1: \begin{abstract}

2: The Variational Monte Carlo method has recently seen important advances through the use of neural network quantum states.

3: While more and more sophisticated ansätze have been designed to tackle a wide variety of quantum many-body problems,

4: modest progress has been made on the associated optimisation algorithms.

5: In this work, we revisit the Kronecker Factored Approximate Curvature, an optimiser

6: that has been used extensively in a variety of simulations.

7: We suggest improvements on the scaling and the direction of this optimiser, and find that they

8: substantially increase its performance at a negligible additional cost.

9: We also reformulate the Variational Monte Carlo approach

10: in a game theory framework, to propose a novel optimiser based on decision geometry.

11: We find that, on a practical test case for continuous systems, this new optimiser consistently outperforms any of the

12: KFAC improvements in terms of stability, accuracy and speed of convergence.

13: Beyond Variational Monte Carlo, the versatility of this approach suggests

14: that decision geometry could provide a solid foundation for accelerating 

15: a broad class of machine learning algorithms.

16: \end{abstract}

17: