Lanczos Tau method

In mathematics, the Lanczos tau method is a spectral method for the numerical solution of differential equations, named after Cornelius Lanczos, who first presented it in 1938. Unlike the Galerkin method, which requires the basis functions to satisfy the boundary conditions a priori, the tau method uses a set of orthogonal polynomials as basis functions. In the Tau method approach, the boundary conditions are imposed on the linear combination of basis functions, while the residual is made orthogonal to as many basis functions as possible — rather than to the entire set.

Overview

The tau method seeks an approximation to the solution of a differential equation by expanding the solution in terms of a set of basis functions, typically orthogonal polynomials such as Chebyshev or Legendre polynomials. The key idea is to require that the residual of the differential equation be orthogonal to as many basis functions as possible, while the boundary or auxiliary conditions are forced to be exactly satisfied.

Consider a differential equation of the form

Dy(x)=f(x),x∈(a,b)

subject to ν appropriate boundary or auxiliary conditions

gi(y)=si,i=1,…,ν.

Here, D is a differential operator that can be, at least approximately, written in the form

Dy=∑i=0νuiddxiy,

ν is the differential problem order, f,gi,si,i=1,…,ν and ui,i=0,…,ν are given. For the tau method to be applied, it is assumed that the solution admits a generalised Fourier series

y(x)=∑j=0∞ajPj(x)

where 𝒫={Pj}j=0∞ is a set of square-integrable orthogonal functions defined on [a,b], and aj are unknown coefficients. Usually 𝒫 is an orthogonal polynomial basis.

The tau formulation

In the tau method, for some integer n>ν, an approximation to y is found in the subspace ℙn spanned by {Pj}j=0n, in the form

yn(x)=∑j=0naj(n)Pj(x)

where the coefficients aj(n) are found by requiring that the boundary conditions hold

gi(yn)=si,i=1,…,ν

and using the remaining n+1−ν degrees of freedom to impose that Dyn agrees with Dy as far as possible.The last condition means that the residual is orthogonal to each Pj, j=0,…,n−ν or, equivalently, forces the residual to be an order n+1−ν Fourier series in 𝒫

τ≡Dy−Dyn=∑j=n+1−ν∞τjPj

In case all coefficients ui in D are polynomials, and if diis the degree of ui, then the residual τ is a polynomial of degree n+max⁡{di−i, i=0,…,ν}.

Operational approach of the tau method

In 1981, Ortiz and Samara proposed an operational approach to the tau method based on operational matrices of differential operators. For a fixed n>ν, let 𝒫n=[P0,P1,…,Pn] be a basis for the subspace ℙn, and let 𝖺n=[a0(n),a1(n),…,an(n)]T be the vector of the Fourier coefficients of yn, then the approximate solution can be written as yn=𝒫n𝖺n.

Let x𝒫n=[xP0,xP1,…,xPn] denote the multiplication of basis elements by x, and let 𝒫'n=[P'0,P'1,…,P'n] denote differentiation of basis elements. These operations can be represented as matrix operations by

x𝒫n=𝒫n𝖬and𝒫'n=𝒫n𝖭

Matrices 𝖬 and 𝖭 can be evaluated from properties of 𝒫n. If 𝒫n is an orthogonal polynomial sequence, then a three-term recurrence relation holds:

{xPj=αjPj+1+βjPj+γjPj−1,j=0,1,…,nP0=1;P−1=0.

Then 𝖬 is the Jacobi matrix 𝖬=[μi,j]i,j=0n+1,n with

μi,j={γj,i=j−1,βj,i=j,αj,i=j+1,0,|i−j|>1

and 𝖭=[ηi,j]i,j=0n,n where ηi,j are the connection coefficients in

P'j=∑i=0j−1ηi,jPi

From matrices 𝖬 and 𝖭, the action of the differential operator over polynomials is translated to the action of the algebraic operator, over the vector of Fourier coefficients, defined by the operational matrix

𝖣=∑i=0νui(𝖬)𝖭i,

in the sense that 𝖣𝖺n=𝖿, where 𝖿=[f0,f1,…,fn] is the vector of Fourier coefficients of the projection of f onto ℙn. If, for the coefficients of D, we have

ui=∑j=0diui,jPj

at least approximately, then

ui(𝖬)=∑j=0diui,jPj(𝖬).

The tau matrix

The action of boundary conditions over the base 𝒫n is represented by the ν×(n+1) matrix 𝖦=[gi,j]i,j=0ν−1,n with

gi,j=gi+1(Pj)

and the vector 𝗌=[s1,s2,…,si]T. Then yn=𝒫n𝖺n, is an approximate solution to

{Dy(x)=f(x),x∈(a,b)gi(y)=si,i=1,…,ν

in the Tau sense if

𝖳n𝖺n=𝖻n

where

𝖳n=[𝖦𝖣n],and𝖻n=[𝗌𝖿n],

𝖣n is matrix 𝖣 truncated to its first n+1−ν rows and 𝖿n is vector 𝖿 truncated to its first n+1−ν elements.

See also

Galerkin method