arXiv reader

Maximum Bipartite Subgraph of Geometric Intersection Graphs

Satyabrata Jana, Anil Maheshwari, Saeed Mehrabi and Sasanka Roy

arXiv
1909.03896v2 · PDF · 17 March 2020
Categories
cs.DM, math.CO
Licence
CC BY 4.0 The site shows the text.

Abstract

We study the Maximum Bipartite Subgraph (MBS) problem, which is defined as follows. Given a set S of n geometric objects in the plane, we want to compute a maximum-size subset S'\subseteq S such that the intersection graph of the objects in S' is bipartite. We first give a simple O(n)-time algorithm that solves the MBS problem on a set of n intervals. We also give an O(n^2)-time algorithm that computes a near-optimal solution for the problem on circular-arc graphs. We show that the MBS problem is NP-hard on geometric graphs for which the maximum independent set is NP-hard (hence, it is NP-hard even on unit squares and unit disks). On the other hand, we give a PTAS for the problem on unit squares and unit disks. Moreover, we show fast approximation algorithms with small-constant factors for the problem on unit squares, unit disks and unit-height rectangles. Finally, we study a closely related geometric problem, called Maximum Triangle-free Subgraph (TFS), where the objective is the same as that of MBS except the intersection graph induced by the set S' needs to be triangle-free only (instead of being bipartite).

Items 39

Theorem

Theorem 1

7F8S

The \mathsf{MBS} problem on a set of n intervals can be solved in O(n) time, assuming that the intervals are already sorted on their right endpoint.

Lemma

Lemma 2

ZMAB

If G is triangle-free, then it can have at most one cycle.

Used by Proof
Proof of Lemma 2

Suppose for the sake of contradiction that G has more than one cycle. Let A_1 and A_2 be two cycles of G. Now, since G is a triangle-free circular-arc graph, the corresponding arcs of the vertices of any cycle in G together cover the circle \mathcal{C}. So, there must exist three distinct vertices v\in A_1, u\in A_1 and w\in A_2 such that v, u, w are pairwise adjacent. Which is a contradiction to the fact that G is triangle-free.

Lemma

Lemma 3

QX4F

If B and T are optimal solutions for the \mathsf{MBS} and \mathsf{MTFS} problems on G, respectively, then |T|-1\leq |B|\leq|T|.

Proof of Lemma 3

Since a bipartite subgraph contains no triangle, |B|\leq|T|. Now, if G[T] (i.e., the subgraph of G induced by T) is odd-cycle free, then it induces a bipartite subgraph. Otherwise, G[T] can have at most one cycle by Lemma 2. If this cycle is odd, then by removing any single vertex form the cycle, we obtain a bipartite subgraph of G with size at least |T|-1.

Uses Lemma 2
Lemma

Lemma 4

P5TF

If H is a maximum-size induced forest in G, then |V(H)|\geq |T|-1.

Theorem

Theorem 5

APUN

Let OPT be a maximum-size induced bipartite subgraph of a circular-arc graph G with n vertices. Then, there is an algorithm that computes an induced bipartite subgraph H of G such that |V(H)|\geq |OPT|-1. The algorithm runs in O(n^2) time.

Lemma

Lemma 6

6I1S

G has an independent set of size at least k if and only if G' has a bipartite subgraph of size at least 2k.

Proof of Lemma 6

Let U be an independent set of G with |U|\geq k. Let H be the subgraph of G' induced by U along with all the corresponding vertices of U in the other copy. Then, H is a bipartite subgraph with size at least 2k. Conversely, if G' has a bipartite subgraph of size at least 2k, then G' must have an independent set of size at least k. By the construction of G', if G' has an independent set of size at least k, then G must have an independent set of size at least k.

Theorem

Theorem 7

0XWF

The \mathsf{MBS} problem is \mathsf{NP}-complete on the classes of geometric intersection graphs for which \mathsf{MIS} is \mathsf{NP}-complete.

Remark

Remark

PCOG

By the definition of parameterized reduction [DBLP:series/mcs/DowneyF99], one can verify that the above reduction is in fact a parameterized reduction and so we have the following result.

Corollary

Corollary 8

FY9B

The \mathsf{MBS} problem is \mathsf{W[1]}-complete on the classes of geometric intersection graphs for which \mathsf{MIS} is \mathsf{W[1]}-complete.

