CMStatistics 2020: Start Registration
View Submission - CMStatistics
B1064
Title: Structural information-complexity transfer in log-Gaussian Cox processes Authors:  Adriana Medialdea - Universidad de Granada (Spain) [presenting]
Jose Miguel Angulo - University of Granada (Spain)
Jorge Mateu - University Jaume I (Spain)
Abstract: Log-Gaussian Cox processes establish a flexible class of spatial point pattern models which allow the representation of a wide variety of dependency effects. Information and complexity measures constitute a useful tool for assessment of stochasticity and structural richness of a system. In this framework, based on a box-counting approach from simulation, diverse scenarios of log-Gaussian Cox processes are analyzed under different parameter configurations of the Matern covariance model, with the aim of characterizing the structural information-complexity transfer from the underlying intensity field to the resulting point pattern. Generalized entropy, divergence and complexity measures are computed, enabling both global and local comparisons of the distributions corresponding to the two phases involved in the realization of the processes. Sensitivity with respect to varying values of deformation parameters is also assessed. Ordinary and relative diversity indices provide a direct interpretation of the structural enrichment from the intensity field to the subsequently generated point pattern.