William Byrd

Jon Doyle's 1978 PhD thesis proposal, 'Reflexive Interpreters,' describes the idea of a reflective problem-solving meta-interpreter with a sophisticated model of its own behavior, semantics, capabilities, and limitations, that would display "self-conscious" behavior. We investigate this idea of a reflexive interpreter, and connect it to ideas from McCarthy, Minsky, and Lenat.

We will investigate what weirdness occurs when John McCarthy's non-deterministic 'amb' operator is added to a LISP interpreter written in a relational style, in the miniKanren constraint logic programming language.

The code for this talk can be found at https://github.com/webyrd/relational-amb

In this talk I will give an introduction to miniKanren, a minimal embedded domain-specific language for constraint logic programming. miniKanren is designed to support relational programming, in which all code is written as mathematical relations, with no distinction between inputs and outputs. miniKanren uses a complete interleaving search, combined with unification and other constraints, to solve for variables representing unknown values. In effect, every miniKanren query is a program synthesis problem.

In this talk I will give an introduction to miniKanren, a minimal embedded domain-specific language for constraint logic programming. miniKanren is designed to support relational programming, in which all code is written as mathematical relations, with no distinction between inputs and outputs. miniKanren uses a complete interleaving search, combined with unification and other constraints, to solve for variables representing unknown values. In effect, every miniKanren query is a program synthesis problem.