Lemma

Lemma 9

SI2R

The \mathsf{MBS} problem has a feasible solution of size k on G if and only if there exists a directed path from s to t with the total weight k.

Proof of Lemma 9

For a given directed st-path with total weight k, let X be the union of all the disks corresponding to the internal vertices of this path. Then, the intersection graph of X is bipartite because the disks in X\cap {\cal M}_i are disjoint from the disks in S\cap {\cal M}_j when j>i+1. Moreover, when j=i+1, the disks in X\cap ({\cal M}_i\cup {\cal M}_j) must form an induced bipartite graph by the definition of an edge in G. Since the total weight of the vertices on the path is k, we have |X|=k. On the other hand, let Y be a feasible solution of size k for the \mathsf{MBS} problem on G. Then, the intersection graph of disks in Y\cap D_i is bipartite, for all i. Hence, selecting the vertices corresponding to Y\cap D_i for all i gives us a path with total weight k from s to t.

Theorem

Theorem 10

4SEV

There exists a \mathsf{PTAS} for \mathsf{MBS} on unit disks that runs in O(n^{O(1)}\cdot n^{O(1/\epsilon)}) time, for any \epsilon>0.

Theorem

Theorem 11

NGQN

There exists a \mathsf{PTAS} for the \mathsf{MBS} problem on unit squares running in O(n^{O(1)}\cdot n^{O(1/\epsilon)}) time, for any \epsilon>0. Moreover, the weighted \mathsf{MBS} problem also admits a \mathsf{PTAS} running within the same time bound on unit disks and unit squares.

Theorem

Theorem 12

4TDS

There exists an O(n\log n) time 2-approximation algorithm for the \mathsf{MBS} problem on a set of n unit-height rectangles in the plane.

Lemma

Lemma 13

CGE7

Let D_i, D_j, D_k \in \mathcal{D} be three disks with centers c_i , c_j and c_k, respectively, where x( c_i ) \leq x ( c_j ) \leq x ( c_k ). If D_i \cap D_j = \emptyset and D_j \cap D_k =\emptyset, then D_i\cap D_k= \emptyset.

Used by Proof, Proof
Proof of Lemma 13

We prove this by contradiction. Suppose that D_i\cap D_k\neq\emptyset. Let R be the rectangle such that R={ \ooalign{3mu\cr3mu\cr} }((x(c_i),0), (x(c_i),1), (x(c_k),1),(x(c_k),0)). Since x( c_i ) \leq x ( c_j ) \leq x ( c_k ), c_j must belong to R. Let m=(x(c_i)+x(c_k))/2. We partition the rectangle R into two rectangles A and B in the following way:

A={ \ooalign{$\sqsubset\mkern3mu$\cr$\mkern3mu\sqsupset$\cr} }((x(c_i),0),(x(c_i),1),(m,1),(m,0)),
B={ \ooalign{$\sqsubset\mkern3mu$\cr$\mkern3mu\sqsupset$\cr} }((m,0), (m,1), (x(c_k),1),(x(c_k),0)).

Since D_i\cap D_k \neq\emptyset, we have d(c_i,c_k)\leq 2. Then, d((x(c_i),0), (m,0))\leq 1 and d((m,0),(x(c_k),0))\leq 1. Now, if c_j\in A then d(c_i,c_j) \leq \sqrt{2} and so D_i \cap D_j \neq \emptyset. Otherwise, if c_j\in B, then d(c_j,c_k) \leq \sqrt{2} and so D_j\cap D_k\neq \emptyset. See Figure ? for an illustration. Since both cases lead to a contradiction, the lemma holds.

Lemma

Lemma 14

GA8B

There is no induced cycle of length at least 5 in G_{\mathcal{D}}.

Proof of Lemma 14

We prove this lemma by contradiction. Let G_{\mathcal{D}} contains an induced cycle C = \langle D_{1}, D_2, \dots, D_r, \dots, D_k \rangle of length k \geq 5. W.l.o.g. we assume that \forall i,~ 2\leq i \leq k,~ x(c_1) \leq x(c_i). Let D_r be the right-most disk according to the increasing order of x-coordinates of the centers of disks in the cycle C. Now there are two disjoint paths \pi_1 and \pi_2 in C between the vertices corresponding to D_1 and D_r. Further we assume that the number of disks in \pi_1 is atmost the number of disks in \pi_2. The other case is analogous. Our proof is based on the number of vertices in \pi_1.

First consider the case that D_1 \cap D_r \neq \phi. In this case \pi_1 consists of disks \{D_1, D_2 (=D_r) \} and \pi_2 consists of disks \{D_2, D_3, \dots, D_k, D_1\}. Note that the number of disks in \pi_2 is atleast 5. So there must be a disk D_i \in \pi_2 such that D_i \cap D_1 = \phi and D_i \cap D_2= \phi. As x (c_1) \leq x(c_i) \leq x(c_2), by Lemma 13, D_1 \cap D_2 = \phi. This contradicts our assumption.

Now we consider the case that \pi_1= (D_1, D_2, \dots, D_r ), where r \geq 3 and \pi_2= (D_1, D_k, \dots, D_{r+1}, D_r). Note that the number of disks in \pi_1 is atleast three and the number of disks in \pi_2 is atleast four. Now we consider following three subcases based on location of centers of D_2,\dots, D_{r-1}.

Case 1: x(c_{k}) \leq x(c_j) \leq x(c_{k-1}) holds for atleast one j \{ 2, \dots, r-1\}. As D_k \cap D_j = \emptyset and D_j \cap D_{k-1} = \emptyset, so by Lemma 13, D_k and D_{k-1} should not intersect each other. This contradicts our assumption that D_k \cap D_{k-1} \neq \phi.

Case 2: x(c_{i}) \leq x(c_k),~ \forall i, ~2 \leq i \leq r-1. As D_{r-1} \cap D_{k} = \emptyset and D_{k} \cap D_r = \emptyset, so by Lemma 13, D_{r-1} and D_r should not intersect each other. This contradicts our assumption that D_{r-1} \cap D_{r} \neq \phi.

Case 3: x(c_{k-1}) \leq x(c_{i}) ,~ \forall i,~ 2 \leq i \leq r-1. As D_1 \cap D_{k-1} = \emptyset and D_{k-1} \cap D_2 = \emptyset, so by Lemma 13, D_1 and D_2 should not intersect each other. This contradicts our assumption that D_1 \cap D_{2} \neq \phi.

Lemma

Lemma 15

1U6L

Let B[i,j,k] >3, and D_{\ell} \in D[i,j,k]. Then, no two disks corresponding to B[j,k,{\ell}] form a triangle with D_i.

Proof (Proof of Lemma 15) of Lemma 15

By assumption B[i,j,k] >3 and D_{\ell} \in \{D_{k+1}, \dots, D_n\} where the subgraph induced by \{ D_i, D_j, D_k, D_{\ell}\} is triangle-free. Let \Psi_{j,k,\ell} be the set of disks that defines the count for B[j,k,\ell]. Since B[i,j,k] >3, ~\Psi_{j,k,\ell} \neq \phi. We need to show that for any pair of disks D, D' \in \Psi_{j,k,\ell}, the subgraph induced by \{D_i, D, D'\} is triangle-free. We prove this by contradiction. Suppose for a contradiction that D and D' be two disks corresponding to B[j,k,\ell] such that D_i,D and D' form a triangle. As the subgraph induced by \{ D_j, D_k, D_{\ell},D, D'\} is triangle-free. By the definition of D[i,j,k], at most one of D and D' belongs to \{D_j, D_k, D_{\ell}\}.

Case 1: \{D, D'\}\cap\{D_j, D_k, D_{\ell}\}\neq\emptyset. Assume w.l.o.g. that D'\in \{D_j, D_k, D_{\ell}\}. We consider several cases.

