1: \begin{abstract}

2:   Maximum entropy inference and learning of graphical models are pivotal tasks

3:   in learning theory and optimization.

4:   This work extends algorithms for these problems, including generalized

5:   iterative scaling (GIS) and gradient descent (GD), to the quantum realm.

6:   While the generalization, known as quantum iterative scaling (QIS), is

7:   straightforward, the key challenge lies in the non-commutative nature of

8:   quantum problem instances, rendering the convergence rate analysis

9:   significantly more challenging than the classical case.

10:   Our principal technical contribution centers on a rigorous analysis of the

11:   convergence rates, involving the establishment of both lower and upper bounds

12:   on the spectral radius of the Jacobian matrix for each iteration of these

13:   algorithms.

14:   Furthermore, we explore quasi-Newton methods to enhance the performance of QIS

15:   and GD\@.

16:   Specifically, we propose using Anderson mixing and the L-BFGS method for QIS

17:   and GD, respectively.

18:   These quasi-Newton techniques exhibit remarkable efficiency gains, resulting

19:   in orders of magnitude improvements in performance.

20:   As an application, our algorithms provide a viable approach to designing

21:   Hamiltonian learning algorithms.

22: \end{abstract}

23: