1 Time Analysis of Recursive Algorithms
To formally reason about time complexities, we will need the following technical lemma.
Lemma 1
We will use Lemma 1 as a shortcut to argue about time complexities of our branching algorithms; let us now briefly explain its intuition. The function T(n) will be the running time bound of the discussed algorithm. The term \(2^{cn\log n / \varDelta (n)}\) in (1) corresponds to a processing time at a single step of the algorithm; note that this is at least polynomial in n as \(\varDelta (n) \le n\). The terms in the \(\max \) in (1) are different branching options chosen by the algorithm. The first one, S(n), is a subcall to a different procedure, such as bounded treewidth subroutine. The second one, \(T(n1) + T(n\lceil \varDelta (n) \rceil )\), corresponds to a twoway branching on a single vertex of degree at least \(\varDelta (n)\). The last one corresponds to an exhaustive branching on a set \(X \subseteq V(G)\) of size k, such that every connected component of \(GX\) has at most \(nk\varDelta (n)\) vertices.
Proof of Lemma 1
For notational convenience, it will be easier to assume that the functions S and T is defined on the whole halfline \(\mathbb {R}_{\ge 0}\) with \(S(x) = S(\lfloor x \rfloor )\) and \(T(x) = T(\lfloor x \rfloor )\).
We now show the following.
Claim
Consider a sequence \(n_0 = n\) and \(n_{i+1} = n_i  \varDelta (n_i)\). Then \(n_i = \mathcal {O}(1)\) for \(i = \mathcal {O}(n / \varDelta (n))\). Here, the big\(\mathcal {O}\)notation hides constants depending on \(\gamma \).
Proof
3 Gyárfás’ PathGrowing Argument
The main (technical but useful) result of this section is the following adaptation of Gyárfás’ proof that \(P_t\)free graphs are \(\chi \)bounded [15].
Lemma 2
Let \(t \ge 2\) be an integer, G be a connected graph with a distinguished vertex \(v_0 \in V(G)\) and maximum degree at most \(\varDelta \), such that G does not contain an induced path \(P_t\) with one endpoint in \(v_0\). Then, for every weight function \(w : V(G) \rightarrow \mathbb {Z}_{\ge 0}\), there exists a set \(X \subseteq V(G)\) of size at most \((t1)\varDelta +1\) such that every connected component C of \(GX\) satisfies \(w(C) \le w(V(G))/2\). Furthermore, such a set X can be found in polynomial time.
Proof
In what follows, a connected component C of an induced subgraph H of G is big if \(w(C) > w(V(G))/2\). Note that there can be at most one big connected component in any induced subgraph of G.
If \(G\{v_0\}\) does not contain a big component, we can set \(X=\{v_0\}\). Otherwise, let \(A_0 = \{v_0\}\) and \(B_0\) be the big component of \(GA_0\). As G is connected, every component of \(GA_0\) is adjacent to \(A_0\), thus \(v_0\in N(B_0)\) holds. We will inductively define vertices \(v_1,v_2,v_3,\ldots \) such that \(v_0,v_1,v_2,\ldots \) induce a path in G.
Given vertices \(v_0,v_1,v_2,\ldots ,v_i\), we define sets \(A_{i+1}\) and \(B_{i+1}\) as follows. We set \(A_{i+1} = N_G[v_0,v_1,\ldots ,v_i]\). If \(GA_{i+1}\) does not contain a big connected component, we stop the construction. Otherwise, we set \(B_{i+1}\) to be the big connected component of \(GA_{i+1}\). During the process we maintain the invariant that \(B_i\) is the big component of \(GA_i\) and that \(v_i \in N(B_i)\). Note that this is true for \(i=0\) by the choice of \(A_0\) and \(B_0\).
It remains to show how to choose \(v_{i+1}\), given vertices \(v_0,v_1,\ldots ,v_i\) and sets \(A_{i+1}\) and \(B_{i+1}\). Note that \(A_{i+1} = A_i \cup N_G[v_i]\) and \(v_i \in N(B_i)\), so \(B_{i+1}\) is the big connected component of \(G[(B_i {\setminus } N_G(v_i))]\). Consequently, we can choose some \(v_{i+1} \in B_i \cap N_G(B_{i+1}) \cap N_G(v_i)\) that satisfies all the desired properties.
For the algorithmic claim, note that the entire proof can be made algorithmic in a straightforward manner. \(\square \)
A balanced separator of a set \(W \subseteq V(G)\) in a graph G is a set \(X \subseteq V(G)\) such that every connected component C of \(GX\) satisfies \(W \cap C \le W/2\). Note that Lemma 2 implies that in a connected \(P_t\)free graph G of maximum degree \(\varDelta \) for every \(W \subseteq V(G)\) there exists a balanced separator of W of size at most \((t1)\varDelta +1\), and such a balanced separator can be found in polynomial time. It is well known that existence of such small balanced separators bounds the treewidth of the graph [13, Theorem 11.17(2)].
Theorem 5
[13] Let G be a graph and \(k \ge 1\). If for every \(W \subseteq V(G)\) of size \(2k+3\) there exists a balanced separator of W of cardinality at most \(k+1\), then G has treewidth at most \(3k+3\).
Theorem 5 applied to \(k = (t1)\varDelta \) implies that a connected \(P_t\)free graph of maximum degree \(\varDelta \) has treewidth at most \(3(t1)\varDelta +3\).
Algorithmically, it is also a standard consequence of Lemma 2 that a tree decomposition of width \(\mathcal {O}(t\varDelta )\) can be obtained in polynomial time. What needs to be observed is that standard 4approximation algorithms for treewidth, which run in time exponential in treewidth, can be made to run in polynomial time if we are given a polynomialtime subroutine for finding the separator X as in Lemma 2. This is immediate from the proof of Theorem 11.17 in [13], but, for completeness, we sketch the proof here.
Corollary 1
A \(P_t\)free graph with maximum degree \(\varDelta \) has treewidth \(\mathcal {O}(t\varDelta )\). Furthermore, a tree decomposition of this width can be computed in polynomial time.
Proof
We follow standard constant approximation algorithm for treewidth, as described in [11, Section 7.6]. This algorithm, given a graph G and an integer k, either correctly concludes that \(\mathrm {tw}(G) > k\) or computes a tree decomposition of G of width at most \(4k+4\).
Let G be a \(P_t\)free graph with maximum degree at most \(\varDelta \). We may assume that G is connected, otherwise we can handle the connected components separately. Let us start by setting \(k := (t1)\varDelta \) so that any application of Lemma 2 gives a set of size at most \(k+1\).
 1.
\(S \le 3k+4\) and \(W {\setminus } S \ne \emptyset \);
 2.
both G[W] and \(G[W {\setminus } S]\) are connected;
 3.
\(S = N_G(W {\setminus } S)\).
The construction of \(\widehat{S}\) is trivial for \(S < 3k+4\), as we can take \(\widehat{S} = S \cup \{v\}\) for an arbitrary \(v \in W {\setminus } S\). The crucial step happens for sets S of size exactly \(3k+4\). Instead of the exponential search of [11, Section 7.6], we invoke Lemma 2 on the graph G[W] and a function \(w:W \rightarrow \{0,1\}\) that puts \(w(v) = 1\) if and only if \(v \in S\). The lemma returns a set \(X \subseteq W\) of size at most \(k+1\) such that every connected component C of \(G[W {\setminus } X]\) contains at most \(3k/2+2\) vertices of S. Since \(G[W {\setminus } S]\) is connected and \((3k/2+2) + (k+1) < 3k+4\), we cannot have \(X \subseteq S\). Consequently, \(\widehat{S} := S \cup X\) satisfies all the requirements.
The algorithm of [11, Section 7.6] returns that \(\mathrm {tw}(G) > k\) only if at some step it encounters pair (W, S) for which it cannot construct the set \(\widehat{S}\). However, our method of constructing \(\widehat{S}\) works for every choice of (W, S), and executes in polynomial time. Consequently, the modified algorithm of [11, Section 7.6] always computes a tree decomposition of width at most \(4k+4 = \mathcal {O}(t\varDelta )\) in polynomial time, as desired. \(\square \)
4 Subexponential Algorithms Based on the PathGrowing Argument
The goal of this section is to use Corollary 2.1 to prove Theorems 1 and 2 stated in the Introduction.
4.1 Independent Set on Graphs Without Long Paths
We first prove the following statement, which implies Theorem 1.
Theorem 6
The MaximumWeight Independent Set problem on an nvertex \(P_t\)free graph can be solved in time \(2^{\mathcal {O}(\sqrt{tn\log n})}\).
Proof
Let G be an nvertex \(P_t\)free graph. We set a threshold \(\varDelta = \varDelta (n) := \sqrt{\frac{n \log (n+1)}{t}}\). If the maximum degree of G is at most \(\varDelta \), we invoke Corollary 1 to obtain a tree decomposition of G of width \(\mathcal {O}(t\varDelta ) = \mathcal {O}(\sqrt{tn\log n})\). By standard dynamic programming techniques, on graphs of bounded treewidth (cf. [11]), adapted to vertexweighted graphs, we solve MaximumWeight Independent Set on G in time \(2^{\mathcal {O}(\sqrt{tn\log n})}\).
