Tesi etd-09252013-124512 |
Link copiato negli appunti
Tipo di tesi
Tesi di laurea magistrale
Autore
DEL TESSANDORO, EMILIO
URN
etd-09252013-124512
Titolo
Binary Decision Diagrams for sequences: definitions, properties and algorithms
Dipartimento
INFORMATICA
Corso di studi
INFORMATICA
Relatori
relatore Grossi, Roberto
relatore Pagli, Linda
controrelatore Prof.ssa Scutellà, Maria Grazia
relatore Pagli, Linda
controrelatore Prof.ssa Scutellà, Maria Grazia
Parole chiave
- algoritmi
- bdd
- binary decision diagram
- stringhe
Data inizio appello
11/10/2013
Consultabilità
Completa
Riassunto
Binary Decision Diagrams (BDDs) are a particular kind of graph for representing boolean functions. In particular they are rooted directed acyclic graphs, where every node represent a binary decision, or equivalently a branch on a certain boolean variable.
There is a wide range of problems where BDDs are suitable because of what is called symbolic analysis. It’s possible in fact to encode the parameters of a system with boolean variables, and in general to encode a problem with a boolean function.
Many flavours of decision diagrams exist in the literature. One of them, the Sequence BDD, has been used recently for representing sets of strings, with interesting results.
The study of these kind of BDDs led to the discovery of new algorithms, mainly conceived for their reduction, that have been studied both theoretically and experimentally. In particular for this last purpose a small BDD package was developed. As a practical application, the problem of indexing substrings has been studied more in depth.
There is a wide range of problems where BDDs are suitable because of what is called symbolic analysis. It’s possible in fact to encode the parameters of a system with boolean variables, and in general to encode a problem with a boolean function.
Many flavours of decision diagrams exist in the literature. One of them, the Sequence BDD, has been used recently for representing sets of strings, with interesting results.
The study of these kind of BDDs led to the discovery of new algorithms, mainly conceived for their reduction, that have been studied both theoretically and experimentally. In particular for this last purpose a small BDD package was developed. As a practical application, the problem of indexing substrings has been studied more in depth.
File
Nome file | Dimensione |
---|---|
thesis_final.pdf | 492.69 Kb |
Contatta l’autore |