Skip to main content
. Author manuscript; available in PMC: 2022 Mar 1.
Published in final edited form as: Can J Stat. 2021 Feb 15;49(1):203–227. doi: 10.1002/cjs.11606
Algorithm 1. Finding regularization parameters in griPEER
Input:matrices:Z,XandQ;vector:y;initial point:λ[0][λQ[0],λR[0]]T;stop criterion:δ>0;function which defines the density:ψ;k1do1.defineβ[k1]andb[k1]by solving:argminβ,b{2i=1n[yi(Xiβ+Zib)ψ(Xiβ+Zib)]+λQ[k1]bTQb+λR[k1]b22};2.θ[k1]Xβ[k1]+Zb[k1],W[k]diag([ψ(θ1[k1])]1,,[ψ(θn[k1])]1);3.definey[k]by puttingyi[k][ψ(θi[k1])]1(yiψ(θi[k1]))+θi[k1],fori=1,,n;4.P[k]IX(XTW1[k]X)1XTW1[k];5.y~[k]P[k]y[k],X[k]P[k]X,Z[k]P[k]Z;6.Ω[k]TZ[k]W[k]1Z[k],q[k]Z[k]TW[k]1y~[k];7.λ[k]argminλ¯0{ln(λQQ+λRIp+Ω[k])(λQQ+λRIp)1q[k]T(λQQ+λRIp+Ω[k])1q[k]};8.kk+1;while{λ[k]λ[k1]λ[k1]>δ}