4.2 Approximation on BroomFree Graphs
We now extend the argumentation in Theorem 6 to (d, t)brooms—however, this time we are able to obtain only an approximation algorithm. Recall that a (d, t)broom \(B_{d,t}\) is a graph consisting of a path \(P_t\) and d additional vertices of degree one, all adjacent to one of the endpoints of the path.
We now prove Theorem 2 from the introduction.
Proof of Theorem 2
Let \(\varDelta (n) = \frac{1}{2dt} \cdot n^{1/4}\); note that such a definition fits the prerequisites of \(\varDelta (n)\) for Lemma 1. In the complexity analysis, we will use Lemma 1 with this \(\varDelta (n)\) and without any function S(n); this will give the promised running time bound. In what follows, whenever we execute a branching step of the algorithm we argue that it fits into one of the subcases of the \(\max \) in (1) of Lemma 1.
As in the proof of Theorem 6, as long as there exists a vertex in G of degree larger than \(\varDelta \), we can branch on such a vertex v: in one subcase, we consider independent sets not containing v (and thus delete v from G), in the other subcase, we consider independent sets containing v (and thus delete N(v) from G). Such a branching step can be conducted in polynomial time, and fits in the second subcase of \(\max \) in (1). Thus, we can assume henceforth that the maximum degree of G is at most \(\varDelta \).
We also assume that G is connected and \(n > (2dt)^4\), as otherwise we can consider every connected component independently and/or solve the problem by bruteforce.
Later, we will also need a more general branching step. If, in the course of the analysis, we identify a set \(X \subseteq V(G)\) such that every connected component of \(GX\) has size at most \(n  \frac{Xn^{1/4}}{2dt}\), then we can exhaustively branch on all vertices of X and independently resolve all connected components of the remaining graph. Such a branching fits into the last case of the \(\max \) in (1), and hence it again leads to the desired time bound \(2^{\mathcal {O}(n^{3/4} \log n)}\) by Lemma 1.
We start with greedily constructing a set \(A_0\) with the following properties: \(G[A_0]\) is connected and \(n^{1/2} \le N[A_0] \le n^{1/2} + \varDelta \). We start with \(A_0\) being a single arbitrary vertex and, as long as \(N[A_0] < n^{1/2}\), we add an arbitrary vertex of \(N(A_0)\) to \(A_0\) and continue. Since G is connected, the process ends when \(N[A_0] \ge n^{1/2}\); since the maximum degree of G is at most \(\varDelta \), we have \(N[A_0] \le n^{1/2} + \varDelta < 2n^{1/2}\).
Let B be the vertex set of the largest connected component of \(GN[A_0]\). If \(B < n  n^{3/4}\), we exhaustively branch on \(X := N[A_0]\), as X is of size at most \(2n^{1/2}\), but every connected component of \(GX\) is of size at most \(n  n^{3/4} \le n \frac{1}{2} X n^{1/4}\). Hence, we are left with the case \(B > n  n^{3/4}\).
Let \(S = N(B)\). Note that \(A_0\) is disjoint from N[B]. Let \(A_1\) be the connected component of \(GS\) that contains \(A_0\). Since \(S \subseteq N(A_0)\), we have that \(N[A_1] \supseteq N[A_0]\); in particular, \(N[A_1] \ge n^{1/2}\) while, as \(B > nn^{3/4}\), we have \(N[A_1] \le n^{3/4}\). Furthermore, since \(S \subseteq N(A_0)\) and \(A_0 \subseteq A_1\), we have \(N(A_1) = S\).
Consider now the following case: there exists \(v \in S\) such that \(N(v) \cap B\) contains an independent set L of size d. Observe that such a vertex v can be found by an exhaustive search in time \(n^{d+\mathcal {O}(1)}\).
Observe that G[D] does not contain a path \(P_t\) with one endpoint in v, as such a path, together with the set L, would induce a \(B_{d,t}\) in G. Consequently, we can apply Lemma 2 to the graph G[D] with the vertex \(v_0=v\) and uniform weight \(w(u) = 1\) for every \(u \in D\), obtaining a set \(X_D \subseteq D\) of size \(X_D \le (t1)\varDelta + 1 \le \frac{1}{2} n^{1/4}\) such that every connected component of \(G[D {\setminus } X]\) has size at most n / 2. We branch exhaustively on the set \(X = X_D \cup (N[L] {\setminus } \{v\})\): this set is of size at most \(n^{1/4}\), while every connected component of \(GX\) is of size at most n / 2 due to the properties of \(X_D\) and the fact that \(D \ge nn^{1/2}\). This finishes the description of the algorithm in the case when there exists \(v \in S\) and an independent set \(L \subseteq N(v) \cap B\) of size d.
