程序代写代做代考 data structure Today:

Today:
ECS 20 — Lecture 8 — Fall 2013 —22 Oct 2013
Phil Rogaway
o Quiz 2
o Some more operations on sets
o How a computer might manipulate sets: dictionaries and disjoint-sets
(INSERT/ IN/DELETE;
UNION/FIND/MAKESET)
Various laws
Prove them by tracing through the definitions
De Morgan’s laws:
 
Proof (of first claim): x  (A  B) c
iff iff iff iff
 x  (A  B)) xA x B)
xA)  xB) xAc  xBc
Be careful!!
Cartesian Product (= Cross product)
A  B = {(a,b): AA, B B} R2 points in the plane
An array of chessmen might be represented by BYTES64
? (A\B)\CA\(B\C)
1

Unordered Product
A & B = {{a,b}: AA, BB} // when I learned graph theory — never saw it since! Power Set
P – Power set operator, unary operator (takes 1 input). P(x) is the “set of all subsets of x”
P (X) = {A: A  X} Example: X = {a, b, c}
Example:
Variant notation: P (X) = 2X
Notation is suggestive of size – For X finite, |P (X) |= 2|X|
Dictionary ADT
and its realization with a list and with a hash table
Want to be able to Insert items into a dictionary and to Lookup if an item is already in the dictionary. (Sometimes want to be able to Delete an item, too.) For concreteness, think of the items we are inserting as strings.
Example: discover how many distinct words are in a book.
Implementation
1) A list of words, each one appearing at most once. 2) A hash table.
Explain how each works.
Show how to modify the hash table to do a frequency count.
Representing a collection of sets in a computers
A different game – we are going to maintain a collection of disjoint sets. We want to be able to figure out if two things are in the same set, or in different sets. For example, each point in the set might represent a person and when we learn that person one and person two know one another – maybe one calls or emails the other – then we combine them. Each set then represents people that know one another through some path of knowing.
2

More interesting applications will come later, when we do graph theory. You want to realize
 find(x) return a canonical name for the unique set containing x. x and y are in the same set iff find(x)=find(y)
 union(x,y) merge the sets containing x and y.
 makeset (x) create a set containing the element x. Return a canonical name for it Naïve implementation: list of elements
Smarter – “union/find data structure”
Union by rank
Collapsing find.
Any sequence of n operations takes n (n) time, for an incredibly slows growing function(n). [Omit big-O because not yet introduced]
Tarjan (1975)
function MakeSet(x) x.parent := x x.rank := 0
function Union(x, y) xRoot := Find(x) yRoot := Find(y) if xRoot == yRoot
return
// x and y are not already in same set. Merge them.
if xRoot.rank < yRoot.rank xRoot.parent := yRoot else if xRoot.rank > yRoot.rank
yRoot.parent := xRoot
else
yRoot.parent := xRoot
3

xRoot.rank := xRoot.rank + 1
The second improvement, called path compression, is a way of flattening the structure of the tree whenever Find is used on it. The idea is that each
function Find(x)
if x.parent != x
x.parent := Find(x.parent) return x.parent

4