A Stochastic Conjugate Gradient Method for Approximation of Functions

TL;DR

This paper introduces a stochastic conjugate gradient method for function approximation that avoids covariance matrix computation, using stochastic sampling for inner products, with proven convergence and applications in power amplifier linearization.

Contribution

It presents a novel stochastic conjugate gradient approach that reduces computational complexity and provides convergence guarantees for function approximation tasks.

Findings

Method converges in probability

Avoids covariance matrix computation

Applicable to power amplifier linearization

Abstract

A stochastic conjugate gradient method for approximation of a function is proposed. The proposed method avoids computing and storing the covariance matrix in the normal equations for the least squares solution. In addition, the method performs the conjugate gradient steps by using an inner product that is based stochastic sampling. Theoretical analysis shows that the method is convergent in probability. The method has applications in such fields as predistortion for the linearization of power amplifiers.