Suppor of the vector x is is the number of non-zero elements in that vector.

The LASSO estimator is defined as follows B(x):= argmin|1Y - Xb][3+1|16|| Exercise 2 Let X E RMX be an orthogonal matrix and to be the (1 + V1 -a)/2 quantile of a N (0,02) distribution 1) Prove the following inequality P(=i supp(3) such that B,(40) #9) sa. This probability called the familywise error rate) is the probability to do not correctly estimate at 0 with LASSO estimator at least one null component of 8. In which case this inequality is an equality? The LASSO estimator is defined as follows B(x):= argmin|1Y - Xb][3+1|16|| Exercise 2 Let X E RMX be an orthogonal matrix and to be the (1 + V1 -a)/2 quantile of a N (0,02) distribution 1) Prove the following inequality P(=i supp(3) such that B,(40) #9) sa. This probability called the familywise error rate) is the probability to do not correctly estimate at 0 with LASSO estimator at least one null component of 8. In which case this inequality is an equality