Dependent unit vectors
Some models are given for dependent unit vectors in \(R^ p\). To construct a bivariate distribution with uniformly distributed marginals, a 2p-dimensional normally distributed random vector with mean 0 and with a suitable covariance matrix is projected onto 2 unit vectors. To construct a bivariate distribution with specified axially symmetric marginals, a unit vector in \(R^ p\) is split into two pieces, the part along the axis (a number lying in [-1,1]) and the part orthogonal to the axis (a unit vector in \(R^{p-1})\). Then models for dependence are imposed on each piece separately. Some suitable measures of correlation for these models are proposed and an example is given.
- Some novel models of distributions over the unit square
- Tractable circula densities from Fourier series
- Inferences based on a bivariate distribution with von Mises marginals
- A weighted spatial median for clustered data
- Recent advances in directional statistics
- A distribution for dependent unit vectors
- An affine‐invariant multivariate sign test for cluster correlated data
- Models for circular data from time series spectra
- On the multivariate spatial median for clustered data
- A distribution for a pair of unit vectors generated by Brownian motion
- Symmetric distributions for dependent unit vectors
- On \(p\times 1\) dependent random variables having each \((p - 1)\times 1\) sub-vector made up of iid observations with examples
- A test for independence of two sets of variables when the number of variables is large relative to the sample size
This page was built for publication: Dependent unit vectors
Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q3324852)