Optimizing personal computer configurations with heuristic-based search methods

Abstract

Given the diversity and limited compatibility for personal computer hardware, obtaining an (sub-)optimal configuration for different usage restricted to some budget limits and other possible criteria can be challenging. In this paper, we firstly formulated these common configuration problems as discrete optimization problems to flexibly add in or modify users' requirements. More interestingly, we proposed two intelligent optimizers: a simple-yet-powerful beam search method and a min-conflict heuristic-based micro-genetic algorithm (MGA) to solve this real-life optimization problem. The heuristic-based MGA consistently outperformed the beam search and branch-and-bound method in most test cases. Furthermore, our work opens up exciting directions for investigation.