1: \begin{abstract}

2: This paper presents a unified geometric framework for the statistical analysis of a general ill-posed linear inverse model which includes  as special cases  noisy compressed sensing, sign vector recovery, trace regression, orthogonal matrix estimation, and noisy matrix completion. We propose computationally feasible convex programs for statistical inference including estimation, confidence intervals and hypothesis testing. A theoretical framework is developed to characterize the local estimation rate of convergence and to provide statistical inference guarantees. Our results are built based on the local conic geometry and duality. 

3: %We also study minimax lower bound for the linear inverse problem restricted to a general convex cone. 

4: The difficulty of statistical inference is captured by the geometric characterization of the local tangent cone through 

5: %the complexity measures -- 

6: the Gaussian width and Sudakov minoration estimate. 

7: 

8: 

9: 

10: \end{abstract}

11: