Paper/document/root.tex
changeset 82 14b12b5de6d3
parent 75 d63baacbdb16
child 83 f438f4dbaada
equal deleted inserted replaced
81:dc879cb59c9c 82:14b12b5de6d3
    34 \begin{abstract} 
    34 \begin{abstract} 
    35 There are numerous textbooks on regular languages. Nearly all of them 
    35 There are numerous textbooks on regular languages. Nearly all of them 
    36 introduce the subject by describing finite automata and 
    36 introduce the subject by describing finite automata and 
    37 only mentioning on the side a connection with regular expressions. 
    37 only mentioning on the side a connection with regular expressions. 
    38 Unfortunately, automata are a hassle for formalisations in HOL-based
    38 Unfortunately, automata are a hassle for formalisations in HOL-based
    39 theorem provers. The reason is that they need to be represented as graphs 
    39 theorem provers. The reason is that they need to be represented as graphs, 
    40 or matrices, neither of which can be defined as inductive datatype. Also 
    40 matrices or functions, none of which are inductive datatypes. Also 
    41 operations, such as disjoint unions of graphs, are not easily formalisiable 
    41 operations, such as disjoint unions of graphs, are not easily formalisiable 
    42 in HOL. In contrast, regular expressions can be defined conveniently 
    42 in HOL. In contrast, regular expressions can be defined conveniently 
    43 as datatype and a corresponding reasoning infrastructure comes for 
    43 as datatype and a corresponding reasoning infrastructure comes for 
    44 free. We show in this paper that a central result from formal 
    44 free. We show in this paper that a central result from formal 
    45 language theory---the Myhill-Nerode theorem---can be recreated 
    45 language theory---the Myhill-Nerode theorem---can be recreated