TY - GEN
T1 - Comparison of Different Open Addressing Hashing Algorithms
AU - Luo, Wenbin
AU - Heileman, Gregory L.
N1 - Publisher Copyright:
© 18th International Conference on Computers and Their Applications 2003, CATA 2003. All rights reserved.
PY - 2003
Y1 - 2003
N2 - Hash functions are among the oldest and most widely used data structures in computer science. Different hash functions exist. So, it is very important to compare their performance. In this paper, we introduced our new hash function which was proposed recently in [1], and compared its performance with two different open addressing hashing algorithms: double hashing and exponential hashing. Double hashing is a widely-used hash function, which offers good performance. Exponential hashing, proposed by Smith et al [2], has been shown through experimental analysis to be superior to double hashing when the data elements are not uniformly distributed. The good performance of exponential hashing is due to its ability of spreading the table elements more randomly than double hashing. While exponential hashing has some desirable characteristics, Smith points out that it produces less than full probe length on 1/2 of the table elements. This is undesirable since it could potentially lead to insertion and search failures. Our new hash function has the ability of spreading the table elements randomly, and produces full probe length on all the table elements at the same time. From this point of view, our new hash function combines the strength of both double hashing and exponential hashing with their weakness eliminated. Experimental results are presented to support our claims, which is followed by some theoretic analysis.
AB - Hash functions are among the oldest and most widely used data structures in computer science. Different hash functions exist. So, it is very important to compare their performance. In this paper, we introduced our new hash function which was proposed recently in [1], and compared its performance with two different open addressing hashing algorithms: double hashing and exponential hashing. Double hashing is a widely-used hash function, which offers good performance. Exponential hashing, proposed by Smith et al [2], has been shown through experimental analysis to be superior to double hashing when the data elements are not uniformly distributed. The good performance of exponential hashing is due to its ability of spreading the table elements more randomly than double hashing. While exponential hashing has some desirable characteristics, Smith points out that it produces less than full probe length on 1/2 of the table elements. This is undesirable since it could potentially lead to insertion and search failures. Our new hash function has the ability of spreading the table elements randomly, and produces full probe length on all the table elements at the same time. From this point of view, our new hash function combines the strength of both double hashing and exponential hashing with their weakness eliminated. Experimental results are presented to support our claims, which is followed by some theoretic analysis.
UR - http://www.scopus.com/inward/record.url?scp=5444248986&partnerID=8YFLogxK
UR - http://www.scopus.com/inward/citedby.url?scp=5444248986&partnerID=8YFLogxK
M3 - Conference contribution
AN - SCOPUS:5444248986
T3 - 18th International Conference on Computers and Their Applications 2003, CATA 2003
SP - 1
EP - 4
BT - 18th International Conference on Computers and Their Applications 2003, CATA 2003
A2 - Debnath, Narayan C.
PB - The International Society for Computers and Their Applications (ISCA)
T2 - 18th International Conference on Computers and Their Applications, CATA 2003
Y2 - 26 March 2003 through 28 March 2003
ER -