Abstract:This paper describes a fast algorithm for transforming Legendre coefficients into Chebyshev coefficients, and vice versa. The algorithm is based on the fast multipole method and is similar to the approach described by Alpert and Rokhlin [SIAM J. Sci. Comput., 12 (1991)]. The main difference is that we utilise a modal Galerkin approach with Chebyshev basis functions instead of a nodal approach with a Lagrange basis. Part of the algorithm is a novel method that facilitates faster spreading of intermediate results through neighbouring levels of hierarchical matrices. This enhancement leads to a method that is approximately 20% faster to execute, due to less floating point operations. We also describe an efficient initialization algorithm that for the Lagrange basis is roughly 5 times faster than the original method for large input arrays. The described method has both a planning and execution stage that asymptotically require O(N) flops. The algorithm is simple enough that it can be implemented in 100 lines of vectorized Python code. Moreover, its efficiency is such that a single-threaded C implementation can transform 1000,000 coefficients in approximately 20 milliseconds on a new MacBook Pro M3, representing about 3 times the execution time of a well-planned (single-threaded) type 2 discrete cosine transform from FFTW (<a class="link-external link-http" href="http://www.fftw.org" rel="external noopener nofollow">this http URL</a>). Planning for the 1000,000 coefficients requires approximately 50 milliseconds.

On Fast Algorithms for the Evaluation of Legendre Coefficients

A new and efficient method for the computation of Legendre coefficients

On Error Bounds for Orthogonal Polynomial Expansions and Gauss-Type Quadrature

On the Optimal Convergence Rates of Chebyshev Interpolations for Functions of Limited Regularity

On the convergence rates of Legendre approximation.

New error bounds for Legendre approximations of differentiable functions

A faster multipole Legendre-Chebyshev transform

Asymptotics on Laguerre or Hermite polynomial expansions and their applications in Gauss quadrature

Optimal Error Estimates for Legendre Expansions of Singular Functions with Fractional Derivatives of Bounded Variation

POINTWISE ERROR ESTIMATES AND LOCAL SUPERCONVERGENCE OF JACOBI EXPANSIONS

General Nondiagonal Cubic Hermite-Padé Approximation to the Exponential Function

On the Optimal Rates of Convergence for Quadratures Derived from Chebyshev Points

On optimal convergence rates of Laguerre polynomial expansions for piecewise functions

An Improved Error Bound on Gauss Quadrature

On the Highly Accurate Evaluation of the Voigt/Complex Error Function with Small Imaginary Argument

The mysteries of the best approximation and Chebyshev expansion for the function with logarithmic regularities

Computing Chebyshev polynomials using the complex Remez algorithm

Accurate Evaluation of Polynomials in Legendre Basis

On Error Bounds of Filon-Clenshaw-Curtis Quadrature for Highly Oscillatory Integrals

Difft: A Fast and Accurate Algorithm for Fourier Transform Integrals of Discontinuous Functions

Optimal pointwise error estimates for piecewise functions expanded with Laguerre polynomials