The splay tree data structure in Go.
Splay trees are self-balancing binary search trees. Accessing a node in the tree causes it and its neighbors to be splayed upwards by means of rotations. Future accesses to this node or to its neighbors will then be cheaper. Splay trees perform best when there is temporal or spatial locality in the access patterns (insertions, lookups or removals). Say, you enter person data by date+time of birth and then you investigate specific age groups. Or you manage virtual memory pages and your applications tend to focus on groups of related pages. Over time they will abandon some groups of memory pages and move on to other groups of memory pages. The implied locality of reference in these examples indicate splay trees as the data structure of choice.
The worst case cost for a single access (insertions, lookups or removals) is O(n), but amortized (averaged) over a sequence of accesses, while starting from an empty tree, the cost will be O(log n). Inserting a sorted sequence of 'n' elements into an empty tree has total cost O(n), which is optimal. Retrieving that same sequence of 'n' elements in either ascending or descending order has total cost O(n), which is again optimal.
##References
"Performance Analysis of BSTs in System Software" by Ben Pfaff, compares the performance of 20 variants of binary search trees (BSTs) and concludes that splay trees perform best when accesses are sequential or clustered.
##Installation
$ go get github.com/gijsbers/go-splaytree
##Copyright and License
Copyright ©2016 Bert Gijsbers except where otherwise noted. All rights reserved. Use of this source code is governed by an Apache 2.0 license that can be found in the LICENSE file.