Practical aspects of the virtual noise convex optimum design approach for correlated responses

2025/04/24 by Hainy, Markus, Müller, Werner G., Pázman, Andrej
#62K05 #Computation (stat.CO) #FOS: Computer and information sciences #Methodology (stat.ME)

paper · doi:10.48550/arxiv.2504.17651

Abstract

In this paper we present several practically-oriented extensions and considerations for the virtual noise method in optimal design under correlation. First we introduce a slightly modified virtual noise representation which further illuminates the parallels to the classical design approach for uncorrelated observations. We suggest more efficient algorithms to obtain the design measures. Furthermore, we show that various convex relaxation methods used for sensor selection are special cases of our approach and can be solved within our framework. Finally, we provide practical guidelines on how to generally approach a design problem with correlated observations and demonstrate how to utilize the virtual noise method in this context in a meaningful way.

Related

• A convex approach to optimum design of experiments with correlated observations
• Correlated optimum design with parametrized covariance function. Justification of the Fisher information matrix and of the method of virtual noise.
• Optimum design for correlated processes via eigenfunction expansions
• Further aspects on an example of D-optimal designs in the case of correlated errors
• Optimal Design of Experiments Subject to Correlated Errors