1: \begin{abstract}

2: This paper is concerned with the selection and estimation of fixed and

3: random effects in linear mixed effects models. We propose a class of

4: nonconcave penalized profile likelihood methods for selecting and

5: estimating important fixed effects. To overcome the difficulty of

6: unknown covariance matrix of random effects, we propose to use a proxy

7: matrix in the penalized profile likelihood. We establish conditions on

8: the choice of the proxy matrix and show that the proposed procedure

9: enjoys the model selection consistency where the number of fixed

10: effects is allowed to grow exponentially with the sample size. We

11: further propose a group variable selection strategy to simultaneously

12: select and estimate important random effects, where the unknown

13: covariance matrix of random effects is replaced with a proxy matrix. We

14: prove that, with the proxy matrix appropriately chosen, the proposed

15: procedure can identify all true random effects with asymptotic

16: probability one, where the dimension of random effects vector is

17: allowed to increase exponentially with the sample size. Monte Carlo

18: simulation studies are conducted to examine the finite-sample

19: performance of the proposed procedures. We further illustrate the

20: proposed procedures via a real data example.

21: \end{abstract}