[Paper Review] On the application of the variational iteration method to a prey and predator model with variable coefficients

Francisco M. Fernández|ArXiv.org|Aug 13, 2008

Mathematical and Theoretical Epidemiology and Ecology Models4 references15 citations

TL;DR

This paper critically evaluates the application of the variational iteration method (VIM) to a prey-predator model with time-dependent coefficients, demonstrating that while VIM produces increasingly accurate approximations, they fail to capture essential physical behaviors such as finite-time singularities. The study shows that simpler Padé approximants of the power series solution outperform VIM results in accuracy and correctly predict the location of poles, revealing VIM's limitations despite its widespread use in recent literature.

ABSTRACT

We discuss an amazing prey--predator model with variable coefficients, analyze its predictions and the accuracy of the variational iteration method used to solve the nonlinear equations.

Motivation & Objective

• To assess the validity and accuracy of the variational iteration method (VIM) in solving a nonlinear prey-predator system with time-varying coefficients.
• To challenge the scientific utility of recent VIM applications in population dynamics, arguing they often produce misleading or nonsensical results.
• To demonstrate that Padé approximants of the power series solution provide significantly better accuracy and physical consistency than VIM-derived expressions.
• To highlight the lack of attention to global dynamical behavior in VIM and homotopy perturbation method (HPM) studies, which focus narrowly on initial-time approximations.

Proposed method

• The author analyzes two specific examples of the VIM applied to a prey-predator model with variable coefficients, comparing VIM results to exact analytical solutions.
• The author computes the exact solutions for both examples, revealing unphysical behaviors such as infinite population growth and sign discontinuities.
• The author derives power series expansions of the exact solutions and constructs Padé approximants (e.g., [2/4], [3/4]) to test their accuracy and pole prediction capabilities.
• The author compares the VIM solutions (up to third-order approximations) with the Padé approximants, showing that Padé methods better capture the location of singularities.
• The author uses the Padé approximant [4/5](z) with z = t² to numerically converge toward the exact pole at t² = 1.386294364.
• The author critiques the VIM's failure to reproduce the pole structure and its tendency to incorrectly predict asymptotic decay to zero, unlike the exact solution's blow-up behavior.

Experimental results

Research questions

• RQ1 Does the variational iteration method (VIM) accurately reproduce the singular behavior of a prey-predator model with time-dependent coefficients?
• RQ2 Can Padé approximants of the power series solution outperform VIM in predicting the location and nature of finite-time singularities?
• RQ3 Why do VIM and homotopy perturbation method (HPM) results remain popular despite their failure to capture global dynamical features of nonlinear systems?
• RQ4 What are the implications of using VIM on models with unphysical initial conditions, such as negative population values?
• RQ5 Why is the scientific community persistently publishing results from VIM and related methods that are mathematically inconsistent with exact solutions?

Key findings

• The VIM solution x₃(t) fails to exhibit the exact solution’s pole at t_c = √(2 ln 2) ≈ 1.177, instead decaying to zero as t → ∞, which contradicts the exact behavior.
• The Padé approximant [2/4](t) = (8(t² - 6))/(t⁴ + 16t² - 24) predicts a pole at t² ≈ 1.3808, converging toward the exact pole t_c² = 1.386294364 with increasing order.
• Higher-order Padé approximants [3/4](z) and [4/5](z) converge to the exact pole with increasing accuracy, achieving agreement to 9 decimal places.
• The VIM solutions do not predict any pole and incorrectly model the long-term behavior as decaying to zero, while the exact solution diverges to infinity at t_c.
• The VIM results are only qualitatively acceptable near t = 0, but fail to describe the system’s true dynamics beyond the initial transient phase.
• The author concludes that VIM results are often mathematically inconsistent and that simpler Padé approximants of the power series are more accurate and physically meaningful than complex VIM expressions.