580 California St., Suite 400
San Francisco, CA, 94104
Key research themes
1. How can the Secant Method be generalized and what are the implications for convergence order and applicability?
This theme explores generalizations of the classic secant method by replacing linear interpolation with higher-degree polynomial interpolations or modifying update strategies. The investigations focus on theoretical convergence orders extending beyond the classical golden ratio, computational efficiency by requiring only one function evaluation per iteration, and applications including complex roots. Understanding these generalizations aids in broadening the scope of the secant method, enhancing convergence speed, and extending applicability to a wider class of nonlinear equations.
Key finding: Introduces a (k+1)-point iteration where the classical secant linear interpolant is replaced by a polynomial of degree k (k ≥ 2), requiring only one evaluation of the function per iteration; demonstrates that the method has... Read more
Key finding: Extends the generalized secant method to the complex plane for finding simple complex roots and establishes that the local convergence order s_k matches that of the real case, thus broadening the method’s applicability while... Read more
Key finding: Proposes a parameterized family of secant-like iterative methods, derived from second-degree polynomial approximations, which retain the super-linear convergence order (approximately 1.618) characteristic of the standard... Read more
Key finding: Derives exact Q- and R-orders of convergence of the classic secant method applied to functions where Newton’s method has exact order p; establishes the secant method’s convergence order as S(p) = (1/2)(−1 + sqrt(1 + 4p)),... Read more
2. How do modifications of the Secant Method improve convergence speed and computational efficiency for nonlinear equation and optimization problems?
This theme covers methodological advancements that utilize secant-type iterative ideas to enhance convergence rates from super-linear to cubic or super-quadratic, reduce computational overhead via diagonal or approximate Jacobian updates, or incorporate fractional calculus frameworks. These innovations aim for faster, more stable numerical solvers applicable in nonlinear root-finding, least squares problems, or optimization contexts, balancing accuracy, convergence speed, and computational cost.
Key finding: Develops a novel predictor-corrector iterative scheme where the predictor step is a secant method approximation, achieving cubic convergence with only one function and derivative evaluation per iteration; analytical... Read more
Key finding: Introduces the T-Secant method which incorporates a non-uniform scaling transformation to the secant equations and adds an extra approximation in each iteration, achieving a super-quadratic convergence rate (~2.618);... Read more
Key finding: Proposes a T-Secant iteration strategy allowing full-rank updates of approximate Jacobians by generating multiple independent trial approximations per step, improving from super-linear to super-quadratic or cubic convergence... Read more
Key finding: Designs a diagonal quasi-Newton method employing a weak secant equation and novel criteria to control diagonal Hessian approximations, ensuring positive definiteness and descent directions; global convergence with Armijo line... Read more
Key finding: Develops a fractional calculus-based extension of the Newton-Raphson method (Fractional Newton-Raphson), which enables iterates starting from real initial values to access the complex plane, thus effectively computing both... Read more
3. What are the practical comparisons and applications of the Secant Method versus other numerical methods in root-finding and parameter estimation?
This theme focuses on empirical and applied studies analyzing the secant method and its variants against other numerical methods like bisection, Brent, conjugate gradient, and finite difference methods in real-world contexts such as solar cell parameter estimation, image processing, and polynomial root finding. Insights concern efficiency, error convergence, robustness to initial guess selection, and computational stability, providing actionable guidance on method selection and adaptation based on problem characteristics.
Key finding: Compares bisection and secant methods for solving nonlinear equations modeling photovoltaic cells, showing the secant method requires fewer iterations and achieves faster convergence when estimating PV parameters across... Read more
Key finding: Develops a conjugate gradient method employing a modified secant condition parameter optimized via penalty functions incorporating both function and gradient information; demonstrates global convergence and achieves efficient... Read more
Key finding: Analyzes how initial guess selection affects root-finding outcomes for polynomials with multiple roots comparing Brent, bisection, and modified secant methods; finds the modified secant method more efficient in converging to... Read more
Key finding: Though centered on Euler’s methods rather than secant, discusses stability and step-size selection challenges in numerical ODE solvers; provides insights into numerical method misconceptions influenced by step size and... Read more
Key finding: Reviews Euler, Taylor, and Runge-Kutta methods for first-order ODEs with focus on accuracy and computational efficiency; through MATLAB and FORTRAN implementations, it demonstrates superior accuracy of RK4 compared to Euler... Read more