  • D'=D_j. Since D_i\cap D\neq \emptyset, by Lemma 13, (i) either D_i \cap D_k\neq\emptyset or D_k\cap D\neq\emptyset as well as (ii) either D_i\cap D_{\ell}\neq\emptyset or D_{\ell}\cap D\neq\emptyset. Similarly, since D_j\cap D\neq\emptyset, by Lemma 13, (i) either D_j \cap D_{\ell}\neq\emptyset or D_{\ell} \cap D\neq\emptyset. Suppose that D_k \cap D\neq\emptyset. Now, if D_{\ell} \cap D\neq\emptyset, then \{D_j,D_k,D_l,D\} forms a K_{1,3}. This contradicts Lemma 16. Therefore, D_{\ell}\cap D=\emptyset; consequently, D_j\cap D_{\ell}\neq\emptyset and D_i\cap D_{\ell}\neq\emptyset. Then, the disks D_i, D_j, D_{\ell} form a triangle. A contradiction. So, D_k and D should not intersect each other. Then, D_i\cap D_k\neq\emptyset and D_j\cap D_k\neq\emptyset. Consequently, the disks D_i, D_j, D_k form a triangle and so B[i,j,k]=0. This contradicts our assumption that B[i,j,k]>3.

  • D'=D_k. Since D_i\cap D\neq\emptyset, by Lemma 13, either D_i\cap D_j\neq\emptyset or D_j\cap D\neq\emptyset. Suppose that D_j\cap D\neq\emptyset. Since D_k\cap D\neq\emptyset, either D_k\cap D_{\ell}\neq\emptyset or D_{\ell}\cap D\neq\emptyset (again by Lemma 13). Let D_{\ell}\cap D\neq \emptyset. Then, \{D_j, D_k, D_{\ell}, D\} forms a K_{1,3}. This contradicts Lemma 16. So, D_{\ell}\cap D=\emptyset, that means that D_k \cap D_{\ell}\neq\emptyset and D_i \cap D_{\ell}\neq \emptyset. Consequently, the disks D_i, D_k, D_{\ell} form a triangle. A contradiction. This means that the disks D_j and D do not intersect each other and so D_i \cap D_j\neq\emptyset. Now, if D_i\cap D_{\ell}\neq\emptyset, then \{D_i, D_j, D_k, D _{\ell}\} forms a K_{1,3}. This contradicts Lemma 16. Thus, the disks D_i and D_{\ell} do not intersect each other and so D_{\ell}\cap D\neq\emptyset. We know that previously D_i\cap D\neq\emptyset and D_i\cap D_j\neq\emptyset. Since D_j\cap D_k=\emptyset and D_k\cap D_{\ell}=\emptyset, we have D_j\cap D_{\ell}=\emptyset. Thus, the disks D_j, D_k, D_{\ell} are now pairwise independent. In this case, the disks D_i, D_j, D_k, D_{\ell}, D together contradict Lemma 18.

  • D'=D_{\ell}. Proof is analogous to the case D'=D_j. Since D_i\cap D\neq \emptyset, by Lemma 13, (i) either D_i \cap D_j\neq\emptyset or D_j\cap D\neq\emptyset as well as (ii) either D_i\cap D_{k}\neq\emptyset or D_{k}\cap D\neq\emptyset. Similarly, since D_i\cap D_{\ell}\neq\emptyset, by Lemma 13, (i) either D_i \cap D_{j}\neq\emptyset or D_{j} \cap D_{\ell}\neq\emptyset. Suppose that D_i \cap D_k\neq\emptyset. Now, if D_{i} \cap D_j\neq\emptyset, then \{D_i,D_j,D_k,D_{\ell}\} forms a K_{1,3}. A contradiction to Lemma 16. Therefore, D_{i}\cap D_j=\emptyset; consequently, D_j\cap D_{\ell}\neq\emptyset and D_{\ell}\cap D\neq\emptyset. Then, the disks D_j, D_{\ell}, D form a triangle. A contradiction. So, D_i and D_j should not intersect each other. Then, D_j\cap D_{\ell}\neq\emptyset and D_{\ell}\cap D\neq\emptyset. Consequently, the disks D_j, D_{\ell}, D form a triangle. This contradicts the fact the subgraph induced by \Psi_{j,k,\ell} is triangle-free.

