A Convergent Inexact Abedin-Kitagawa Iteration Method for Monge-Amp\`ere Eigenvalue Problems

In this paper, we propose an inexact Aleksandrov solution based Abedin-Kitagawa iteration (AKI) method for solving (real) Monge-Amp{\`e}re eigenvalue problems. The proposed approach utilizes the convergent Rayleigh inverse iterative formulation introduced by Abedin and Kitagawa as the prototype. More importantly, it employs an error tolerance criterion of inexact Aleksandrov solutions to approximately solve the subproblems without spoiling the convergence, which becomes the most crucial issue for the efficient implementation of the iterative method. For the two-dimensional case, by properly taking advantage of the flexibility rendered by the proposed inexact approach and a convergent fixed-point-based approach to solve the subproblems, considerable advancements in computational efficiency can be achieved by the inexact AKI method with its convergence under the ${\cal C}^{2,\alpha}$ boundary condition being rigorously established. Numerical experiments are conducted to demonstrate the efficiency of the proposed inexact AKI method. The numerical results suggest that the inexact AKI method can be more than eight times faster than the original AKI method, at least for all the tested problems.