# An Evolutionary Framework for Automatic and Guided Discovery of Algorithms

## TL;DR

This paper introduces AAD, an evolutionary framework that uses Problem Guided Evolution to automatically discover complex algorithms, outperforming previous methods in generating high-complexity solutions across various problems.

## Contribution

The paper presents a novel PGE approach that guides evolutionary discovery of algorithms, enabling the synthesis of complex programs without relying solely on fitness functions.

## Key findings

- AAD can generate Python code for 29 array/vector problems.
- PGE enables discovery of algorithms with complexity comparable or higher than state-of-the-art.
- AAD demonstrates adaptability and innovative problem-solving in constrained environments.

## Abstract

This paper presents Automatic Algorithm Discoverer (AAD), an evolutionary framework for synthesizing programs of high complexity. To guide evolution, prior evolutionary algorithms have depended on fitness (objective) functions, which are challenging to design. To make evolutionary progress, instead, AAD employs Problem Guided Evolution (PGE), which requires introduction of a group of problems together. With PGE, solutions discovered for simpler problems are used to solve more complex problems in the same group. PGE also enables several new evolutionary strategies, and naturally yields to High-Performance Computing (HPC) techniques. We find that PGE and related evolutionary strategies enable AAD to discover algorithms of similar or higher complexity relative to the state-of-the-art. Specifically, AAD produces Python code for 29 array/vector problems ranging from min, max, reverse, to more challenging problems like sorting and matrix-vector multiplication. Additionally, we find that AAD shows adaptability to constrained environments/inputs and demonstrates outside-of-the-box problem solving abilities.