μKanren (MuKanren), which is also known as microKanren, is a declarative-relational programming language, that was originally presented in the book The Reasoned Schemer. In microKanren You would solve a problem by declaring it as an end goal Goal that consists of relations between variables and states, and then you let microKanren handle how to solve it.
This is a Python 3 port of Tom Stuart's μKanren implementation in Ruby.
If you find it difficult to play around with this library, then I urge you to watch Tom's video as it builds up a set of interesting ideas from simple constructs. If you're feeling adventurous, then head over to examples/, and try to build on them.
Differences to Tom's Ruby project:
- The List data structure is renamed to Sequence, since Python has a list data structure already (Ruby has arrays).
- The Ruby blocks are implemented as Python lambda functions.
- Enumerators are implemented as normal named functions which yield values (generators/iterators).
- State, Variable, Goal and Relations are all incapsulated in the micro module.
- Pair and Sequence are implemented in the sequence module.
All the examples from the talk can be found under the examples/ directory.
Suppose I have two variables: x, and y. If x and y multiplied gives me 24, what could x and y be?
empty_state = State() goal = Goal.with_variables(lambda x, y: multiply( x, y, peano.to_peano(24))) # encode 24 in peano's arithmetics states = goal.pursue_in(empty_state) states = itertools.islice(filter(None, states), 8) # take 8 states only # Print the results for state in states: print(list(map(peano.from_peano, state.results(2)))) # map x and y from peano to decimal
[1, 24] [2, 12] [3, 8] [4, 6] [6, 4] [8, 3] [12, 2] [24, 1]
Numbers in this microKanren implementation are encoded using Peano's arithmetic (see: peano module) when supplied to Goals and then decoded when returned as outputs.
This project was developed on OSX 10 and using Python 3.5.1, I haven't debugged it on other platforms. This is for educational purposes only. If you find any bugs or other problems with the code or even would like to contribute, please open an issue.