Concluding Remarks and Open Questions
Much progress has been made on persistent data structures since the seminal paper of Driscoll et al. . This progress has three folds: In developing general techniques to make any data structure persistent, in making speciﬁc data structures persistent, and in emerging algorithmic applications. Techniques developed to address these challenges sometimes proved useful for other applications as well.
This algorithmic ﬁeld still comprise intriguing challenges. In developing general techniques to make data structures persistent, a notable challenge is to ﬁnd a way to make the time slowdown of the node splitting method worst case. Another interesting research track is how to restrict the operations that combine versions in a conﬂuently persistent set- ting so that better time bounds, or simpler simulations, are possible. We also believe that the techniques and data structures developed in this ﬁeld would prove useful for numerous forthcoming applications.