Barboza, Francisco MárcioMedeiros, Guilherme Felipe de Oliveira2025-03-102025-03-102025-01-24MEDEIROS, Guilherme Felipe de Oliveira. Comparação empírica da convergência de algoritmos baseados em Gradiente Conjugado Não Linear. Orientador: Francisco Márcio Barboza. 2025. 31 f. Trabalho de Conclusão de Curso (Bacharelado em Sistemas de Informação) - Departamento de Computação e Tecnologia, Centro de Ensino Superior de Seridó (CERES), Universidade Federal do Rio Grande do Norte, Caicó, 2025.https://repositorio.ufrn.br/handle/123456789/62973This study explores the Nonlinear Conjugate Gradient Method, an extension of the Conjugate Gradient Method, with the aim of comparing the convergence of the Fletcher-Reeves and Polak-Ribière nonlinear optimization algorithms. The research conducts an empirical comparison between the two algorithms, evaluating their performance and accuracy using the Rosenbrock and Beale test functions. The comparison is crucial for identifying the most suitable algorithm for different optimization contexts and includes the analysis of method convergence, the behavior of the beta parameter over iterations, trajectories on contour plots, the number of iterations, computational time, and solution accuracy through absolute and relative errors. The results show that, for the proposed tests, the Fletcher-Reeves algorithm converges faster in terms of the number of iterations and computational time, while the Polak-Ribière algorithm demonstrates greater accuracy in the solutionGradiente Conjugado Não LinearMétodos iterativosOtimizaçãoNonlinear Conjugate GradientIterative methodsOptimizationbachelorThesisCNPQ::CIENCIAS EXATAS E DA TERRA::MATEMATICACNPQ::CIENCIAS EXATAS E DA TERRA::CIENCIA DA COMPUTACAO