1: \begin{abstract}

2: 	We propose a global convergent numerical method to reconstruct the initial condition of a nonlinear parabolic equation from the measurement of both Dirichlet and Neumann data on the boundary of a bounded domain. 

3: The first step in our method is to derive, from the nonlinear governing parabolic equation, a nonlinear systems of elliptic partial differential equations (PDEs) whose solution yields directly the solution of the inverse source problem. We then establish a contraction mapping-like iterative scheme to solve this system.  The convergence of this iterative scheme is rigorously proved by employing a Carleman estimate and the argument in the proof of the traditional contraction mapping principle.

4: This convergence is fast in both theoretical and numerical senses.  

5: Moreover, our method, unlike the methods based on optimization, does not require a good initial guess of the true solution.

6: Numerical examples are presented to verify these results.

7: \end{abstract}

8: