dekexu / computationbook

Example code for Understanding Computation

Understanding Computation example code

This is the example code for Understanding Computation, an O’Reilly book about computation theory. (Here’s a sample chapter.) Ruby 1.9 or 2.0 is required.

Right now it’s a pretty rough dump of code from the book. Each chapter has its own directory:

• Chapter 2: The Meaning of Programs
  • syntax
  • parser
  • small-step operational semantics
  • big-step operational semantics
  • denotational semantics
• Chapter 3: The Simplest Computers
  • deterministic and nondeterministic finite automata
  • regular expressions
• Chapter 4: Just Add Power
  • deterministic and nondeterministic pushdown automata
• Chapter 5: The Ultimate Machine
  • deterministic Turing machines
• Chapter 6: Programming with Nothing
  • FizzBuzz with procs
  • the lambda calculus
• Chapter 7: Universality is Everywhere
  • partial recursive functions
  • the SKI combinator calculus
  • Iota
  • tag systems
  • cyclic tag systems
• Chapter 8: Impossible Programs
• Chapter 9: Programming in Toyland
  • abstract interpretation: arithmetic with signs
  • type systems

Each directory contains definitions of the classes implemented in that chapter. There’s also a file named after each chapter (e.g. just_add_power.rb) that can be required to load all the code for that chapter.

For example:

$ irb -I.
>> require 'universality_is_everywhere'
=> true
>> identity = SKICall.new(SKICall.new(S, K), SKICall.new(K, K))
=> S[K][K[K]]
>> x = SKISymbol.new(:x)
=> x
>> expression = SKICall.new(identity, x)
=> S[K][K[K]][x]
>> while expression.reducible?; puts expression; expression = expression.reduce; end; puts expression
S[K][K[K]][x]
K[x][K[K][x]]
K[x][K]
x
=> nil

If you run bundle install to install Treetop, you can try out the parsers:

$ bundle exec irb -I.
>> require 'treetop'
=> true
>> Treetop.load('the_meaning_of_programs/parser/simple')
=> SimpleParser
>> require 'the_meaning_of_programs'
=> true
>> program = SimpleParser.new.parse('while (x < 5) { x = x * 3 }').to_ast
=> «while (x < 5) { x = x * 3 }»
>> program.reduce(x: Number.new(3))
=> [«if (x < 5) { x = x * 3; while (x < 5) { x = x * 3 } } else { do-nothing }», {:x=>«3»}]
>> program.evaluate(x: Number.new(3))
=> {:x=>«9»}
>> program.to_ruby
=> "-> e { while (-> e { (-> e { e[:x] }).call(e) < (-> e { 5 }).call(e) }).call(e); e = (-> e { e.merge({ :x => (-> e { (-> e { e[:x] }).call(e) * (-> e { 3 }).call(e) }).call(e) }) }).call(e); end; e }"
>> eval(program.to_ruby).call(x: 3)
=> {:x=>9}

$ bundle exec irb -I.
>> require 'treetop'
=> true
>> Treetop.load('programming_with_nothing/lambda_calculus/lambda_calculus')
=> LambdaCalculusParser
>> require 'programming_with_nothing'
=> true
>> two = LambdaCalculusParser.new.parse('-> p { -> x { p[p[x]] } }').to_ast
=> -> p { -> x { p[p[x]] } }
>> require 'universality_is_everywhere'
=> true
>> two.to_ski
=> S[S[K[S]][S[K[K]][I]]][S[S[K[S]][S[K[K]][I]]][K[I]]]
>> two.to_ski.to_iota
=> ɩ[ɩ[ɩ[ɩ[ɩ]]]][ɩ[ɩ[ɩ[ɩ[ɩ]]]][ɩ[ɩ[ɩ[ɩ]]][ɩ[ɩ[ɩ[ɩ[ɩ]]]]]][ɩ[ɩ[ɩ[ɩ[ɩ]]]][ɩ[ɩ[ɩ[ɩ]]][ɩ[ɩ[ɩ[ɩ]]]]][ɩ[ɩ]]]][ɩ[ɩ[ɩ[ɩ[ɩ]]]][ɩ[ɩ[ɩ[ɩ[ɩ]]]][ɩ[ɩ[ɩ[ɩ]]][ɩ[ɩ[ɩ[ɩ[ɩ]]]]]][ɩ[ɩ[ɩ[ɩ[ɩ]]]][ɩ[ɩ[ɩ[ɩ]]][ɩ[ɩ[ɩ[ɩ]]]]][ɩ[ɩ]]]][ɩ[ɩ[ɩ[ɩ]]][ɩ[ɩ]]]]
>> inc, zero = SKISymbol.new(:inc), SKISymbol.new(:zero)
=> [inc, zero]
>> expression = SKICall.new(SKICall.new(two.to_ski.to_iota, inc), zero)
=> ɩ[ɩ[ɩ[ɩ[ɩ]]]][ɩ[ɩ[ɩ[ɩ[ɩ]]]][ɩ[ɩ[ɩ[ɩ]]][ɩ[ɩ[ɩ[ɩ[ɩ]]]]]][ɩ[ɩ[ɩ[ɩ[ɩ]]]][ɩ[ɩ[ɩ[ɩ]]][ɩ[ɩ[ɩ[ɩ]]]]][ɩ[ɩ]]]][ɩ[ɩ[ɩ[ɩ[ɩ]]]][ɩ[ɩ[ɩ[ɩ[ɩ]]]][ɩ[ɩ[ɩ[ɩ]]][ɩ[ɩ[ɩ[ɩ[ɩ]]]]]][ɩ[ɩ[ɩ[ɩ[ɩ]]]][ɩ[ɩ[ɩ[ɩ]]][ɩ[ɩ[ɩ[ɩ]]]]][ɩ[ɩ]]]][ɩ[ɩ[ɩ[ɩ]]][ɩ[ɩ]]]][inc][zero]
>> expression = expression.reduce while expression.reducible?
=> nil
>> expression
=> inc[inc[zero]]

If you have any questions, please get in touch via Twitter or email. If you find any bugs or other programs with the code, please open an issue.