4 ms·
Computing with Trees Not Numbers
I came across this meetup page (https://www.meetup.com/FP-Syd/events/vcqlmpyzdbkc/) where one of the topics is "Computing with Trees not Numbers". The description look intriguing as well.
It talk about tree calculus. Has that been around for long? I can't find much about it on the internet
Full Description:
Tree calculus is a calculus of combinations with a single ternary operator
(for building nodes) that has three reduction rules, according to whether
its first argument is a leaf, stem or fork. Its normal forms are the
unlabeled binary trees, which provide a better basis for computation than
the natural numbers: there is no need to encode and decode programs as
numbers, as the binary trees are simultaneously functions to be applied and
structures to be traversed. There are compositional, reduction-preserving
embeddings of the traditional models of computation into tree calculus. Also,
it supports self-interpretation without the need for quotation to convert
programs to trees. Tree calculus supports a type system that satisfies
subject reduction.
- tlack 8y agoSounds interesting. I found some of his other work which seems related: https://www.researchgate.net/scientific-contributions/70708830_Barry_Jay https://www.researchgate.net/scientific-contributions/707088...
- magicmouse 8y agowell, instead of trees of the classical S-expression type that you are referring to here, which are what LISP is built on. I suggest you investigate a graph database like Neo4J which is way more fun than lousy S-expressions. But what are in the trees are numbers or strings (which are arrays of numbers). Frankly all the computer has are bits which form numbers, so that is not something you are going to get away from.
- agumonkey 8y agoaren't binary digits conflated trees ?