search

PanPalitta / phase_estimation

This project apply reinforcement learning algorithms based on DE and PSO to optimize adaptive quantum-phase estimation.
http://panpalitta.github.io/phase_estimation/
GNU General Public License v3.0
9 stars 5 forks source link

Clean up accept-reject criteria #12

Closed PanPalitta closed 8 years ago

PanPalitta commented 9 years ago

The accept-reject criteria should be dealt with, in terms of how data are transferred between classes at least, as one of the objective functions in a multiple objective setting. Accept-reject criteria are patched in at the moment.

peterwittek commented 8 years ago

How are we doing with this?

PanPalitta commented 8 years ago

The criteria are now moved to Phase class (as they are specific to the problem), and OptAlg class access these criteria through T_condition function and error_condition function from within check_success function through Problem class. The only part that I haven't done is generalizing the input to these function. I don't have a good idea for this at the moment.

peterwittek commented 8 years ago

I agree this is a bit messy at the moment. If I understand correctly, memory_fitarray is relevant only to Phase, so perhaps we could delegate all related logic to that class. The main loop variable numvar could be a variable in OptAlg, and the while loop could terminate with a condition like while (opt->done()) which would in turn do a similar evaluation. This way you could also move the heuristic bits between lines 114 and 140, and then between 144 and 167 in main.cpp to OptAlg.

PanPalitta commented 8 years ago

I have tried to further generalize the criteria. The error_condition I think is generalized enough to be used with other problems. However, I don't know how to generalize T_condition at the moment.

As for your suggestion, I would suggest that we defer the change for now as memory_fitarray stores results from many numvar whereas Phase and OptAlg are instantiated and deleted for every numvar. So as it stands, I think memory_fitarray would belong to a bigger structure (the main function) at the moment.

peterwittek commented 8 years ago

Okay, that is fair enough. I reckon this is clean enough. As far as I am concerned, we can close this issue.