Can you bring me to the research showing b-tree is equally performant? Is postgres taking this research into account?
Not trying to be rude, but you've been asking rather general questions; our mailing list is archived, searchable, and probably a better use of everyone's time for you to consult prior to posting. Per your question, to my knowledge, there is no active work on changing our primary b-tree index structure, which is based on Lehman and Yao's b-link tree. Given the maturity of our current implementation, I think it would be rather difficult to improve upon it in terms of performance, especially considering concurrency-related issues.