comp20121 the implementation and power of computer
play

COMP20121 The Implementation and Power of Computer Languages Power - PowerPoint PPT Presentation

COMP20121 The Implementation and Power of Computer Languages Power Part http://www.cs.man.ac.uk/ petera/2121/index.html . Peter Aczel room: CS2.52, tel: 56155 petera@cs.man.ac.uk email: Department of Computer Science, University of


  1. COMP20121 The Implementation and Power of Computer Languages ‘Power’ Part http://www.cs.man.ac.uk/ ∼ petera/2121/index.html . Peter Aczel room: CS2.52, tel: 56155 petera@cs.man.ac.uk email: Department of Computer Science, University of Manchester COMP20121 - Section 0 – p.1/307

  2. COMP20121, ‘power’ part: section 1 lectures 2,3: Finite automata for regular languages LECTURE FOUR - Some properties of Regular Languages - Start of section 2: Context-free Grammars COMP20121 - Section 1 – p.74/307

  3. Properties of regular languages Proposition • Every finite language is regular. COMP20121 - Section 1 – p.75/307

  4. Properties of regular languages Proposition • Every finite language is regular. • If L is a regular language then so are L n (for all n ∈ N ), L ∗ , L and L R . COMP20121 - Section 1 – p.75/307

  5. Properties of regular languages Proposition • Every finite language is regular. • If L is a regular language then so are L n (for all n ∈ N ), L ∗ , L and L R . • If L and L ′ are regular languages then so are L ∪ L ′ , L ∩ L ′ and L · L ′ . COMP20121 - Section 1 – p.75/307

  6. Proofs—I We first show that every finite language is regular: COMP20121 - Section 1 – p.76/307

  7. Proofs—I We first show that every finite language is regular: If { α 1 , α 2 , . . . , α n } is a finite language then α 1 | α 2 | · · · | α n is a pattern for the language. COMP20121 - Section 1 – p.76/307

  8. Proofs—II If L is regular then so are L n and L ∗ : COMP20121 - Section 1 – p.77/307

  9. Proofs—II If L is regular then so is L n : For if p is a pattern for L then ? is a pattern for L n , COMP20121 - Section 1 – p.77/307

  10. Proofs—II If L is regular then so is L n : For if p is a pattern for L then pp · · · p is a � �� � n pattern for L n , and COMP20121 - Section 1 – p.77/307

  11. Proofs—II If L is regular then so is L n : For if p is a pattern for L then pp · · · p is a � �� � n pattern for L n , and ? is one for L ∗ . COMP20121 - Section 1 – p.77/307

  12. Proofs—II If L is regular then so is L n : For if p is a pattern for L then pp · · · p is a � �� � n pattern for L n , and p ∗ is one for L ∗ . COMP20121 - Section 1 – p.77/307

  13. Proofs—II If L is regular then so is its complement, L . This follows from Assessed Exercise 1(b) and Theorem 1.5. COMP20121 - Section 1 – p.77/307

  14. Proofs—II If L is regular then so is L R . This is an exercise. COMP20121 - Section 1 – p.77/307

  15. Proofs—III If L and L ′ are regular languages then so is L ∪ L ′ : COMP20121 - Section 1 – p.78/307

  16. Proofs—III If L and L ′ are regular languages then so is L ∪ L ′ : If p is a pattern for L and p ′ one for L ′ then ? is a pattern for L ∪ L ′ . COMP20121 - Section 1 – p.78/307

  17. Proofs—III If L and L ′ are regular languages then so is L ∪ L ′ : If p is a pattern for L and p ′ one for L ′ then p | p ′ is a pattern for L ∪ L ′ . COMP20121 - Section 1 – p.78/307

  18. Proofs—III If L and L ′ are regular languages then so is L · L ′ : COMP20121 - Section 1 – p.78/307

  19. Proofs—III If L and L ′ are regular languages then so is L · L ′ : If p is a pattern for L and p ′ one for L ′ then ? is a pattern for L · L ′ . COMP20121 - Section 1 – p.78/307

  20. Proofs—III If L and L ′ are regular languages then so is L · L ′ : If p is a pattern for L and p ′ one for L ′ then pp ′ is a pattern for L · L ′ . COMP20121 - Section 1 – p.78/307

  21. Proofs—III If L and L ′ are regular languages then so is L ∩ L ′ . This is again an exercise. COMP20121 - Section 1 – p.78/307

  22. Long words in regular languages Assume that a DFA has n states. Then every word of length at least n will go through (at least) one state twice. COMP20121 - Section 1 – p.79/307

  23. Long words in regular languages Assume that a DFA has n states. Then every word of length at least n will go through (at least) one state twice. Very long word x 1 x n . . . . . . state n + 1 state 1 COMP20121 - Section 1 – p.79/307

