0% found this document useful (0 votes)
771 views19 pages

Karp

reducibility among combinatorial problems

Uploaded by

MikKeOrtiz
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF or read online on Scribd
0% found this document useful (0 votes)
771 views19 pages

Karp

reducibility among combinatorial problems

Uploaded by

MikKeOrtiz
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF or read online on Scribd
You are on page 1/ 19
t REDUCIBILITY AMONG COMBINATORIAL PROBLEMS Richard M. Karp University of California at Berkeley Abstract: A large class of computational problems involve the determination of properties of graphs, digraphs, integers, arrays of integers, finite families of finite sets, boolean formulas and elements of other countable domains. ‘Through simple encodings from such domains into the set of words over a finite alphabet these problems can be converted into language recognition problems, and we can inquire into their computational complexity. It is reasonable to consider such a problem satisfactorily solved when an algorithm for its solution is found which terminates within a number of steps bounded by a polynomial in the length of the input. We show that a large number of classic unsolved problems of cover- ing, matching, packing, routing, assignment and sequencing are equivalent, in the sense that either each of them possesses a polynomial-bounded algorithm or none of them does. «INTRODUCTION All the general methods presently known for computing the chromatic number of a graph, deciding whether a graph has a Hamilton circuit, or solving a system of linear inequalities in which the variables are constrained to be 0 or 1, require a combinatorial search for which the worst case time requirement grows exponentially with the length of the input. In this paper we give theorems which strongly suggest, but do not imply, that these problems, as well as many others, will remain intractable perpetually. This research was partially supported by National Science Founda~ tion Grant GJ-474. 86 RICHARD M. KARP We are specifically interested in the existence of algorithms that are guaranteed to terminate in a number of steps bounded by a polynomial in the length of the input, We exhibit a class of well- known combinatorial problems, including those mentioned above, which are equivalent, in the sense that a polynomial-bounded algo- rithm for any one of them would effectively yield a polynomial- bounded algorithm for all. We also show that, if these problems do possess polynomial-bounded algorithms then all the problems in an unexpectedly wide class (roughly speaking, the class of problems solvable by polynomial-depth backtrack search) possess polynomial- bounded algorithms. The following is a brief summary of the contents of the paper. For the sake of definiteness our technical development is carried out in terms of the recognition of languages by one-tape Turing machines, but any of a wide variety of other abstract models of computation would yield the same theory, Let I* be the set of all finite strings of O's and 1's. A subset of I* is called a language. Let P be the class of languages recognizable in polynomial time by one-tape deterministic Turing machines, and let NP be the class of languages recognizable in polynomial time by one-tape nondeterministic Turing machines. Let fl be the class of functions from =* into £* computable in polynomial time by one-tape Turing machines. Let L and M be languages. We say that L=M (L is reducible to M) if there is a function f € Il such that f(x) @M*xé€L, If Me P and L&M then Le P. We call L and M equivalent if L=M and ML, Call L (polynomial) complete if Le NP and every language in NP is reducible to L. Either all complete languages are in P, or none of them are. The former alternative holds if and only if P = NP. The main contribution of this paper is the demonstration that a large number of classic difficult computational problems, arising in fields such as mathematical programming, graph theory, combina tories, computational logic and switching theory, are complete (and hence equivalent) when expressed in a natural way as language recognition problems. This paper was stimulated by the work of Stephen Cook (1971), and rests on an important theorem which appears in his paper. The author also wishes to acknowledge the substantial contributions of Eugene Lawler and Robert Tarjan. 2, THE CLass P There is a large class of important computational problems which involve the determination of properties of graphs, digraphs, integers, finite families of finite sets, boolean formulas and REDUCIBILITY AMONG COMBINATORIAL PROBLEMS, 87 elements of other countable domains. It is a reasonable working hypothesis, championed originally by Jack Edmonds (1965) in connec~ tion with problems in graph theory and integer programming, and by now widely accepted, that such a problem can be regarded as tract- able if and only if there is an algorithm for its solution whose running time is bounded by a polynomial in the size of the input. In this section we introduce and begin to investigate the class of problems solvable in polynomial time. We begin by giving an extremely general definition of "deter- ministic algorithm", computing a function from a countable domain D into a countable range R. For any finite alphabet A, let A* be the set of finite strings of elements of A; for x€ A*, let Ig(x) denote the length of x. A deterministic algorithm A is specified by: a countable set D (the domain) a countable set R (the range) a finite alphabet A such that A*AR = an encoding function E: D > a* a transition function A* + AFUR , The computation of A on input x €D is the unique sequence YysYgeees Such that y, = E(x), y,,, = T(y,) for all i and, if the sequence is finite and ends with y,, then y, € R. Any string occurring as an element of a computStion is cafled an instantaneous description. If the computation of A on input x is finite and of length t(x), then t(x) is the running time of A on input x. A is terminating if all its computations are finite. A terminating algorithm A computes the function f4: DR such that £,(x) is the last element of the computation of A on x. If R= {ACCEPT,REJECT} then A is called a recognition algorithm. A recognition algorithm in which D = E* is called a string recognition algorithm, If A is a string recognition algorithm then the language recognized by A is {xe D*| £4(x) = ACCEPT}. If D=R-=Z* then A is called a string mapping algorithm. A terminating algorithm A with domain D = © operates in Polynomial time time if there is a polynomial p(-) ‘such that, for every xe I*, t(x) < p(Ig(x)). To discuss algorithms in any practical context we must spe- cialize the concept of deterministic algorithm. Various well known classes of string recognition algorithms (Markov algorithms, one-tape Turing machines, multitape and multihead Turing machines, 88 RICHARD M. KARP tandom access machines, etc.) are delineated by restricting the func- tions E and T to be of certain very simple types. These definitions are standard (Hopcroft & Ullman (1969)] and will not be repeated here. It is by now commonplace to observe that many such classes are equi- valent in their capability to recognize languages; for each such class of algorithms, the class of languages recognized is the class of recursive languages. This invariance under changes in definition is part of the evidence that recursiveness is the cor- rect technical formulation of the concept of decidability. The class of languages recognizable by string recognition algorithms which operate in polynomial time is also invariant under a wide range of changes in the class of algorithms. For example, any language recognizable in time p(+) by a multihead or multitape Turing machine is recognizable in time p*(+) by a one-tape Turing machine. Thus the class of languages recognizable