We are left with the complementary case, where for every \(v \in S\), the maximum independent set in \(N(v) \cap B\) is of size less than d. We perform the following operation: by exhaustive search, we find a maximum independent set \(I_A\) in \(GB\) and greedily take it to the solution; that is, recurse on \(GN[I_A]\) and return the union of \(I_A\) and the independent set found by the recursive call in \(GN[I_A]\). Since \(B > nn^{3/4}\), the exhaustive search runs in \(2^{n^{3/4}} n^{\mathcal {O}(1)}\) time, fitting the first summand of the right hand side in (1). As a result, the graph reduces by at least one vertex, and hence the remaining running time of the algorithm fits into the second case of the \(\max \) in (1). This gives the promised running time bound. It remains to argue about the approximation ratio; to this end, it suffices to show the following claim.
Claim
If I is a maximum independent set in G and \(I'\) is a maximum independent set in \(GN[I_A]\), then \(I  I' \le dI_A\).
Proof
Let \(J = I {\setminus } N[I_A]\). Clearly, J is an independent set in \(GN[I_A]\), and thus \(J \le I'\). It suffices to show that \(I  J \le dI_A\), that is, \(I \cap N[I_A] \le dI_A\).
This finishes the proof of Theorem 2. \(\square \)
5 Scattered Set
We prove Theorem 4 in this section. The algorithm for Scattered Set for \(P_t\)free graphs hinges on the following combinatorial bound.
Lemma 3
For every \(t\ge 2\) and for every \(P_t\)free graph with m edges, we have that G has treewidth \(\mathcal {O}(t\sqrt{m})\).
Proof
Let X be the set of vertices of G with degree at least \(\sqrt{m}\). The sum of the degrees of the vertices in X is at most 2m, hence we have \(X\le 2m/\sqrt{m}=2\sqrt{m}\). By the definition of X, the graph \(GX\) has maximum degree less than \(\sqrt{m}\). Thus by Corollary 1, the treewidth of \(GX\) is \(\mathcal {O}(t\sqrt{m})\). As removing a vertex can decrease treewidth at most by one, it follows that G has treewidth at most \(\mathcal {O}(t\sqrt{m})+X=\mathcal {O}(t\sqrt{m})\). \(\square \)
5.1 Lower Bounds for Scattered Set
A standard consequence of the ETH and the socalled Sparsification Lemma is that there is no subexponentialtime algorithm for MIS even on graphs of bounded degree (see, e.g., [11]):
Theorem 7
Assuming the ETH, there is no \(2^{o(n)}\)time algorithm for MIS on nvertex graphs of maximum degree 3.
A very simple reduction can reduce MIS to 3Scattered Set for \(P_5\)free graphs, showing that, assuming the ETH, there is no algorithm subexponential in the number of vertices for the latter problem. This proves Theorem 3 stated in the Introduction.
Proof of Theorem 3
Given an nvertex medge graph G with maximum degree 3 and an integer k, we construct a \(P_5\)free graph \(G'\) with \(n+m=\mathcal {O}(n)\) vertices such that \(\alpha (G)=\alpha _3(G')\). This reduction proves that a \(2^{o(n)}\)time algorithm for 3Scattered Set could be used to obtain a \(2^{o(n)}\)time algorithm for MIS on graphs of maximum degree 3, and this would violate the ETH by Theorem 7.
We may assume that G has no isolated vertices. The graph \(G'\) contains one vertex for each vertex of G and additionally one vertex for each edge of G. The m vertices of \(G'\) representing the edges of G form a clique. Moreover, if the endpoints of an edge \(e\in E(G)\) are \(u,v\in V(G)\), then the vertex of \(G'\) representing e is connected with the vertices of \(G'\) representing u and v. This completes the construction of \(G'\). It is easy to see that \(G'\) is \(P_5\)free: an induced path of \(G'\) can contain at most two vertices of the clique corresponding to E(G) and the vertices of \(G'\) corresponding to the vertices of G form an independent set.
