TTMS29. Least squares means and their extensions
Speaker: Dr. Sejong Kim, Chungbuk National University, Korea
Email: [email protected]
Abstract: it is an interesting problem to define a barycenter (or mean) of a finite number of points in a metric space. A natural and canonical barycenter is the least squares mean, at which weighted sum of squared distances to given variables is minimized.
We study and compare mainly two kinds of the least squares mean, Cartan mean and Wasserstein mean, on the cone of positive definite Hermitian matrices. Finally we discuss their extensions to positive operators.