1: \begin{abstract}

2:     Computing tasks may often be posed as optimization problems.

3:     The objective functions for real-world scenarios are often nonconvex and/or nondifferentiatiable. State-of-the-art methods for solving these problems typically only guarantee convergence to local minima. 

4:     This work presents Hamilton-Jacobi-based Moreau Adaptive Descent (HJ-MAD), a zero-order algorithm with guaranteed convergence to global minima, assuming continuity of the objective function.

5:     The core idea is to compute gradients of the Moreau envelope of the objective (which is ``piece-wise convex'') with adaptive smoothing parameters. 

6:     Gradients of the Moreau envelope are approximated by employing the Hopf-Lax formula for the viscous Hamilton-Jacobi equation.

7:     Provided numerical examples illustrate global convergence.

8: \end{abstract}

9: