QuaRL source code

[JMuff22]

Hi this is not an issue more of a question/discussion. I have read the Quartz paper and the QuaRL paper. I found the quarl repository https://github.com/quantum-compiler/Quarl but noticed it is several commits behind this fork.

• If I want to use QuaRL which repository branch should I use?
• Are the techniques presented in the QuaRL paper present in this repository? If so where?

Thanks!

[xumingkuan]

Thanks for your interest in the works of our research group!

We implemented Quarl as a fork of Quartz at the beginning, and then, when I decided to add more features to Quartz, I did not maintain the codebase related to Quarl. So although the Quartz repo contains more commits, it is not guaranteed that Quarl still works here. So I believe that you should use the Quarl repo.

I'm not sure which branch you should use or where the techniques are. @zikun-li should be able to answer these questions.

[JMuff22]

I am not sure there is the Quarl artifact on github somewhere but it is on Zenodo: https://zenodo.org/records/10463907

[JMuff22]

@xumingkuan just so I know which is a better starting point, would you recommend to use Quartz now over Quarl as Quartz has more features? Have all of the methods in Quarl been incorporated into the Quartz code repository?

[xumingkuan]

If you want to develop new methods, I recommend using Quartz because Quartz has more features and the codebase is better maintained.

If you only want to optimize circuits, however, Quartz has not covered all optimizations Quarl's reinforcement learning finds. If the budget of training an RL model is not a concern, Quarl probably still outperforms Quartz.