in polynomial time by one-tape Turing machines is the same as the class recognizable by the ostensibly more powerful multihead or multitape Turing machines. Similar remarks apply to random access machines, Definition 1. P is the class of languages recognizable by one-tape Turing machines which operate in polynomial time. Definition 2. I is the class of functions from E* into I* defined by one-tape Turing machines which operate in polynomial time. The reader will not go wrong by identifying P with the class of languages recognizable by digital computers (with unbounded backup storage) which operate in polynomial time and fl with the class of string mappings performed in polynomial time by such computers. Remark. If £: £*+Z* is in M then there is a polynomial p(+) such that 1g(£(x)) < p(1g(x)). We next introduce a concept of reducibility which is of cen- tral importance in this paper. Definition 3. Let L and M be languages, Then L «= M (Lis reducible to M) if there is a function f € Ml such that f(x) EM*x EL. Lemma 1. If L«=M and MEP then Le P. Proof. The following is a polynomial-time bounded algorithm to decide if x €L: compute £(x); then test in polynomial time whether f(x) € M. We will be interested in the difficulty of recognizing subsets of countable domains other than I*. Given such a domain D, REDUCIBILITY AMONG COMBINATORIAL PROBLEMS: a9 there is usually a natural one-one encoding e: D+ Z*. For exam- ple we can represent a positive integer by the string of O's and 1's comprising its binary representation, a 1-dimensional integer array as a list of integers, a matrix as a list of 1-dimensional arrays, etc.; and there are standard techniques for encoding lists into strings over a finite alphabet, and strings over an arbitrary finite alphabet as strings of 0's and 1's. Given such an encod~ ing e: D+2*, we say that a set TCD is recognizable in poly- nomial time if e(T) € P. Also, given sets TCD and UCD', and encoding functions e: D+ Z* and e!: >=* we say T € L'” and lg(y) < p(1g(x))}. REDUCIBILITY AMONG COMBINATORIAL PROBLEMS. a We refer to L as the language derived from L‘?) py p-bounded existential quantification. Definition 4, NP is the set of languages derived from ele- ments of PCY by polynomial-bounded existential quantification. There is an alternative characterization of NP in terms of nondeterministic Turing machines, A nondeterministic recognition algorithm A is specified by: a countable set D (the domain) a finite alphabet A such that A* M{ACCEPT,REJECT} = > an encoding function E: D + A* a transition relation +t C A*x (\*U {ACCEPT,REJECT}) such that, for every y. € A*, the set {| € Tt} has fewer than ka elements, where ka is a constant. A computation of A on input x €D is a sequence y,,y7,... such that yy = E(x), € T for all i, and, if the sequence is finite and ends with y,, then y, € {ACCEPT,REJECT}. A string y € O* which occurs in some computation is an instantaneous description. A finite computation ending in ACCEPT is an accepting computation. Input x is accepted if there is an accepting computation for x. If D=2* then A is a nondeter- ministic string recognition algorithm and we say that A operates in polynomial time if there is a polynomial p(+) such that, when- ever A acespts' x, there is an accepting computation for x of length < p(1g(x)). A nondeterministic algorithm can be regarded as a process which, when confronted with a choice between (say) two alternatives, can create two copies of itself, and follow up the consequences of both courses of action. Repeated splitting may lead to an exponen- tially growing number of copies; the input is accepted if any sequence of choices leads to acceptance. The nondeterministic 1-tape Turing machines, multitape Turing machines, random-access machines, etc, define classes of nondeterministic string recognition algorithms by restricting the encoding function E and transition relation Tt to particularly simple forms. All these classes of algorithms, restricted to oper- ate in polynomial time, define the same class of languages. More over, this class is NP. Theorem 1. L€ NP if and only if L is accepted by a non~ deterministic Turing machine which operates in polynomial time. Proof. = Suppose L€ NP. Then, for sone L(2) € P(2) and some polynomial p, L is obtained from L‘2) by p-bounded exis- tential quantification. We can construct a nondeterministic 92 RICHARD M. KARP machine which first guesses the successive digits of a s ing y of length < p(1g(y)) and then tests whether € L(2), such a machine clearly recognizes L in polynomial time “Suppose L is accepted by a nondeterministic Turing machine T which operates in time p. Assume without loss of generality that, for any instantaneous description Z, there are at most two instantaneous descriptions that may follow Z (i.e., at most two primitive transitions are applicable). Then the se— quence of choices of instantaneous descriptions made by T in a given computation can be encoded as a string y of 0's and 1's, such that 1g(y) < p(1g(x)). : Thus we can construct a deterministic Turing machine T', with Z*xE* as its domain of inputs, which, on input , simulates the action of T on input x with the sequence of choices y. Clearly T' operates in polynomial time, and L is obtained by polynomial bounded existential quantification from the set of pairs of strings accepted by 1! The class NP is very extensive. Loosely, a recognition problem is in NP if and only if it can be solved by a backtrack search of polynomial bounded depth. A wide range of important computational problems which are not known to be in P are obvious- ly in NP. For example, consider the problem of determining whe- ther the nodes of a graph G can be colored with k colors so that no two adjacent nodes have the same color. A nondeterministic algorithm can simply guess an assignment of colors to the nodes and then check (in polynomial time) whether all pairs of adjacent nodes have distinct colors. In view of the wide extent of NP, the following theorem due to Cook is remarkable. We define the satisfiability problem as follows: SATISFIABILITY INPUT: Clauses C1,C7,...,C, PROPERTY: The conjunction of the given clauses is satisfiable; i.e., there is a set S$ C {x],x7,...,%,5%),%25++-,%,) such that a) § does not contain a Complementary pair of literals and b) SAC. #6, k= 1,2,..05p- Theorem 2 (Cook), If L€NP then L« SATISFIABILITY. The theorem stated by Cook (1971) uses a weaker notion of reducibility than the one used here, but Cook's proof supports the present statement. Corollary 1. P = NP © SATISFIABILITY € P. REDUCIBILITY AMONG COMBINATORIAL PROBLEMS 9 Proof. If SATISFIABILITY € P then, for each Le NP, Le P, since L = SATISFIABILITY. If SATISFIABILITY ¢ P, then, since clearly SATISFIABILITY € NP, P # NP. Remark. If P= NP then NP is closed under complementation and polynomial-bounded existential quantification. Hence it is also closed under polynomial-bounded universal quantification, It follows that a polynomial-bounded analogue of Kleene's Arithmetic Hierarchy [Rogers (1967)] becomes trivial if P = NP. Theorem 2 shows that, if there were a polynomial-time algo~ rithm to decide membership in SATISFIABILITY then every problem solvable by a polynomial-depth backtrack search would also be solvable by a polynomial-time algorithm. This is strong circum- stantial evidence that SATISFIABILITY ¢ P. 4, COMPLETE PROBLEMS The main object of this paper is to establish that a large number of important computational problems can play the role of SATISFIABILITY in Cook's theorem. Such problems will be called complete. Definition 5. The language L is (polynomial) complete if a) Le NP and b) SATISFIABILITY = L. Theorem 3. Either all complete languages are in P, or none of them are. The former alternative holds if and only if P = NP. We can extend the concept of completeness to problems defined over countable domains other than I*, Definition 6, Let D be a countable domain, e a "standard" one-one encoding e: D+Z* and T a subset of D, Then T is complete if and only if e(D) is complete. Lemma 2, Let D and D’ be countable domains, with one-one encoding functions e and e'. Let TCD and T'CD'. Then T#T" if there is a function F: DD’ such that a) F(x) €T!@xeT iy and b) there is a function £€ I such that £(x) = e'(F(e™'(x))) whenever e'(F(e7l(x))) is defined. The rest of the paper is mainly devoted to the proof of the following theorem. 94 RICHARD M. KARP Main Theorem. All the problems on the following list are complete. . SATISFIABILITY COMMENT: By duality, this problem is equivalent to deter- mining whether a disjunctive normal form expression is a tautology. 2. O-1 INTEGER PROGRAMMING INPUT: integer matrix C and integer vector d PROPERTY: There exists a O-1 vector x such that Cx = 3. CLIQUE INPUT: graph G, positive integer k PROPERTY: G has a set of k mutually adjacent nodes. SET PACKING INPUT: Family of sets {S,}, positive integer & PROPERTY: {s,} contains “2 mutually disjoint sets. NODE COVER INPUT: graph G', positive integer & PROPERTY: There is a set RCN' such that [R| < 2 and every are is incident with some node in R. 6. SET COVERING INPUT: finite family of finite sets {S,}, positive integer k PROPERTY: There is a subfamily {1,} C {Sj} containing < k sets such that Ur, = US. 7. FEEDBACK NODE SET INPUT: digraph H, positive integer k PROPERTY: ‘There is a set RCV _ such that every (directed) cycle of H contains a node in R. 8. FEEDBACK ARC SET INPUT: digraph H, positive integer k PROPERTY: There is a set SCE such that every (directed) cycle of H contains an are in Ss. os DIRECTED HAMILTON CIRCUIT INPUT: digraph H PROPERTY: H has a directed cycle which includes each node exactly once. 10, UNDIRECTED HAMILTON CIRCUIT INPUT: graph G PROPERTY: G has a cycle which includes each node exactly once. REDUCIBILITY AMONG COMBINATORIAL PROBLEMS. 95 le lz. 13. 1s, 15. 16. a, 18. 19. that SATISFIABILITY WITH AT MOST 3 LITERALS PER CLAUSE INPUT: Clauses Dj,D2,...,Dp, each consisting of at most 3 literals from the set’ {uy,u2,---,u,) U (G,02,...,0,) PROPERTY: The set {D,,D5,...,D,} is satisfiable. CHROMATIC NUMBER INPUT: graph G, positive integer k PROPERTY: There is a function 4: N+ Z, such that, if-u and v are adjacent, then $(u) # o(v). CLIQUE COVER INPUT: graph G', positive integer & PROPERTY: N' is the union of £& or fewer cliques. EXACT COVER INPUT: family {Sj} of subsets of a set {uj, i= 1,2,...,t) PROPERTY: There i8 a subfamily {1} C {sj} such that the =uUs, = = sets Tj are disjoint and UT, 35 uy, £ = 1,2,..-,t}. HITTING SET INPUT: family {Uj} of subsets of {s,, j = 1,2,...,r} PROPERTY: There is a set W such that, for each i, [win ul al. STEINER TREE INPUT: graph G, RCN, weighting function w: A> Z, positive integer k PROPERTY: G has a subtree of weight Z, positive integer W PROPERTY: There is a set S CN such that 2 w(tuvi) >w {u,v}ea ues ves It is clear that these problems (or, more precisely, their encodings into I*), are all in NP. We proceed to give a series of explicit reductions, showing that SATISFIABILITY is reducible to each of the problems listed. Figure 1 shows the structure of the set of reductions. Each line in the figure indicates a reduc- tion of the upper problem to the lower one. To exhibit a reduction of a set TCD toaset T'Cop', we specify a function F: D+ D' which satisfies the conditions of Lemma 2. In each case, the reader should have little difficulty in verifying that F does satisfy these conditions. SATISFIABILITY = 0-1 INTEGER PROGRAMMING Lif xy eC, cy 7 fd At Re cy 0 otherwise j L=1,2,...5p 1,2,....0 = 1-(the number of complemented variables in C,) , i= 1,2,.0.5Pe SATISFIABILITY = CLIQUE N= {| 6 is a literal and occurs in cy} A= {{,}| i # § and a # §} k =p, the number of clauses. CLIQUE = SET PACKING Assume N= {1,2,...,n}, The elements of the sets 81,S2,-++,S_ are those two-element sets of nodes {i,j} not in A. sy =€4,5)] (4,5) € A i= 1,2,...,0 Lek. 98 . RICHARD M. KARP CLIQUE © NODE COVER G' is the complement of 6. Q= |N| -k NODE COVER « SET COVERING Assume N' = {1,2,...,n}, The elements are the arcs of G'. Sj 4s the set of arcs incident with node j. k= 2. NODE COVER = FEEDBACK NODE SET ven! E= {] {u,v} e at} kek NODE COVER « FEEDBACK ARC SET v = N'x {0,1} E = {<,>| u €N'} U {<,>| {u,v} € A") k= a NODE COVER « DIRECTED HAMILTON CIRCUIT Without loss of generality assume A' = Zp. V = fay,az,...,ag} U {| u € N’ is incident with ie At and a € {0,1}} E = {<,>| € v} U {<,>| i € A', u and v are incident with he ae {0,1}} U {<,>] u is incident with i and j and dh, i,ag>] 1 < £ < 2 and Yh > 4 such that u is inci- dent with h} U {>| 1 < £ < 2 and dh < i such that u is inci- dent with h} . DIRECTED HAMILTON CIRCUIT « UNDIRECTED HAMILTON CIRCUIT N = vx {0,1,2} A= {{,},{,}| ue v} U {{,}] € E} SATISFIABILITY « SATISFIABILITY WITH AT MOST 3 LITERALS PER CLAUSE Replace a clause 0]U0)U-+-Uo,, where the oj are literals and m>3, by (04 Yon Yuz) (03 U +++ Uoq VEL) Gy Vuz)+++ Guy), where uj; is anew variable. Repeat this transformation until no clause has more than three literals. REDUCIBILITY AMONG COMBINATORIAL PROBLEMS SATISFIABILITY WITH AT MOST 3 LITERALS PER CLAUSE ‘© CHROMATIC NUMBER Assume without loss of generality that m> 4. 2Vnh sgh U (yy v2 9062 9¥q N= (uzsugsseejtgh U {3,595 U {Dy ,D9,.++,D,} A= (Cug,0g}] 491,2,...5n} U {{vgsvy}] 45) U Clvgsxy}] 445} U Covers] 45) U {Cup sDg}] ug ¢ Dg) U H;,D,)] a e Dg) k=rtl CHROMATIC NUMBER « CLIQUE COVER G' is the complement of G Lak, CHROMATIC NUMBER « EXACT COVER The set of elements is NUAU{cu,e,£>| u is incident with e andl ] e is incident with u} ; for each e€€ A and each pair £), £2 such that 1S fp ek, 1S fp,£#£;} U{| ext.) , where u and v are the two nodes incident with e. EXACT COVER « HITTING SET 7 The hitting set problem has sets Uj and elements Sj, such that 55 € Uy * uy € Sy. EXACT COVER « STEINER TREE N= {no} V{s5} Ulug} R = {ng} U{uy} A= (lng.53}} ULES; ,ug}] uj € $3) w({ng,S3}) = [85] w({S;,uz}) = 0 k= |{uj}] EXACT COVER « 3-DIMENSIONAL MATCHING Without loss of generality assume iss] > 2 foreach j. Let T= {] ue 8,). Let a be an arbitrary one-one function 100 RICHARD M. KARP from {uj} into T. Let m: T+ T be a permutation such that, for each fixed j, {| uy € 85} is a cycle of 1, U = {,>] € TH U {| for all i, B#a(uy)} . EXACT COVER « KNAPSACK = : . f ifues Let d= |{sj}|+1. Let €5; ¢ iE uy es} Let r= |{sj)|, ay =] ejga** and b= KNAPSACK ® SEQUENCING par, Ty = Py = ay, KNAPSACK © PARTITION srt? cet ay, 1 1,2,...,0 cet 7 B41 Cepg 7 CE a+ l-b een PARTITION = MAX CUT N= {1,2,...,5} A= {(i,j}] 1€N, 7 €N, i # 5} w(t, j} = eres = fer c® we [Zr ci] Some of the reductions exhibited here did not originate with the present writer. Cook (1971) showed that SATISFIABILITY « SATISFIABILITY WITH AT MOST 3 LITERALS PER CLAUSE. The reduction SATISFIABILITY © CLIQUE is implicit in Cook (1970), and was also known to Raymond Reiter. The reduction NODE COVER « FEEDBACK NODE SET was found by the Algorithms Seminar at the Cornell University Computer Science Department. The reduction NODE COVER * FEEDBACK ARC SET was found by Lawler and the writer, and Lawler discovered the reduction EXACT COVER « 3-DIMENSIONAL MATCHING REDUCIBILITY AMONG COMBINATORIAL PROBLEMS 101 The writer discovered that the exact cover problem was redu- cible to the directed traveling-salesman problem on a digraph in which the arcs have weight zero or one. Using refinements of the technique used in this construction, Tarjan showed that EXACT COVER « DIRECTED HAMILTON CIRCUIT and, independently, Lawler showed that NODE COVER © DIRECTED HAMILTON CIRCUIT . The reduction DIRECTED HAMILTON CIRCUIT = UNDIRECTED HAMILTON CIRCUIT was pointed out by Tarjan. Below we list three problems in automata theory and language theory to which every complete problem is reducible. These pro- blems are not known to be complete, since their membership in NP is presently in doubt. The reader unacquainted with automata and language theory can find the necessary definitions in Hopcroft and Uliman (1969). EQUIVALENCE OF REGULAR EXPRESSIONS INPUT: A pair of regular expressions over the alphabet {0,1} PROPERTY: The two expressions define the same language. EQUIVALENCE OF NONDETERMINISTIC FINITE AUTOMATA INPUT: A pair of nondeterministic finite automata with input alphabet {0,1} PROPERTY: The two automata define the same language. CONTEXT-SENSITIVE RECOGNITION INPUT: A context-sensitive grammar [ and a string x PROPERTY: x is in the language generated by [. First we show that SATISFIABILITY WITH AT MOST 3 LITERALS PER CLAUSE « EQUIVALENCE OF REGULAR EXPRESSIONS The reduction is made in two stages. In the first stage we con- struct a pair of regular expressions over an alphabet A = {uj,u2, +eesUnsty,iz,...,0,}. We then convert these regular expressions to regular expressions over {0,1}. The first regular expression is A™A* (more exactly, A is written out as (uytuyts+stuytiyt++++5,), and A" represents n copies of the expression for A concatenated together). ‘The se- cond regular expression is ant UD (atu, ata i=L * Uata.atu.a* a* U Oa, *u,A*) U & 90.) 102 RICHARD M. KARP where Ata A* if Dy = 02 A*S {ART QA _U ANG QA*GA* if = 01 Ua. RO IATT DALY Aa Sy 1 Vag OCD,) = 4 A*G{A%T 2G 3A* U'A*T]A%G 4O4T 2A" U AKG DAYG ANT ZA UA *596* a aya U A%030%010*0 50% U Ato 3A*0 20% yh if Dy = 91VoQVo, . Now let m be the least positive integer > log |A|, and let be a 1-1 function from A into {0,1}". Replace?each regular expression by a regular expression over {0,1}, by making the substitution a+ $(a) for each occurrence of each element of A. EQUIVALENCE OF REGULAR EXPRESSIONS = EQUIVALENCE OF NONDETERMINISTIC FINITE AUTOMATA There are standard polynomial-time algorithms [Salomaa (1969) ] to convert a regular expression to an equivalent nondeterministic automaton. Finally, we show that, for any L € MP, L & CONTEXT-SENSITIVE RECOGNITION + Suppose Lis recognized in time p() by a nondeterministic Turing machine. Then the following language L over the alphabet {0,1,#} is accepted by a nondeterministic linear bounded automaton which simulates the Turing machine: cia £4? A809), yP L8G0)) | xe. Hence L is context-sensitive and has a context-sensitive grammar F. Thus xeL iff Fy gP eGo), yp Ce (x)) is an acceptable input to CONTEXT-SENSITIVE RECOGNITION. We conclude by listing the following important problems in NP which are not known to be complete. GRAPH ISOMORPHISM INPUT: graphs G and G! PROPERTY: G is isomorphic to G', NONPRIMES INPUT: positive integer k PROPERTY: k is composite. LINEAR INEQUALITIES INPUT: integer matrix C, integer vector d PROPERTY: Cx > d has a rational solution. REDUCIBILITY AMONG COMBINATORIAL PROBLEMS 103 APPENDIX 1 Notation and Terminology Used in Problem Specification PROPOSITIONAL CALCULUS HyeMyreeee%_ WpyseeyU, propositional variables RaReeok Bd. complements of ae Lasse = propositional variables 8,04 literals €1,C2,.0.,Cp D1,D2,.+.,Dr clauses Ce S (x12 50+ eRe X1 E2006 Xp) Dy © {uy ,uz,.++,Upstys07,+..,0,,} A clause contains no complementary pair of literals. SCALARS, VECTORS, MATRICES 2 the positive integers = the set of p-tuples of positive integers 2p the set {0,1,...,p-1} kyW elements of Z the ordered pair (a,) (yj) 4 vectors with nonnegative integer components (c,,) € matrices with integer components tj GRAPHS AND DIGRAPHS SETS G = (N,A) G' = (V',a') finite graphs N,N! sets of nodes A,A' sets of arcs s,t,u,v nodes e,{u,v} arcs GX = {{u,v}| ue X and ve X} cut If s€X and teX, (X,X) isa s-t cut. wi A*+Z w':A'>*Z weight functions The weight of a subgraph is the sum of the weights of its arcs. H= (V,E) digraph V_ set of nodes, E set of arcs e, ares > the empty set the number of elements in the finite set S$ s ih {T,} {U,} finite families of finite sets

You might also like