Bias and synergy in the self-consistent approach of data analysis of ion beam techniques

TL;DR

This paper investigates how combining multiple ion beam analysis techniques through a self-consistent, multi-objective optimization approach can reduce bias, improve accuracy, and enhance information extraction in data analysis.

Contribution

It introduces a systematic evaluation of biases in the weighted-sum method for data fusion in ion beam analysis, demonstrating convergence to true values and the importance of measurement accuracy.

Findings

Bias converges to true value with better statistics.

Joint analysis inherits the most accurate measurement's precision.

Some measurement combinations provide more valuable information.

Abstract

Using multiple ion beam analysis measurements, or techniques, combined with self-consistent data processing, generally allows extracting more (or more accurate) information from the measurements than processing separately data from single measurements. Solving ambiguities, improving the final depth resolution, defining constraints and extending applicability are the main strengths of the data-fusion approach. It basically consists in formulating a multi-objective minimization problem that can be tackled by the adoption of the weighted-sum method. A simulation study is reported in order to evaluate the systematic error inserted in the analysis by the choice of a specific objective function, or even by the weights or normalization adopted in the weighted-sum method. We demonstrate that the bias of the analyzed objective functions asymptotically converges to the true value for better statistics. We also demonstrate that the joint analysis inherits the accuracy of the most accurate measurement, establishing a rank of information content, where some combinations of measurements are more valuable than others, i.e. when processed together they provide more information by means of a better constraint for the multi-objective optimization.