Simple Models of Quantum Finite Automata
Total Page:16
File Type:pdf, Size:1020Kb
Masaryk University Faculty of Informatics Simple models of quantum finite automata Bachelor’s Thesis Martin Frian Brno, Spring 2018 Masaryk University Faculty of Informatics Simple models of quantum finite automata Bachelor’s Thesis Martin Frian Brno, Spring 2018 This is where a copy of the official signed thesis assignment and a copy ofthe Statement of an Author is located in the printed version of the document. Declaration Hereby I declare that this paper is my original authorial work, which I have worked out on my own. All sources, references, and literature used or excerpted during elaboration of this work are properly cited and listed in complete reference to the due source. Martin Frian Advisor: prof. RNDr. Jozef Gruska DrSc. i Acknowledgements I would like to thank my supervisor prof. RNDr. Jozef Gruska, DrSc. for his guidance and valuable advice. iii Abstract We present in a uniform way several models of classical finite automata and quantum finite automata. We have studied five main models of quantum finite automata in detail. Three other models of quantum finite automata were also briefly presented. We have also presented basic results concerning the power of each model, closure properties, mutual relations, and results concerning decision problems. iv Keywords finite automata, quantum finite automata, classical state, quantum state, regular languages, acceptance probability, quantum information processing, quantum mechanics v Contents Introduction 1 1 Finite automata 3 2 Basic tools of QIP 9 3 Quantum finite automata 13 3.1 Measure-many one-way quantum finite automata ...... 13 3.2 Measure-once one-way qunatum finite automata ...... 17 3.3 Two-way quantum finite automata .............. 20 3.4 Enhanced quantum finite automata .............. 22 3.5 Quantum finite automata with open time evolution ..... 24 3.6 Other models of quantum finite automata .......... 25 4 Conlusion 27 Bibliography 29 vii Introduction In the classical world, each regular language can be recognized (de- scribed) by a finite automaton. Finite automata model can be seenas an abstract machine with an input tape, on which an input is written, each input symbol in a single cell, read by the reading head. Finite automaton can be in exactly one of a fixed finite set of states at any moment of discrete time. Depending on the input, this machine model can go from one state to another and this is called the transition. Ac- cording to the used transition function we have the two most basic models of finite automata, deterministic finite automata (DFA) and non-deterministic finite automata (NFA). In fact, despite the different definitions of these two models, their computational power issame. Trivially, every DFA is NFA too and it was proven that for each NFA there exists a DFA recognizing the same language. However such a DFA could have exponentially larger number of states than the NFA under recognition. Another computational model of finite automata are probabilistic finite automata (PFA). It is well known, that PFA have same computa- tional power as DFA at a certain definition of acceptance. In the quantum world, there exists also various types of computa- tional models of finite automata. Informally, quantum finite automata (QFA) are quantum analog of PFA. However, instead of using clas- sical states, they use quantum states. Depending on used quantum resources, QFA might be more powerful and they might have better time or space complexity than PFA or DFA. So far there are several models of QFA that use simple quantum tools but generate only a subset of regular languages. There is also a model of QFA that generates exactly the whole class of regular languages but use the whole range of quantum resources. Both of these models are not fully satisfactory. Hence we need a model of QFA, which recognises exactly the whole class of regular languages using as minimum of quantum tools as possible. Therefore the aim of this thesis is to provide an overview of the current well known main models of quantum finite automata as well as basic results concerning their power, mutual relations and decidability of basic decision problems. 1 This thesis is divided into four chapters. In the first chapter we present classical deterministic finite automata, their properties and several basic results concerning decidability of decision problems. We also show modifications of deterministic finite automata, namely non- deterministic finite automata, two-way deterministic finite automata and probabilistic finite autoamta. In the second chapter we provide a brief introduction to basic tools widely used in a construction of quan- tum finite automata. In the third chapter we present in detail current known models of quantum finite automata, namely measure-many one-way quantum finite automata, measure-once one-way quantum finite automata, two-way quantum finite automata, enhanced quan- tum finite automata and quantum finite automata with opentime evolution. We also briefly present another models of quantum finite automata. 2 1 Finite automata Deterministic finite automata (DFA) are basic models of classical in- formation processing devices. They consist of: 1. a finite memory of so-called states. One of the states is said to be an initial state. A subset of states is said to be a set of accepting states; 2. a set of input symbols called the alphabet; 3. a transition function, which specifies the action of the automaton during computations. For each state and each input symbol determines the next state of the automaton. Formally such an automaton is defined as follows. Definition 1. A deterministic finite automaton A (DFA) is a 5-tuple (S, S, d, s0, Sa), where ∙ S is a finite set of states ∙ S is an input alphabet ∙ d : S × S ! S is a transition function ∙ s0 2 S is an initial state ∙ Sa ⊆ S is a set of accepting states Processing of an input (string) w by a DFA A is informally defined as follows. The automaton starts processing w in the initial state and reads as its input the leftmost symbol of w. The automaton then pro- ceeds in discrete time steps as follows until it gets into an accepting state or finishes processing of the whole input string: for the current state and next symbol of the input string it comes to a new state of its memory that is determined by its transition function and moves to read next symbol of its input string. This process is formally defined via extended transition function. Definition 2. An Extended transition function is a function dˆ : S × S* ! S satisfying the following three conditions: 3 1. Finite automata ∙ d(s, e) = s ∙ dˆ(s, a) = d(s, a) ∙ dˆ(s, aw) = dˆ(d(s, a), w) for any a 2 S, w 2 S+ and s 2 S. A language accepted by a DFA is defined as follows. Definition 3. Let A = (S, S, d, s0, Sa) be a DFA, the language accepted or recognized by A, denote as L(A), is * L(A) = fw 2 S j dˆ(s0, w) 2 Sag One of the key results of finite automata theory is that the class of languages accepted by DFA is the class of regular languages. They are the smallest set of languages containing all finite languages and closed under the following operations: ∙ union, ∙ concatenation, ∙ iteration. DFA are of large importance for several reasons: ∙ It is easy to determine in polynomial time the smallest DFA equivalent to a given A. This smallest DFA is always unique. ∙ Main decision problems, namely the emptiness problem and the membership problem are decidable in polynomial time. The emptiness problem is in addition easily reduced to the reacha- bility problem on graphs. ∙ Equivalence problems, such as the equivalence of two states or two DFA are decidable in polynomial time too. ∙ The above model of DFA is very robust. Several of its natural modifications, presented bellow, accept again exactly the class of regular languages. Here are the main variants of the above model of DFA. 4 1. Finite automata Non-deterministic finite automata Informally, transitions in non-deterministic finite automaton are based on non-deterministic choosing of the next state. The transition function maps each pair state and input symbol to the set of potential next states. The automaton nondeterministically chooses one of them. A formal definition of NFA is as follows. Definition 4. A nondeterministic finite automaton (NFA) is a 5-tuple (S, S, d, s0, Sa), where meaning of all elements is same as in the Defini- tion 1 except the transition function that is defined as follows d : S × S !P(S)1. Again, the transition function could be extended in a similar way like in the Definition 2. dˆ : Q × S* !P(Q) satisfies the following conditions: ∙ dˆ(s, e) = fsg ∙ dˆ(s, wa) = S d(p, a) p2dˆ(s,w) for any s 2 S, a 2 S and w 2 S*. The language L(A) accepted by a NFA A is defined as follows * L(A) = fw 2 S j dˆ(s0, w) \ Sa 6= Æg Two-way finite automata Another generalization of DFA are two-way deterministic finite au- tomata (2DFA). The only difference between 2DFA and DFA is that in 2DFA the automaton could move in both directions while reading an input word or can stay on the same position during a transition. Now we provide a formal definition of 2DFA. Definition 5. A deterministic two-way finite automaton (2DFA) A is a 5-tuple A = (S, S, d, s0, Sa) where meaning of each element is same as in Definition 1, except the transition function defined as follows d : S × S ! S × f , #, !g 1. Here P(S) denotes the powerset of the set S 5 1.