USVT {randnet}

estimates the network probability matrix by the improved universal singular value thresholding

Description

estimates the network probability matrix by the universal singular value thresholding of Chatterjee (2015), with the improvement mentioned in Zhang et. al. (2017).

Usage

USVT(A)

Arguments

A

adjacency matrix

Details

Instead of using the original threshold in Chatterjee (2015), the estimate is generated by taking the n^(1/3) leading spectral components. The method was mentioned in Zhang et. al. (2017) and suggested by an anonymous reviewer.

Value

The estimated probability matrix.

Author(s)

Tianxi Li and Can M. Le
Maintainer: Tianxi Li [e-mail]

References

S. Chatterjee. Matrix estimation by universal singular value thresholding. The Annals of Statistics, 43(1):177-214, 2015. Y. Zhang, E. Levina, and J. Zhu. Estimating network edge probabilities by neighbourhood smoothing. Biometrika, 104(4):771-783, 2017.

See Also

LSM.PGD

Examples

dt <- RDPG.Gen(n=600,K=2,directed=TRUE)


A <- dt$A


fit <- USVT(A)