hash

 Questions and Answers

Q: In the implementation of chained hash tables presented in this chapter, the actual hash code used for accessing the table is the hash code modulo the table size. Why is this?

A: This transformation ensures that the hash coding does not position us past the end of the table. Although the hash function should ensure this itself, it is worthwhile for the hash table implementation to provide the guarantee as well, especially since the hash function is provided by the caller. However, this is not the same reason that the modulo is performed when double hashing a key in an open-addressed hash table. In this case, the process of double hashing may produce a hash coding that falls outside of the bounds of the table, even for two auxiliary hash functions each producing hash codings within the table. This is because the two hash codings are added together.

Q: Why are hash tables good for random access but not sequential access? For example, in a database system in which records are to be accessed in a sequential fashion, what is the problem with hashing?

A: Hash tables are excellent for random access because each key hashes us precisely to where we need to be in the table to access the data, or at least within a few steps when a collision occurs. However, hash tables do not support sequential access. After hashing to some position, we have no way to determine where the next smallest or largest key resides. Compare this with a linked list containing elements that are sorted. Assuming some initial position in the list, the next key is easy to determine: we simply look at the next element in the list.

Q: What is the worst-case performance of searching for an element in a chained hash table? How do we ensure that this case will not occur?

A: A chained hash table performs the worst when all elements hash into a single bucket. In this case, searching for an element is O (n), where n is the number of elements in the table. A ridiculous hash function that would result in this performance is h (k) = c, where c is some constant within the bounds of the hash table. Selecting a good hash function ensures that this case will not occur. If the hash function approximates uniform hashing well, we can expect to locate an element in constant time.

Q: What is the worst-case performance of searching for an element in an open-addressed hash table? How do we ensure that this case will not occur?

A: The worst-case performance of searching for an element in an open-addressed hash table occurs once the hash table is completely full and the element we are searching for is not in the table. In this case, searching for an element is an O (m) operation, where m is the number of positions in the table. This case can occur with any hash function. To ensure reasonable performance in an open-addressed hash table, we should not let the table become more than 80% full. If we choose a hash function that approximates uniform hashing well, we can expect performance consistent with what is presented in Table 8.1.