/GraphicalLasso

Graphical Lasso and EM algorithm on confounding model

Primary LanguagePython

You can first view Question.pdf and hw2_YuanLiu.pdf to know the questions and answers.

This is the PGM homework2

``Emalgorithm.py'' This algorithm is an EM Algorithm for Confounded Heterogenous Data y \sim X\beta +Zu+\epsilon u \sim N(0, I \sigma_u^2) Y \sim N(X\beta, ZZ^T \epsilon_\sigma^2 + I \sigma_u^2) You can test it by python Emalgorithm_run.py

``GoWild'' is the code for problem1.3

``GraphicalLass.py'' is based on the paper: Sparse inverse covariance estimation with the graphical lasso You can run it by python GraphicalLasso.py

``Kronecker_product_one.py'' contains the function used in problem 2.2.3. the input of function get_GL_condition(c) is c, it well return the value of max||(Omega \kr \Omega){eS} (Omega \kr \Omega)^{-1}{S,S}||_1 You can run it by python Kronecker_product_one.py

``Kronecker_product_two.py'' contains the function used in problem 2.3.2. the input of function get_GL_condition(c) is c, it well return the value of max||Tau_{e,S} Tau_{S,S}^{-1}||_1 You can run it by python Kronecker_product_two.py