Unified Software Design Patterns for Simulated Annealing

TL;DR

This paper introduces a unified, object-oriented programming framework in Python for implementing and extending simulated annealing algorithms, promoting best practices, interoperability, and rapid iteration across diverse optimization problems.

Contribution

It presents a novel unified design pattern for simulated annealing that enhances flexibility, reusability, and scalability in optimization algorithm implementations.

Findings

Unified framework facilitates rapid prototyping of SA variants

Object-oriented design improves code interoperability and reproducibility

Blueprint for best practices in data-driven optimization libraries

Abstract

Any optimization algorithm programming interface can be seen as a black-box function with additional free parameters. In this spirit, simulated annealing (SA) can be implemented in pseudo-code within the dimensions of a single slide with free parameters relating to the annealing schedule. Such an implementation, however, necessarily neglects much of the structure necessary to take advantage of advances in computing resources and algorithmic breakthroughs. Simulated annealing is often introduced in myriad disciplines, from discrete examples like the Traveling Salesman Problem (TSP) to molecular cluster potential energy exploration or even explorations of a protein's configurational space. Theoretical guarantees also demand a stricter structure in terms of statistical quantities, which cannot simply be left to the user. We will introduce several standard paradigms and demonstrate how these can be "lifted" into a unified framework using object-oriented programming in Python. We demonstrate how clean, interoperable, reproducible programming libraries can be used to access and rapidly iterate on variants of Simulated Annealing in a manner which can be extended to serve as a best practices blueprint or design pattern for a data-driven optimization library.