1: \begin{abstract}

2: We consider a sparse linear regression model with unknown symmetric error under the high-dimensional setting.

3: The true error distribution is assumed to belong to the locally $\beta$-H\"{o}lder class with an exponentially decreasing tail, which does not need to be sub-Gaussian. 

4: We obtain posterior convergence rates of the regression coefficient and the error density, which are nearly optimal and adaptive to the unknown sparsity level. 

5: Furthermore, we derive the semi-parametric Bernstein-von Mises (BvM) theorem to characterize asymptotic shape of the marginal posterior for regression coefficients.

6: Under the sub-Gaussianity assumption on the true score function, strong model selection consistency for regression coefficients are also obtained, which eventually asserts the frequentist's validity of credible sets.

7: %\bch To the best of our knowledge, our work is the first that has obtained posterior convergence rates, the BvM theorem and strong model selection consistency in high-dimensional regression models without the sub-Gaussian assumption. \ech

8: %\keywords{High-dimensional semi-parametric model \and posterior convergence rate \and Bernstein-von Mises theorem \and strong model selection consistency}

9: % \PACS{PACS code1 \and PACS code2 \and more}

10: % \subclass{MSC code1 \and MSC code2 \and more}

11: \end{abstract}

12: