Rayleigh–Ritz method

Rayleigh–Ritz method. The Rayleigh–Ritz method is a fundamental variational method in applied mathematics and computational physics for approximating the eigenvalues and eigenfunctions of self-adjoint operators, particularly those arising from Sturm–Liouville theory. It transforms an infinite-dimensional eigenvalue problem into a finite-dimensional generalized eigenvalue problem by projecting the operator onto a carefully chosen finite-dimensional subspace, known as the trial space. The method is named for its progenitors, Lord Rayleigh and Walther Ritz, who developed its theoretical foundations in the late 19th and early 20th centuries.

● Overview

The core principle of the Rayleigh–Ritz method is the minimization of the Rayleigh quotient, a scalar functional derived from the underlying operator. This approach is deeply rooted in the calculus of variations and provides a systematic technique for obtaining upper bounds to the lowest eigenvalues of a system. It is extensively used in fields such as quantum mechanics, structural analysis, and acoustics to solve problems where exact analytical solutions are intractable. The method's efficacy relies on the selection of trial functions that satisfy essential boundary conditions and span a subspace that can effectively capture the behavior of the true eigenfunctions.

● Mathematical formulation

Consider a self-adjoint operator \( \mathcal{L} \) defined on a Hilbert space \(\mathcal{H}\), with an associated inner product \(\langle \cdot, \cdot \rangle\). The eigenvalue problem is \(\mathcal{L} u = \lambda u\). The Rayleigh quotient \( R[u] \) is defined as \( R[u] = \frac{\langle u, \mathcal{L} u \rangle}{\langle u, u \rangle} \) for any admissible function \( u \). According to the min-max theorem, the minimum of \( R[u] \) over all admissible functions gives the smallest eigenvalue. In the Rayleigh–Ritz procedure, one approximates the solution \( u \) as a linear combination \( u_n = \sum_{i=1}^n c_i \phi_i \), where the \( \phi_i \) are linearly independent trial functions from a subspace of \(\mathcal{H}\). Substituting this ansatz into the Rayleigh quotient leads to a generalized eigenvalue problem of the form \( \mathbf{A} \mathbf{c} = \lambda \mathbf{B} \mathbf{c} \), where the matrices \(\mathbf{A}\) and \(\mathbf{B}\) have entries \( A_{ij} = \langle \phi_i, \mathcal{L} \phi_j \rangle \) and \( B_{ij} = \langle \phi_i, \phi_j \rangle \).

● Procedure and algorithm

The implementation follows a defined sequence. First, a set of \( n \) trial functions \( \{ \phi_i \} \) is selected, typically requiring them to satisfy essential boundary conditions of the problem and to be linearly independent. Second, the matrices \(\mathbf{A}\) and \(\mathbf{B}\), often called the stiffness matrix and mass matrix in finite element method contexts, are computed by evaluating the necessary inner products. Third, the resulting generalized eigenvalue problem is solved using standard numerical linear algebra techniques, such as those found in LAPACK. The computed eigenvalues \( \tilde{\lambda}_k \) provide approximations to the true eigenvalues \( \lambda_k \), and the corresponding eigenvectors \( \mathbf{c}_k \) yield approximations to the eigenfunctions via the linear combination. The accuracy generally improves as the dimension \( n \) of the trial space increases.

● Applications

The method has profound applications across scientific and engineering disciplines. In quantum mechanics, it is used to approximate energy levels of systems like the hydrogen atom or in Hartree–Fock method calculations. Within structural engineering, it is fundamental to estimating natural frequencies and mode shapes in finite element analysis of buildings, bridges, and aircraft wings. It is also employed in electromagnetism for solving Maxwell's equations in resonant cavities and in fluid dynamics for stability analysis. The Galerkin method, a broader projection technique, can be viewed as an extension of this variational approach for more general operators.

● Advantages and limitations

A primary advantage is that it provides rigorous upper bounds to the lowest eigenvalues, a consequence of the min-max theorem. The approximations converge to the exact eigenvalues as the trial space becomes complete. Computationally, it reduces a continuous problem to a manageable matrix eigenvalue problem. However, its accuracy is heavily dependent on the choice of trial functions; poor choices can lead to slow convergence or inaccurate results. The quality of the upper bound for higher eigenvalues is not guaranteed. Furthermore, the method is primarily applicable to self-adjoint problems with a variational formulation, limiting its direct use for non-self-adjoint or nonlinear operators without significant modification.

● Variants and related methods

Several important techniques are directly related or derived from the core principles. The Galerkin method generalizes the projection idea to a wider class of problems, not necessarily self-adjoint. The finite element method can be interpreted as a Rayleigh–Ritz method with piecewise polynomial trial functions defined over a mesh. The Ritz method often refers to the application of the same variational principle to static problems, minimizing potential energy. Other related approaches include the Kantorovich method, the Trefftz method, and spectral methods, which use global basis functions like Chebyshev polynomials or Fourier series as the trial space.

Category:Numerical analysis Category:Variational methods Category:Computational physics