Successive Standardization of Rectangular Arrays

TL;DR

This paper studies the mathematical properties and convergence behavior of successive standardization of rectangular arrays, demonstrating rapid convergence to a standardized form with mean zero and standard deviation one.

Contribution

It provides a rigorous proof of convergence for Efron's algorithm and clarifies the conditions under which it rapidly converges, extending previous work with corrected arguments.

Findings

Convergence occurs for almost all arrays with Lebesgue measure 1.

The standardized array has zero means and unit standard deviations for rows and columns.

Convergence is typically rapid, often exponential in the number of iterations.

Abstract

In this note we illustrate and develop further with mathematics and examples, the work on successive standardization (or normalization) that is studied earlier by the same authors in Olshen and Rajaratnam (2010) and Olshen and Rajaratnam (2011). Thus, we deal with successive iterations applied to rectangular arrays of numbers, where to avoid technical difficulties an array has at least three rows and at least three columns. Without loss, an iteration begins with operations on columns: first subtract the mean of each column; then divide by its standard deviation. The iteration continues with the same two operations done successively for rows. These four operations applied in sequence completes one iteration. One then iterates again, and again, and again,.... In Olshen and Rajaratnam (2010) it was argued that if arrays are made up of real numbers, then the set for which convergence of these successive iterations fails has Lebesgue measure 0. The limiting array has row and column means 0, row and column standard deviations 1. A basic result on convergence given in Olshen and Rajaratnam (2010) is true, though the argument in Olshen and Rajaratnam (2010) is faulty. The result is stated in the form of a theorem here, and the argument for the theorem is correct. Moreover, many graphics given in Olshen and Rajaratnam (2010) suggest that but for a set of entries of any array with Lebesgue measure 0, convergence is very rapid, eventually exponentially fast in the number of iterations. Because we learned this set of rules from Bradley Efron, we call it "Efron's algorithm". More importantly, the rapidity of convergence is illustrated by numerical examples.