Skip to content

Latest commit

 

History

History
20 lines (14 loc) · 1013 Bytes

README.md

File metadata and controls

20 lines (14 loc) · 1013 Bytes
Raw

covcomp

This is an R package for implementing the covariance completion algorithm in [1]. The variable Cov is a discretization of the partial covariance $K_{\Omega}$ in the form of a matrix and Omega is a matrix of 1s and 0s of the same dimensions as Cov which describes the domain $\Omega$ of $K_{\Omega}$.

  1. For manual choice of truncation parameters used in the function:
Compln(Cov, Omega, VectorOfTuning)

Here VectorOfTuning is a vector of integers which serve as truncation parameters in the matrix multiplication involved. Naturally, the length of this vector should be at least $m-1$ when we are dealing with an $m$-serrated domain $\Omega$.

  1. For automatic choice of truncation parameters in accordance with the fraction of variance explained (FVE) criterion:
ComplnFVE(Cov, Omega, FVE)

References

[1] Waghmare, K.G. and Panaretos, V.M., 2021. The Completion of Covariance Kernels. arXiv preprint arXiv:2107.07350. (https://arxiv.org/abs/2107.07350)