If S is an independent set of G, then we claim that the corresponding vertices of \(G'\) are at distance at least 3 from each other. Indeed, no two such vertices have a common neighbor: if \(u,v\in S\) and the corresponding two vertices in \(G'\) have a common neighbor, then this common neighbor represents an edge e of G whose endpoints are u and v, violating the assumption that S is independent. Conversely, suppose that \(S'\subseteq V(G')\) is a set of k vertices with pairwise distance at least 3 in \(G'\). If \(k\ge 2\), then all these vertices represent vertices of G: observe that for every edge e of G, the vertex of \(G'\) representing e is at distance at most 2 from every other (nonisolated) vertex of \(G'\). We claim that \(S'\) corresponds to an independent set of G. Indeed, if \(u,v\in S'\) and there is an edge e in \(G'\) with endpoints u and v, then the vertex of \(G'\) representing e is a common neighbor of u and v, a contradiction. \(\square \)
Next we give negative results on the existence of algorithms for Scattered Set that have running time subexponential in the number of edges. To rule out such algorithms, we construct instances that have bounded degree: then being subexponential in the number of vertices or the number of edges are the same. We consider first clawfree graphs. The key insight here is that Scattered Set with \(d=3\) in line graphs (which are clawfree) is essentially the Induced Matching problem, for which it is easy to prove hardness results.
Theorem 8
Assuming the ETH, dScattered Set does not have a \(2^{o(n)}\) algorithm on nvertex clawfree graphs of maximum degree 6 for any fixed \(d\ge 3\).
Proof
Given an nvertex graph G with maximum degree 3, we construct a clawfree graph \(G'\) with \(\mathcal {O}(dn)\) vertices and maximum degree 4 such that \(\alpha _d(G')=\alpha (G)\). Then by Theorem 7, a \(2^{o(n)}\)time algorithm for dScattered Set for nvertex clawfree graphs of maximum degree 4 would violate the ETH.
The construction is slightly different based on the parity of d; let us first consider the case when d is odd. Let us construct the graph \(G^+\) by attaching a path \(Q_v\) of \(\ell =(d1)/2\) edges to each vertex \(v\in V(G)\); let us denote by \(e_{v,1}\), \(\dots \), \(e_{v,\ell }\) the edges of this path such that \(e_{v,1}\) is incident with v. The graph \(G'\) is defined as the line graph of \(G^+\), that is, each vertex of \(G'\) represents an edge of \(G^+\) and two vertices of \(G'\) are adjacent if the corresponding two vertices share an endpoint. It is well known that line graphs are clawfree. As \(G^+\) has \(\mathcal {O}(dn)\) edges and maximum degree 4 (recall that G has maximum degree 3), the line graph \(G'\) has maximum degree 6 with \(\mathcal {O}(dn)\) vertices an edges. Thus an algorithm for Scattered Set with running time \(2^{o(n)}\) on nvertex clawfree graphs of maximum degree 3 could be used to solve MIS on nvertex graphs with maximum degree 3 in time \(2^{o(n)}\), contradicting the ETH.
If there is an independent set S of size k in G, then we claim that the set \(S'=\{e_{v,\ell }\mid v\in S\}\) is a dscattered set of size k in \(G'\). To see this, suppose for a contradiction that there are two vertices \(u,v\in S\) such that the vertices of \(G'\) representing \(e_{u,\ell }\) and \(e_{v,\ell }\) are at distance at most \(d1\) from each other. This implies that there is a path in \(G^+\) that has at most d edges and whose first and last edges are \(e_{u,\ell }\) and \(e_{v,\ell }\), respectively. However, such a path would need to contain all the \(\ell \) edges of path \(Q_u\) and all the \(\ell \) edges of \(Q_v\), hence it can contain at most \(d2\ell =1\) edges outside these two paths. But u and v are not adjacent in \(G^+\) by assumption, hence more than one edge is needed to complete \(Q_u\) and \(Q_v\) to a path, a contradiction.
Conversely, let \(S'\) be a distanced scattered set in \(G'\), which corresponds to a set \(S^+\) of edges in \(G^+\). Observe that for any \(v\in V(G)\), at most one edge of \(S^+\) can be incident to the vertices of \(Q_v\): otherwise, the corresponding two vertices in the line graph \(G'\) would have distance at most \(\ell <d\). It is easy to see that if \(S^+\) contains an edge incident to a vertex of \(Q_v\), then we can always replace this edge with \(e_{v,\ell }\), as this can only move it farther away from the other edges of \(S^+\). Thus we may assume that every edge of \(S^+\) is of the form \(e_{v,\ell }\). Let us construct the set \(S=\{v\mid e_{v,\ell }\in S^+\}\), which has size exactly k. Then S is independent in G: if \(u,v\in S\) are adjacent in G, then there is a path of \(2\ell +1=d\) edges in \(G^+\) whose first an last edges are \(e_{v,\ell }\) and \(e_{u,\ell }\), respectively, hence the vertices of \(G'\) corresponding to them have distance at most \(d1\).
