General Splay: A Basic Theory and Calculus

  • Authors:

  • G. F. Georgakopoulos;D. J. McClurkin

  • Affiliations:

  • -;-

  • Venue:
  • ISAAC '99 Proceedings of the 10th International Symposium on Algorithms and Computation
  • Year:
  • 1999

Quantified Score

Hi-index 0.00

Visualization

Abstract

For storage and retrieval applications where access frequencies are biased, uniformly-balanced search trees may be suboptimal. Splay trees address this issue, providing a means for searching which is statically optimum and conjectured to be dynamically optimum. Subramanian explored the reasons for their success, expressing local transformations as templates and giving sufficient criteria for a template family to exhibit amortized O(log N) performance. We present a different formulation of the potential function, based on progress factors along edges. Its decomposition w.r.t. a template enables us to relax all of Subramanian's conditions. Moreover it illustrates the reasons why template-based self-adjustment schemes work, and provides a straightforward way of evaluating the efficiency of such schemes.