Non Periodic Trigonometric Polynomial Approximation

TL;DR

This paper introduces a novel non-periodic trigonometric polynomial basis for approximating non-periodic functions, offering spectral accuracy and improved numerical integration over traditional polynomial methods.

Contribution

It proposes replacing polynomial basis functions with sine-based functions, overcoming polynomial approximation limitations and enhancing spectral accuracy and quadrature precision.

Findings

Spectral accuracy in approximating analytic functions

Higher numerical integration accuracy than Legendre quadrature

Effective approximation for non-periodic functions

Abstract

The suitable basis functions for approximating periodic function are periodic, trigonometric functions. When the function is not periodic, a viable alternative is to consider polynomials as basis functions. In this paper we will point out the inadequacy of polynomial approximation and suggest to switch from powers of $x$ to powers of $\sin(px)$ where $p$ is a parameter which depends on the dimension of the approximating subspace. The new set does not suffer from the drawbacks of polynomial approximation and by using them one can approximate analytic functions with spectral accuracy. An important application of the new basis functions is related to numerical integration. A quadrature based on these functions results in higher accuracy compared to Legendre quadrature.