1: \begin{abstract}

2: In this paper, we introduce a new nonlinear evolution partial

3: differential equation for sparse deconvolution problems. The

4: proposed PDE has the form of continuity equation that arises in

5: various research areas, e.g. fluid dynamics and optimal

6: transportation, and thus has some interesting physical and

7: geometric interpretations. The underlying optimization model that

8: we consider is the standard $\ell_1$ minimization with linear

9: equality constraints, i.e. $\min_u\{\|u\|_1 : Au=f\}$ with $A$

10: being an under-sampled convolution operator. We show that our PDE

11: preserves the $\ell_1$ norm while lowering the residual

12: $\|Au-f\|_2$. More importantly the solution of the PDE becomes

13: sparser asymptotically, which is illustrated numerically.

14: Therefore, it can be treated as a natural and helpful plug-in to

15: some algorithms for $\ell_1$ minimization problems, e.g. Bregman

16: iterative methods introduced for sparse reconstruction problems in

17: \cite{Yin:2008p611}. Numerical experiments show great improvements

18: in terms of both convergence speed and reconstruction quality.

19: \end{abstract}

20: