What Is a Splay Tree?

A splay tree is an optimized binary search tree, or node-based data tree, which is self-adjusting and better for accessing recently searched elements and nodes. Five functions can be performed on the splay tree, allowing the user to manipulate the nodes. This tree has a very small footprint, so little memory is needed to store the data. A disadvantage to this tree is that it is built linearly, so nodes stored toward the bottom will take longer to access.

Splay trees are binary trees that store nodes of data; this is usually binary data, though they also can hold files. Unlike a regular binary search tree, which allows users to search through the nodes, a splay tree moves itself around to make searching much faster. Any recently opened nodes will be arranged near the top of the tree, so less time is needed to find and open the node. This rearranging means splay trees are useful for caches — computer memory that holds recently accessed data — and for algorithms made to eliminate unused data.

Five functions can be used on the tree. The splay operation is like a join operation, because one node’s access becomes connected with another node. The split function takes one node and splits it into two or more nodes. With join, two nodes are turned into one. Insert takes part of a node and inserts it into another, while the delete function erases a section of the node from the splay tree.

A splay tree uses very little memory, which allows users to make large trees without taking up a massive amount of hard drive space. Splay trees are simple, and do not require much code to build, so they do not require the same amount of memory that more complex trees require. Bookkeeping information, which is usually necessary for other trees to keep track of data positioning, is unneeded because of the tree’s self-rearranging nature.

While the splay tree takes up little memory and can easily access recent nodes, speed can be an issue. Nodes can only be arranged in a linear fashion, meaning some nodes will be low on the tree, while recent nodes are at the top. These bottom nodes will be difficult to reach, because the tree has to search from top to bottom until the bottom nodes are found. This is because there is no bookkeeping data, resulting in slow searches for low nodes.