If \(d\ge 4\) is even, then the proof is similar, but we obtain the graph \(G^+\) by first subdividing each edge and attaching paths of length \(\ell =d/21\) to each original vertex. The proof proceeds in a similar way: if u and v are adjacent in G, then \(G^+\) has a path of \(2\ell +2=d\) edges whose first and last edges are \(e_{v,\ell }\) and \(e_{u,\ell }\), respectively, hence the vertices of \(G'\) corresponding to them have distance at most \(d1\). \(\square \)
There is a wellknown and easy way of proving hardness of MIS on graphs with large girth: subdividing edges increases girth and the size of the largest independent set changes in a controlled way.
Lemma 4
If there is an \(2^{o(n)}\)time algorithm for MIS on nvertex graphs of maximum degree 3 and girth more than g for any fixed \(g>0\), then the ETH fails.
Proof
Let g be a fixed constant and let G be a simple graph with n vertices, m edges, and maximum degree 3 (hence \(m=\mathcal {O}(n)\)). We construct a graph \(G'\) by subdividing each edge with 2g new vertices. We have that \(G'\) has \(n'=\mathcal {O}(n+gm)=\mathcal {O}(n)\) vertices, maximum degree 3, and girth at least \(3(2g+1)>g\). It is known and easy to show that subdividing the edges this way increases the size of the maximum independent set exactly by gm. Thus a \(2^{o(n')}\) time algorithm for \(n'\)vertex graphs of maximum degree 3 and girth at least g could be used to give a \(2^{o(n)}\)time algorithm for nvertex graphs of maximum degree 3, hence the ETH would fail by Theorem 7. \(\square \)
We use the lower bound of Lemma 4 to prove lower bounds for Scattered Set on \(C_t\)free graphs.
Theorem 9
Assuming the ETH, dScattered Set does not have a \(2^{o(n)}\) algorithm on nvertex \(C_t\)free graphs with maximum degree 3 for any fixed \(t\ge 3\) and \(d\ge 2\).
Proof
Let G be an nvertex medge graph of maximum degree 3 and girth more than t. We construct a graph \(G'\) the following way: we subdivide each edge of G with \(d2\) new vertices to create a path of length \(d1\), and attach a path of length \(d1\) to each of the \((d2)m=\mathcal {O}(dn)\) new vertices created. The resulting graph has maximum degree 3, \(\mathcal {O}(d^2n)\) vertices and edges, and girth more than \((d1)t\) (hence it is \(C_t\)free). We claim that \(\alpha _d(G')=\alpha (G)+m(d2)\) holds. This means that an \(2^{o(n')}\)time algorithm for Scattered Set\(n'\)vertex \(C_t\)free graphs with maximum degree 3 would give a \(2^{o(n)}\)time algorithm for nvertex graphs of maximum degree 3 and girth more than t and this would violate the ETH by Lemma 4.
To see that \(\alpha _d(G')=\alpha (G)+m(d2)\) holds, consider first an independent set S of G. When constructing \(G'\), we attached \(m(d2)\) paths of length \(d1\). Let \(S'\) contain the degree1 endpoints of these \(m(d2)\) paths, plus the vertices of \(G'\) corresponding to the vertices of S. It is easy to see that any two vertices of \(S'\) have distance at least d from each other: S is an independent set in G, hence the corresponding vertices in \(G'\) are at distance at least \(2(d1)\ge d\) from each other, while the degree1 endpoints of the paths of length \(d1\) are at distance at least d from every other vertex that can potentially be in \(S'\). This shows \(\alpha _d(G')\ge \alpha (G)+m(d2)\). Conversely, let \(S'\) be a set of vertices in \(G'\) that are at distance at least d from each other. The set \(S'\) contains two types of vertices: let \(S'_1\) be the vertices that correspond to the original vertices of G and let \(S'_2\) be the vertices that come from the \(m(d2)d\) new vertices introduced in the construction of \(G'\). Observe that \(S'_2\) can be covered by \(m(d2)\) paths of length \(d1\) and each such path can contain at most one vertex of \(S'\), hence at most \(m(d2)\) vertices of \(S'\) can be in \(S'_2\). We claim that \(S'_1\) can contain at most \(\alpha (G)\) vertices, as \(S'\cap S'_1\) corresponds to an independent set of G. Indeed, if u and v are adjacent vertices of G, then the corresponding two vertices of \(G'\) are at distance \(d1\), hence they cannot be both present in \(S'\). This shows \(\alpha _d(G')\le \alpha (G)+m(d2)\), completing the proof of the correctness of the reduction. \(\square \)
As the following corollary shows, putting together Theorems 8 and 9 implies Theorem 4(2).
Corollary 2
If H is a graph having a component that is not a path, then, assuming the ETH, dScattered Set has no \(2^{o(n+m)}\)time algorithm on nvertex medge Hfree graphs for any fixed \(d\ge 3\).
Proof
Suppose first that H is not a forest and hence some cycle \(C_t\) for \(t\ge 3\) appears as an induced subgraph in H. Then the class of Hfree graphs is a superset of \(C_t\)free graphs, which means that statement follows from Theorem 9 (which gives a lower bound for a more restricted class of graphs).
Assume therefore that H is a forest. Then it must have a component that is a tree, but not a path, hence it has a vertex v of degree at least 3. The neighbors of v are independent in the forest H, which means that the claw \(K_{1,3}\) appears in H as an induced subgraph. Then the class of Hfree graphs is a superset of clawfree graphs, which means that statement follows from Theorem 8 (which gives a lower bound for a more restricted class of graphs). \(\square \)
Notes
Acknowledgements
References
 1.Agnarsson, G., Damaschke, P., Halldórsson, M.M.: Powers of geometric intersection graphs and dispersion algorithms. Discrete Appl. Math. 132(1–3), 3–16 (2003)MathSciNetCrossRefzbMATHGoogle Scholar
 2.Alekseev, V.: The Effect of Local Constraints on the Complexity of Determination of the Graph Independence Number. CombinatorialAlgebraic Methods in Applied Mathematics, pp. 3–13. Gorkiy University Press, Gorkiy (1982). (in Russian)Google Scholar
 3.Alekseev, V.E.: Polynomial algorithm for finding the largest independent sets in graphs without forks. Discrete Appl. Math. 135(1–3), 3–16 (2004)MathSciNetGoogle Scholar
 4.Bacsó, G., Marx, D., Tuza, Z.: \({H}\)free graphs, independent sets, and subexponentialtime algorithms. In: Guo, J., Hermelin, D. (eds.) 11th International Symposium on Parameterized and Exact Computation, IPEC 2016, August 24–26, 2016, Aarhus, Denmark, vol. 63 of LIPIcs, pp. 3:1–3:12. Schloss Dagstuhl—LeibnizZentrum fuer Informatik (2016)Google Scholar
 5.Bafna, V., Narayanan, B., Ravi, R.: Nonoverlapping local alignments (weighted independent sets of axisparallel rectangles). Discrete Appl. Math. 71, 41–53 (1996)MathSciNetCrossRefzbMATHGoogle Scholar
 6.Berman, P.: A \(d/2\) approximation for maximum weight independent set in \(d\)claw free graphs. In: Halldórsson, M. (ed.) Algorithm Theory—SWAT 2000, 7th Scandinavian Workshop on Algorithm Theory, Bergen Norway, July 2000, Proceedings, pp. 214–219. Springer (2000)Google Scholar
 7.Bhattacharya, B.K., Houle, M.E.: Generalized maximum independent sets for trees in subquadratic time. In: Algorithms and Computation, 10th International Symposium, ISAAC ’99, Chennai, India, December 16–18, 1999, Proceedings, pp. 435–445 (1999)Google Scholar
 8.Brandstädt, A., Mosca, R.: Maximum weight independent sets for (\({P}_7\), triangle)free graphs in polynomial time. CoRR, arXiv:1511.08066 (2015)
 9.Brause, C.: A subexponentialtime algorithm for the maximum independent set problem in \({P}_t\)free graphs. Discrete Appl. Math. 231, 113–118 (2017)MathSciNetCrossRefzbMATHGoogle Scholar
 10.Chalermsook, P., Laekhanukit, B., Nanongkai, D.: Independent set, induced matching, and pricing: connections and tight (subexponential time) approximation hardnesses. In: 54th Annual IEEE Symposium on Foundations of Computer Science, FOCS 2013, 26–29 October, 2013, Berkeley, CA, USA, pp. 370–379 (2013)Google Scholar
 11.Cygan, M., Fomin, F.V., Kowalik, L., Lokshtanov, D., Marx, D., Pilipczuk, M., Pilipczuk, M., Saurabh, S.: Parameterized Algorithms. Springer, Berlin (2015)CrossRefzbMATHGoogle Scholar
 12.Eto, H., Guo, F., Miyano, E.: Distance\(d\) independent set problems for bipartite and chordal graphs. J. Comb. Optim. 27(1), 88–99 (2014)MathSciNetCrossRefzbMATHGoogle Scholar
 13.Flum, J., Grohe, M.: Parameterized Complexity Theory. Texts in Theoretical Computer Science. An EATCS Series. Springer, Berlin (2006)Google Scholar
 14.Grzesik, A., Klimošová, T., Pilipczuk, M., Pilipczuk, M.: Polynomialtime algorithm for maximum weight independent set on \({P}_6\)free graphs. CoRR, arXiv:1707.05491 (2017)
 15.Gyárfás, A.: Problems from the world surrounding perfect graphs. Zastowania Matematyki Appl. Math. XIX(3–4), 413–441 (1987)MathSciNetzbMATHGoogle Scholar
 16.Halldórson, M. M.: Approximating discrete collections via local improvements. In: Proceedings of the Sixth Annual ACMSIAM Symposium on Discrete Algorithms (SODA ’95), pp. 160–169. SIAM (1995)Google Scholar
 17.Impagliazzo, R., Paturi, R., Zane, F.: Which problems have strongly exponential complexity? J. Comput. Syst. Sci. 63(4), 512–530 (2001)MathSciNetCrossRefzbMATHGoogle Scholar
 18.Lokshtanov, D., Marx, D., Saurabh, S.: Lower bounds based on the exponential time hypothesis. Bull. EATCS 105, 41–72 (2011)MathSciNetzbMATHGoogle Scholar
 19.Lokshtanov, D., Pilipczuk, M., van Leeuwen, E.J.: Independence and efficient domination on \({P}_{\text{6}}\)free graphs. In: Proceedings of the TwentySeventh Annual ACMSIAM Symposium on Discrete Algorithms, SODA 2016, Arlington, VA, USA, January 10–12, 2016, pp. 1784–1803 (2016)Google Scholar
 20.Lokshtanov, D., Vatshelle, M., Villanger, Y.: Independent set in \({P}_5\)free graphs in polynomial time. In: Proceedings of the TwentyFifth Annual ACMSIAM Symposium on Discrete Algorithms, SODA 2014, Portland, OR, USA, January 5–7, 2014, pp. 570–581 (2014)Google Scholar
 21.Lozin, V.V., Milanic, M.: A polynomial algorithm to find an independent set of maximum weight in a forkfree graph. J. Discrete Algorithms 6(4), 595–604 (2008)MathSciNetCrossRefzbMATHGoogle Scholar
 22.Lozin, V.V., Mosca, R.: Independent sets in extensions of \(2{K}_2\)free graphs. Discrete Appl. Math. 146(1), 74–80 (2005)MathSciNetCrossRefzbMATHGoogle Scholar
 23.Marx, D., Pilipczuk, M.: Optimal parameterized algorithms for planar facility location problems using Voronoi diagrams. In: Algorithms—ESA 2015—23rd Annual European Symposium, Patras, Greece, September 14–16, 2015, Proceedings, pp. 865–877 (2015)Google Scholar
 24.Minty, G.J.: On maximal independent sets of vertices in clawfree graphs. J. Comb. Theory Ser. B 28(3), 284–304 (1980)MathSciNetCrossRefzbMATHGoogle Scholar
 25.Randerath, B., Schiermeyer, I.: On maximum independent sets in \({P}_5\)free graphs. Discrete Appl. Math. 158(9), 1041–1044 (2010)MathSciNetCrossRefzbMATHGoogle Scholar
 26.Rosenkrantz, D.J., Tayi, G.K., Ravi, S.S.: Facility dispersion problems under capacity and cost constraints. J. Comb. Optim. 4(1), 7–33 (2000)MathSciNetCrossRefzbMATHGoogle Scholar
 27.Sbihi, N.: Algorithme de recherche d’un stable de cardinalite maximum dans un graphe sans etoile. Discrete Math. 29(1), 53–76 (1980)MathSciNetCrossRefzbMATHGoogle Scholar
 28.Thilikos, D.M.: Fast subexponential algorithms and compactness in planar graphs. In: Algorithms—ESA 2011—19th Annual European Symposium, Saarbrücken, Germany, September 5–9, 2011. Proceedings, pp. 358–369 (2011)Google Scholar
 29.Yu, G., Goldschmidt, O.: On locally optimal independent sets and vertex covers. Naval Res. Logistics 43, 737–748 (1996)MathSciNetCrossRefzbMATHGoogle Scholar
 30.Zuckerman, D.: Linear degree extractors and the inapproximability of max clique and chromatic number. Theory Comput. 3(1), 103–128 (2007)MathSciNetCrossRefzbMATHGoogle Scholar
Copyright information
Open AccessThis article is distributed under the terms of the Creative Commons Attribution 4.0 International License (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made.