Case 2: \{D, D'\}\cap\{D_j, D_k, D_{\ell}\}=\emptyset. Since D_i \cap D \neq \emptyset, by Lemma 13, either D_i \cap D_j\neq\emptyset or D_j\cap D\neq\emptyset. Assume that D_j\cap D\neq\emptyset. If D_j and D' intersects, then we get a triangle formed by D_j, D,D'. This contradicts the definition of B[j,k,\ell]. But, since D_i\cap D'\neq\emptyset, by Lemma 13, either D_i\cap D_j\neq\emptyset or D_j\cap D'\neq\emptyset. We have already shown that D_j\cap D'=\emptyset and so D_i\cap D_j\neq\emptyset. Then, the disks \{D_i, D_j, D\} forms a triangle. Observe that this is Case 1 and so D_j\cap D=\emptyset. By a similar argument, one can show that D_k\cap D=\emptyset and D_{\ell}\cap D=\emptyset. By Lemma 13, since D_i\cap D\neq\emptyset, D_i\cap D_j\neq\emptyset, D_i \cap D_k \neq \emptyset, D_i \cap D_{\ell} \neq \emptyset. This implies that \{D_j, D_k, D_{\ell}\} are pairwise independent. Then \{D_i, D_j, D_k, D _{\ell}\} forms a K_{1,3}. This contradicts to Lemma 16.

Lemma

Lemma 16

PXJW

Let D_i, D_j, D_k, D_{\ell}\in \mathcal{D} be four disks with x( c_i ) \leq x ( c_j ) \leq x ( c_k) \leq x(c_{\ell}). If D_i \cap D_{\ell} \neq \emptyset then the subgraph induced by \{D_i, D_j, D_k,D_{\ell}\} is K_{1,3}-free.

Used by Proof, Proof
Proof of Lemma 16

We prove the lemma by contradiction. Suppose that the subgraph induced by \{D_i, D_j, D_k, D_{\ell}\} is isomorphic to K_{1,3}. As D_i \cap D_{\ell} \neq \emptyset, either (i) D_i intersects both D_j and D_k, where D_j, D_k, D_{\ell} are pairwise independent, or (ii) D_{\ell} intersects both D_j and D_k, where D_i, D_j, D_{k} are pairwise independent,. We prove the lemma for (i); the proof for (ii) is similar. Consider the rectangle R={ \ooalign{3mu\cr3mu\cr} }((x(c_i),0), (x(c_i),1), (x(c_{\ell}),1),(x(c_{\ell}),0))). Since x( c_i ) \leq x ( c_j ) \leq x ( c_k) \leq x(c_{\ell}), we have that c_j and c_k must belong to R. Let m=(x(c_i)+x(c_{\ell}))/2. We partition the rectangle R into two rectangles A and B in the following way:

A ={ \ooalign{$\sqsubset\mkern3mu$\cr$\mkern3mu\sqsupset$\cr} }((x(c_i),0), (x(c_i),1), (m,1), (m, 0)),
B= { \ooalign{$\sqsubset\mkern3mu$\cr$\mkern3mu\sqsupset$\cr} }((m,0), (m,1), (x(c_{\ell}),1), (x(c_{\ell}),0)).

Since D_i \cap D_{\ell} \neq \emptyset, we have d(c_i,c_{\ell})\leq 2. Therefore, d((x(c_i),0), (m,0))\leq 1 and d((m,0),(x(c_{\ell}),0))\leq 1. If c_j, c_k\in A or c_j, c_k\in B, then d(c_j,c_k)\leq\sqrt{2}<2 and so D_j\cap D_k \neq \emptyset. This contradicts that D_j, D_k, D_{\ell} are pairwise independent. Otherwise, c_j\in A and c_k \in B. Since c_{\ell}\in B so d(c_k,c_{\ell}) \leq \sqrt{2}<2. Therefore, D_k\cap D_{\ell}\neq \emptyset, this contradicts that D_j, D_k, D_{\ell} are pairwise independent.

Lemma

Lemma 17

FS5A

The graph G_{\mathcal{D}} is K_{1,4}-free.

Proof of Lemma 17

We prove the lemma again by contradiction. Suppose there is an induced subgraph K_{1,4} with vertex set \{D_i\colon 1\leq i\leq 5\}. Assume w.l.o.g. that x( c_1 ) \leq x ( c_2 ) \leq x ( c_3) \leq x(c_4) \leq x(c_5). In \{D_i\colon 1\leq i\leq 5\}, there must be four pair-wise independent disks and one disk (say, D) that intersects each of that four disks. If D=D_1~ \text{or}~D_4. Then the subgraph induced by \{D_i\colon 1\leq i\leq 4\} is isomorphic to K_{1,3}. This contradicts Lemma 16. Similarly if D=D_2 ~\text{or}~D_5. Then the subgraph induced by \{D_j\colon 2\leq j\leq 5\} is isomorphic to K_{1,3}. This contradicts Lemma 16. Hence only possibility is that D=D_3 and the edge set in K_{1,4} is \{(D_1, D_3),(D_2, D_3),(D_3, D_4),(D_3, D_5)\}.

Now, consider the rectangle R={ \ooalign{3mu\cr3mu\cr} }((x(c_3)-2,0),(x(c_3)-2,1),(x(c_3)+2,1),(x(c_3)+2,0))). We partition R into three rectangles A, B and C as follows:

A ={ \ooalign{$\sqsubset\mkern3mu$\cr$\mkern3mu\sqsupset$\cr} }((x(c_3)-2,0), (x(c_3)-2,1), (x(c_3)-0.5,1), (x(c_3)-0.5,0)),
\hspace{5mm} B ={ \ooalign{$\sqsubset\mkern3mu$\cr$\mkern3mu\sqsupset$\cr} }((x(c_3)-0.5,0), (x(c_3)-0.5,1), (x(c_3)+0.5,1), (x(c_3)+0.5,0)),
C ={ \ooalign{$\sqsubset\mkern3mu$\cr$\mkern3mu\sqsupset$\cr} }((x(c_3)+0.5,0), (x(c_3)+0.5,1), (x(c_3)+2,1), (x(c_3)+2,0)).

Since c_i belongs to R, for 1\leq i\leq 5, by the pigeonhole principle, one of A, B and C must contain two of \{c_1, c_2, c_4, c_5\}. The distance between any two points in the rectangles A, B or C is \leq 2. Therefore, \{D_1, D_2, D_4, D_5\} are not pairwise independent. Hence, G_{\mathcal{D}} is K_{1,4}-free.

Lemma

Lemma 18

EXUZ

Let D_1, D_2, D_3, D_4, D_5 be five disks in \mathcal{D} with x( c_i ) \leq x ( c_{i+1} ) where 1 \leq i \leq 4 . If D_1\cap D_5 \neq\emptyset then the disks D_2, D_3, D_4 do not form an independent set.

Used by Proof
Proof of Lemma 18

The proof proceeds analogous to the proof of Lemma 17. Since D_1 and D_5 intersect, d(c_1,c_5)\leq 2, where c_1 and c_5 are the centres of the disks D_1 and D_5, respectively. Now, consider the rectangle

\hspace{13mm} R={ \ooalign{$\sqsubset\mkern3mu$\cr$\mkern3mu\sqsupset$\cr} }((x(c_1),0), (x(c_1),1), (x(c_5),1),(x(c_5),0)),

and let m=(x(c_1)+x(c_5))/2. We divide R into two rectangles A and B, where

A ={ \ooalign{$\sqsubset\mkern3mu$\cr$\mkern3mu\sqsupset$\cr} }((x(c_1),0),(x(c_1),1),(m,1),(m,0))
B={ \ooalign{$\sqsubset\mkern3mu$\cr$\mkern3mu\sqsupset$\cr} }((m,0),(m,1),(x(c_5),1),(x(c_5),0)).

Now, since each of \{c_i\colon 1\leq i\leq 5\} belongs to R, one of A and B must contain two of \{c_2, c_3, c_4\}. The distance between any two points in the rectangle A or B is at most two. Therefore D_2, D_3, D_4 do not form an independent set.

Theorem

Theorem 19

9VH4

Let \mathcal{D} be a set of n unit-disks intersecting a straight line where all the centers lie on one side of the line. In O(n^4) time and O(n^3) space, we can find maximum sized \mathcal{D'} \subseteq \mathcal{D} such that \mathcal{D'} induces a bipartite subgraph.

Theorem

Theorem 20

G2CN

Let \mathcal{D} be a set of n unit disks intersecting a line in the plane and let G_{\mathcal{D}} be the intersection graph of \mathcal{D}. An induced bipartite subgraph of size at least |\mathsf{MBS}(G_{\mathcal{D}})|/2 can be computed in O(n+|E(G_{\mathcal{D}})|) time, where \mathsf{MBS}(G_{\mathcal{D}}) is an optimal solution for the \mathsf{MBS} problem on G_{\mathcal{D}}.

Observation

Observation 1

0MK4

Let D \in \mathcal{D}_i and D' \in \mathcal{D}_j be two disks in \mathcal{D}, where 1 \leq i,j \leq k. If |i-j| \geq 2 then D and D' do not intersect.

Theorem

Theorem 21

SI4Y

Let \mathcal{D} be a set of n unit disks in the plane and let G_{\mathcal{D}} be the intersection graph of \mathcal{D}. An induced bipartite subgraph of size at least |\mathsf{MBS}(G_{\mathcal{D}})|/3 can be computed in O(n^4) time, where \mathsf{MBS}(G_{\mathcal{D}}) is a maximum induced bipartite subgraph of G_{\mathcal{D}}.

Theorem

Theorem 22

ALBR

Let \mathcal{D} be a set of n unit disks in the plane and let G_{\mathcal{D}}=(V,E) be the intersection graph of \mathcal{D}. An induced bipartite subgraph of size at least |\mathsf{MBS}(G_{\mathcal{D}})|/(2\log n) can be computed in O(n\log n +|E|) time, where \mathsf{MBS}(G_{\mathcal{D}}) denotes a maximum induced bipartite subgraph in G_{\mathcal{D}}.

Lemma

Lemma 23

SXQK

[rim1995rectangle] G=(V,E) has an independent set of size \geq m if and only if H has an independent set of size m+3|E|.

Lemma

Lemma 24

BDFZ

H has an independent set of size \geq k if and only if G_H has a triangle-free subgraph on \geq k+7|E| vertices.

Proof of Lemma 24

Let H have an independent set of k vertices. Let B' \subseteq B be the set of rectangles corresponding to these k vertices. Note that the rectangles in R are independent. We take the subgraph H' of G_H induced by the vertices corresponding to the rectangles B' \cup R. Now H' is triangle-free. Because if H' is not triangle-free then there is an intersection between two black rectangles in B', that leads to a contradiction. As |R|=7|E|, so the claim holds.

Now we show the other direction. Let G_H have a triangle-free subgraph H' on \geq k+7|E| vertices. Let X(H')= B_1 \cup R_1 be the set of rectangles corresponding to the vertices of H', where B_1 \subseteq B and R_1 \subseteq R . Also, let R_2 \subseteq (R \setminus R_1) be the set of those red rectangles that has at most one adjacent rectangle in B_1. Then clearly the graph with underlying rectangles B_1 \cup R_1 \cup R_2 has no triangles. So each rectangle in R \setminus (R_1 \cup R_2) has exactly two neighbours in B_1. Let E(B_1) denote the set of edges in the subgraph induced by the vertices corresponding to the rectangles in B_1. Note that if there is a pair of adjacent rectangles b_i and b_j in B_1 then the rectangle R_{i,j} should be part of R_3. It implies that |E(B_1)|= |R_3| so |E(B_1)|= |R|-|R_1 \cup R_2|. Now |B_1| + | R_1 \cup R_2| \geq k+ 7|E|, so |B_1| - (7|E|- | R_1 \cup R_2|) \geq k. This implies |B_1| - (|R|- | R_1 \cup R_2|) \geq k, hence |B_1| - |E(B_1)| \geq k. Now in B_1 we repeatedly remove the rectangles from B_1 to get an independent set of rectangles. Finally, within at most |E(B_1)| steps, we are left with a set of independent rectangles in B_1 with size \geq k. So there exists an independent set of size \geq k in H.

Lemma

Lemma 25

B08L

G=(V,E) has an independent set of size \geq m if and only if G_H has a triangle-free subgraph on m+10|E| vertices.

Theorem

Theorem 26

107O

The \mathsf{MTFS} problem is \mathsf{NP}-complete on axis-parallel rectangle intersection graphs.

Source and licence

The text on this page is from arXiv:1909.03896v2 by Satyabrata Jana, Anil Maheshwari, Saeed Mehrabi and Sasanka Roy, under the licence CC BY 4.0. The corpus made these changes: it read the items out of the LaTeX source, it expanded the macros of the paper, it put each proof after its statement, and it converted the text to HTML. The formulas are the TeX of the paper, and KaTeX renders them. A formula or a reference can look different from the PDF. The page is published under CC BY 4.0.

The record of each item is in analysis/objects/1909/1909.03896.jsonl.