Skip to contents

Distance between two symmetric, positive-definite matrices

spd.dist.Rd

Function implements several distance measures between symmetric, positive-definite matrices.

Usage

spd.dist(x, y, method = "euclidean", ...)

Arguments

x, y

Symmetric, positive-definite matrices

method

The distance measure. See details.

Details

Allowable distance measures are

  • "euclidean": The Frobenius norm of the difference x-y.

  • "logeuclidean": The Frobenius norm of the difference log(x)-log(y) in the tangent space..

  • "cholesky": The Frobenius norm of the difference between the cholesky factors of x and y. Not affinely invariant.

  • "riemannian": The Riemmanian distance proposed by Barachant, et al. (2013)

  • "stein": The square root of Jensen-Bregman log determinant divergence