A class of iterative methods with third-order convergence to solve nonlinear equations

作者:

Highlights:

摘要

Algebraic and differential equations generally co-build mathematical models. Either lack or intractability of their analytical solution often forces workers to resort to an iterative method and face the likely challenges of slow convergence, non-convergence or even divergence. This manuscript presents a novel class of third-order iterative techniques in the form of xk+1=gu(xk)=xk+f(xk)u(xk) to solve a nonlinear equation f with the aid of a weight function u. The class currently contains an invert-and-average (gKia), an average-and-invert (gKai), and an invert-and-exponentiate (gKe) branch. Each branch has several members some of which embed second-order Newton's (gN), third-order Chebychev's (gC) or Halley's (gH) solvers. Class members surpassed stand-alone applications of these three well-known methods. Other methods are also permitted as auxiliaries provided they are at least of second order. Asymptotic convergence constants are calculated. Assignment of class parameters to non-members carries them to a common basis for comparison. This research also generated a one-step “solver” that is usable for post-priori analysis, trouble shooting, and comparison.

论文关键词:Algebraic equation solvers,Iterative methods,Fixed-point iterations,Simulation,Convergence order,Direct substitution,Partial substitution,Newton's method,Halley's method,Nonlinear equations,Convergence acceleration

论文评审过程:Received 25 July 2006, Revised 3 February 2007, Available online 21 February 2007.

论文官网地址:https://doi.org/10.1016/j.cam.2007.02.001