Articles
Comparative Analysis of Population Control Strategies in Genetic Algorithms Applied to the N-Queens Problem
An undergraduate thesis comparing full replacement, partial replacement, elitism and age-based population control, using a common fitness evaluation budget for the N-Queens problem.
This page introduces the undergraduate thesis article by Vinícius Rubin and Igor Yepes, from Instituto Federal Farroupilha — Campus Frederico Westphalen. The full scientific paper is available in Portuguese; this page provides an English overview.
Read the full paper in PDF (Portuguese) · Explore the repository and LaTeX sources
Problem and objective
The N-Queens problem consists of placing N queens on an N × N board so that none attack each other. The study investigates how different policies for preserving and replacing individuals affect convergence, solution quality and computational cost in genetic algorithms applied to this problem.
A permutation representation eliminates row and column conflicts. The fitness function evaluates diagonal conflicts, seeking a configuration without attacks.
Strategies compared
| Strategy | Population control policy |
|---|---|
| Full replacement | Replace the entire population each generation. |
| Partial replacement | Replace 25% of the population. |
| Elitism | Preserve the best 10% of individuals. |
| Age-based control | Replace individuals according to lifespan, inspired by GAVaPS, with a variable population size. |
Experimental protocol
The comparison keeps genetic operators constant and uses the same fitness evaluation budget across policies:
- An initial population of 100 individuals.
- Binary tournament selection and OX crossover.
- Swap mutation at a rate of 10%.
- 100 independent runs per configuration.
- Each run ends when it finds a valid solution or reaches 2 × P × N evaluations, where P is the initial population size and N is the number of queens.
The analysis considers success rate, solution quality, evaluations consumed, generations and execution time. The paper describes the experimental tool as a Python implementation using NumPy, Numba and Matplotlib.
Main findings
Under the evaluated conditions, age-based control offered a good balance between quality and efficiency for intermediate instances. Elitism performed better on larger instances, at a higher time cost. Full and partial replacement lost performance as the problem size increased.
These results highlight the influence of population renewal policies on evolutionary search. The original PDF contains the charts and full discussion.
Available materials and limitations
The repository contains the paper PDF, LaTeX sources, BibTeX bibliography and charts. Experiment code, seeds and raw data are not included; compiling the document rebuilds the paper but does not reproduce the experiments.
The conclusions apply to the problem and parameters studied. The paper identifies the use of a single problem, fixed parameters and the lack of in-depth formal statistical analysis as limitations.
Original paper
Original title: Análise Comparativa de Estratégias de Controle Populacional em Algoritmos Genéticos Aplicados ao Problema das N-Rainhas.
Authors: Vinícius Rubin and Igor Yepes.
Institution: Instituto Federal Farroupilha — Campus Frederico Westphalen.
Read the original PDF (Portuguese) and browse the research materials on GitHub.
The date displayed on this page refers to its addition to the website on October 14, 2026.