数据结构翻译.doc

(a,b)树和B树 A different method to keep the height of the trees small is to allow tree nodes of higher degree. This idea was introduced as B-trees by Bayer and McCreight (1972) and turned out to be very fruitful. It was originally intended as external memory data structure, but we will see in Section 3.4 that it has interesting uses also as normal main memory data structure. The characteristic of external memory is that access to it is very slow, compared to main memory, and is done in blocks, units much larger than single main memory locations, which are simultaneously transferred into main memory. In the 1970s, computers were still very memory limited but usually already had a large external memory, so that it was a necessary consideration how a structure operates when a large part of it is not in main memory, but on external memory. This situation is now less important, but it is still relevant for database applications, where B-tree variants are still much used as index structures。 不同的方法保持树的高度是允许树节点。它原本是外部存数据结构,但我们在3.4节中看到的,外部存储器的特点是与主存储器相比访问非常慢同时转移到主内存在1970年代,计算机内存仍非常有限,但通常已经有了一个很大的外部存储器所以这是一个必要的考虑结构大的一部分,不是在主内存,是在外部存储器如何运作。这种情况现在不那么重要,但它仍然-树变仍多用作索引结构普通二进制搜索树作为外部内存结构每个树节点可以在不同外部内存块,只有前面的块已经从外部存储器中检索它成为已知所以我们需要尽可能多的外部内存块访问树的高度,树的想法是把每一块作为一个节点的高度。在最初的版本中每个节点和2?1之间然后平衡的标准,所有的树叶都应该在同一深度(logn) block changes can be necessary.For a main memory data structure, the number of changes in rebalancing makes little difference, although it has been studied in many papers; but for an external memory structure it is essential because all changed blocks have to be written again to the external memory device. So these trees, known as (a,b)-trees, are the method of choice. 从Bayer and McCreight 提出的平衡算法中，度的间隔在a到2a-1之间是最小的间隔。因为每一块的空间最多是2a-1个元素，并且至少有一半是符合这种方式的。这听起来像一个不错的选择来优化空间利用率。≥2a-1然后平衡改变复杂度为O（1）的块。而对于b = 2?1,最初的选择(logn)的块的是必要的。主内存数据结构,平衡的数量变化不大,虽然它已经在许多论文研究;但对于外部内存结构至关重要,因为所有改变块外部存储设备。所以这些树,称为(a,b)树,是选择的方法。 An (a, b)-tree is a nonbinary search tree in which all leaves have the same