1: \begin{abstract} 

2: The batch exponentiated gradient (EG) method provides a principled approach to convex smooth minimization on the probability simplex or the space of quantum density matrices.

3: However, it is not always guaranteed to converge.

4: Existing convergence analyses of the EG method require certain quantitative smoothness conditions on the loss function, e.g., Lipschitz continuity of the loss function or its gradient, but those conditions may not hold in important applications.

5: In this paper, we prove that the EG method with Armijo line search always converges for \emph{any} convex loss function with a locally Lipschitz continuous gradient. 

6: Because of our convergence guarantee, the EG method with Armijo line search becomes the fastest guaranteed-to-converge algorithm for maximum-likelihood quantum state estimation, on the real datasets we have.

7: \end{abstract}