I can try and modify this myself, but wanted to check w/ @HaroldRKingsberg
For one of the graphs, I'd like to compare the computation cost for achieving a certain accuracy +/- epsilon, for the MLMC and MC methods.
If I'm understanding things correctly, currently the LayeredMCOptionSolver supports this sort of functionality (i.e. it runs until a desired accuracy condition is met). However, it appears that the NaiveMCOptionSolver runs for a set number of steps. Would it be possible to tweak the NaiveMCOptionSolver so that it runs until a convergence requirement is met, and then the number of steps required is returned?
I can try and modify this myself, but wanted to check w/ @HaroldRKingsberg
For one of the graphs, I'd like to compare the computation cost for achieving a certain accuracy +/- epsilon, for the MLMC and MC methods.
If I'm understanding things correctly, currently the LayeredMCOptionSolver supports this sort of functionality (i.e. it runs until a desired accuracy condition is met). However, it appears that the NaiveMCOptionSolver runs for a set number of steps. Would it be possible to tweak the NaiveMCOptionSolver so that it runs until a convergence requirement is met, and then the number of steps required is returned?