  24. Long words in regular languages Assume that a DFA has n states. Then every word of length at least n will go through (at least) one state twice. Length of word ≥ n , where n is the number of states . . . i . . . i . . . COMP20121 - Section 1 – p.79/307

  25. Long words in regular languages Assume that a DFA has n states. Then every word of length at least n will go through (at least) one state twice. Length of word ≥ n , where n is the number of states . . . i . . . . . . . . . COMP20121 - Section 1 – p.79/307

  26. The Pumping Lemma Lemma (The Pumping Lemma) Let L be an infinite regular language. Then there exists a natural number n > 0 such that every word α of L consisting of at least n characters contains a ‘pumping section’ in the following sense. COMP20121 - Section 1 – p.80/307

  27. The Pumping Lemma The word α can be split up into three parts, say λ , µ , and ν such that • α = λµν and µ � = ǫ ; • the length of λµ is at most n and so that all words of the form λµ k ν, for k ∈ N , belong to L . COMP20121 - Section 1 – p.81/307

  28. The Pumping Lemma–II The processing of α begins The processing of α ends COMP20121 - Section 1 – p.82/307

  29. The Pumping Lemma–II The processing of α begins The processing of α ends p COMP20121 - Section 1 – p.82/307

  30. The Pumping Lemma–II The processing of α begins The processing of α ends p λ is processed COMP20121 - Section 1 – p.82/307

  31. The Pumping Lemma–II The processing of α begins The processing of α ends µ is processed p λ is processed COMP20121 - Section 1 – p.82/307

  32. The Pumping Lemma–II The processing of α begins The processing of α ends µ is processed p ν is processed λ is processed COMP20121 - Section 1 – p.82/307

  33. Processing λν The processing of λν begins The processing of α ends µ is processed p ν is processed λ is processed COMP20121 - Section 1 – p.83/307

  34. Processing λν The processing of λν begins The processing of α ends µ is processed p ν is processed λ is processed COMP20121 - Section 1 – p.83/307

  35. Processing λµ 2 ν The processing of λν begins The processing of α ends µ is processed p ν is processed λ is processed COMP20121 - Section 1 – p.84/307

  36. Processing λµ 2 ν The processing of α begins The processing of α ends µ is processed p ν is processed λ is processed COMP20121 - Section 1 – p.84/307

  37. Processing λµ 2 ν The processing of α begins The processing of α ends µ is processed p ν is processed λ is processed COMP20121 - Section 1 – p.84/307

  38. Processing λµ 2 ν The processing of λν begins The processing of α ends µ is processed p ν is processed λ is processed COMP20121 - Section 1 – p.84/307

  39. Pumping Lemma: Example The language L = { 0 i 1 i | i ∈ N } is not regular. • Assume n is as in the Pumping Lemma. COMP20121 - Section 1 – p.85/307

  40. Pumping Lemma: Example The language L = { 0 i 1 i | i ∈ N } is not regular. • Assume n is as in the Pumping Lemma. • Consider 0 n 1 n ; assume λ , µ , ν as in the Pumping Lemma. COMP20121 - Section 1 – p.85/307

  41. Pumping Lemma: Example • Since the length of λµ is at most n and µ � = ǫ , we know that n n � �� � � �� � 00 · · · · · · 1 · · · 1 . 00 � �� � ���� µ λ COMP20121 - Section 1 – p.86/307

  42. Pumping Lemma: Example • Since the length of λµ is at most n and µ � = ǫ , we know that n n � �� � � �� � 00 · · · · · · 1 · · · 1 . 00 � �� � ���� µ λ • But then λν = λµ 0 ν is not in L : <n n � �� � � �� � 00 · · · 1 · · · 1 . 0 � �� � λ COMP20121 - Section 1 – p.86/307

  43. Section 1 summary • Regular languages are describable via patterns (regular expressions). COMP20121 - Section 1 – p.87/307

  44. Section 1 summary • Regular languages are describable via patterns (regular expressions). • Regular languages are precisely the languages accepted by finite deterministic automata and COMP20121 - Section 1 – p.87/307

  45. Section 1 summary • Regular languages are describable via patterns (regular expressions). • Regular languages are precisely the languages accepted by finite deterministic automata and • these are precisely the languages accepted by finite non-deterministic automata. COMP20121 - Section 1 – p.87/307

  46. Section 1 summary • Matching a word to a pattern is a recursive process, while checking acceptance with a deterministic automaton is straightforward. COMP20121 - Section 1 – p.88/307

  47. Section 1 summary • Matching a word to a pattern is a recursive process, while checking acceptance with a deterministic automaton is straightforward. • The Pumping Lemma says that infinite regular languages contain strings where a ‘pumping section’ can be repeated arbitrarily often. COMP20121 - Section 1 – p.88/307

  48. Section 1 summary • Regular languages are not powerful enough for the parsing of programming languages. COMP20121 - Section 1 – p.89/307

Recommend


More recommend