0 Up votes0 Down votes

43 views5 pagesalgos

Nov 16, 2015

© © All Rights Reserved

DOCX, PDF, TXT or read online from Scribd

algos

© All Rights Reserved

43 views

algos

© All Rights Reserved

- Steve Jobs
- Wheel of Time
- NIV, Holy Bible, eBook
- NIV, Holy Bible, eBook, Red Letter Edition
- Cryptonomicon
- The Woman Who Smashed Codes: A True Story of Love, Spies, and the Unlikely Heroine who Outwitted America's Enemies
- Contagious: Why Things Catch On
- Crossing the Chasm: Marketing and Selling Technology Project
- Everybody Lies: Big Data, New Data, and What the Internet Can Tell Us About Who We Really Are
- Zero to One: Notes on Start-ups, or How to Build the Future
- Console Wars: Sega, Nintendo, and the Battle that Defined a Generation
- Dust: Scarpetta (Book 21)
- Hit Refresh: The Quest to Rediscover Microsoft's Soul and Imagine a Better Future for Everyone
- The Innovators: How a Group of Hackers, Geniuses, and Geeks Created the Digital Revolution
- Crushing It!: How Great Entrepreneurs Build Their Business and Influence—and How You Can, Too
- Make Time: How to Focus on What Matters Every Day
- Algorithms to Live By: The Computer Science of Human Decisions
- Wild Cards

You are on page 1of 5

1 : Using the linked-list representation of disjoint sets and the weightedunion heuristic, a sequence of m MAKE-SET, UNION, and FIND-SET operations, n of

which are MAKE-SET operations, takes O(m + n lg n) time.

Proof We start by computing, for each object in a set of size n, an upper bound on the

number of times the object's pointer back to the representative has been updated.

Consider a fixed object x. We know that each time x's representative pointer was

updated, x must have started in the smaller set. The first time x's representative

pointer was updated, therefore, the resulting set must have had at least 2 members.

Similarly, the next time x's representative pointer was updated, the resulting set must

have had at least 4 members. Continuing on, we observe that for any k n, after x's

representative pointer has been updated lg k times, the resulting set must have at

least k members. Since the largest set has at most n members, each object's

representative pointer has been updated at most lg n times over all the UNION

operations. We must also account for updating the head and tail pointers and the list

lengths, which take only (1) time per UNION operation. The total time used in

updating the n objects is thus O(n lg n).

The time for the entire sequence of m operations follows easily. Each MAKE-SET and

FIND-SET operation takes O(1) time, and there are O(m) of them. The total time for

the entire sequence is thus O(m + n lg n).

=====Adapt the aggregate proof of Theorem 21.1 to obtain amortized time bounds of

O(1) for MAKE-SET and FIND-SET and O(lg n) for UNION using the linked-list

representation and the weighted-union heuristic.

Proof: We want to show that we can assign O(1) charges to MAKE-SET and FINDSET and an O(lg n) charge to UNION such that the charges for a sequence of these

operations are enough to cover the cost of the sequence.O(m+n lg n), according to

the theorem. When talking about the charge for each kind of operation, it is helpful to

also be able to talk about the number of each kind of operation. Consider the usual

sequence of m MAKE-SET, UNION, and FIND-SET operations, n of which are MAKESET operations, and let l < n be the number of UNION operations. (Recall the

discussion in Section 21.1 about there being at most n 1 UNION operations.) Then

there are n MAKE-SET operations, l UNION operations, and m n l FIND-SET

operations. The theorem didn.t separately name the number l of UNIONs; rather, it

bounded the number by n. If you go through the proof of the theorem with l UNIONs,

you get the time bound O(ml+l lg l) = O(m+l lg l) for the sequence of operations. That

is, the actual time taken by the sequence of operations is at most c(m +l lg l), for some

constant c. Thus, we want to assign operation charges such that (MAKE-SET charge)

n

+(FIND-SET charge) (m n l)

+(UNION charge) l

Sum c(m + l lg l) ,

so that the amortized costs give an upper bound on the actual costs. The following

assignments work, where c is some constant c: MAKE-SET: c; FIND-SET: c;

UNION: c(lg n + 1)

Substituting into the above sum, we get

cn + c(m n l) + c(lg n + 1)l = cm + cl lg n = c(m + l lg n)

> c(m + l lg l) .

===== Suggest a simple change to the UNION procedure for the linked-list

representation that removes the need to keep the tail pointer to the last object in each

list. Whether or not the weighted-union heuristic is used, your change should not

change the asymptotic running time of the UNION procedure. (Hint: Rather than

appending one list to another, splice them together.)

Proof: Let.s call the two lists A and B, and suppose that the representative of the new

list will be the representative of A. Rather than appending B to the end of A, instead

splice B into A right after the rst element of A. We have to traverse B to update

representative pointers anyway, so we can just make the last element of B point to the

second element of A.

=====Give a sequence of m MAKE-SET, UNION, and FIND-SET operations, n of

which are MAKE-SET operations, that takes (m lg n) time when we use union by

rank only

Proof: You need to nd a sequence of m operations on n elements that takes _(m lg

n) time. Start with n MAKE-SETs to create singleton sets {x1} , {x2} , . . . , {xn}. Next

perform the n 1 UNION operations shown below to create a single set whose tree

has depth lg n.

UNION(x1, x2) n/2 of these

UNION(x4, x8) n/8 of these

UNION(x3, x4)

UNION(x12, x16)

UNION(x5, x6)

UNION(x20, x24)

...

...

UNION(xn1, xn)

UNION(xn4, xn)

UNION(x2, x4) n/4 of these

UNION(x6, x8)

...

UNION(x10, x12)

UNION(xn/2, xn) 1 of these

...

UNION(xn2, xn)

Finally, perform m 2n + 1 FIND-SET operations on the deepest element in the tree.

Each of these FIND-SET operations takes _(lg n) time. Letting m 3n, we have more

than m/3 FIND-SET operations, so that the total cost is _(m lg n).

===== Show that any sequence of m MAKE-SET, FIND-SET, and LINK operations,

where all the LINK operations appear before any of the FIND-SET operations, takes

only O(m) time if both path compression and union by rank are used. What happens

in the same situation if only the path-compression heuristic is used?

With the path-compression heuristic, the sequence of m MAKE-SET, FIND-SET, and

LINK operations, where all the LINK operations take place before any of the FIND-

SET operations, runs in O(m) time. The key observation is that once a node x appears

on a nd path, x will be either a root or a child of a root at all times thereafter.

We use the accounting method to obtain the O(m) time bound. We charge a MAKESET operation two dollars. One dollar pays for the MAKE-SET, and one dollar remains

on the node x that is created. The latter pays for the rst time that x appears on a nd

path and is turned into a child of a root.We charge one dollar for a LINK operation.

This dollar pays for the actual linking of one node to another.

We charge one dollar for a FIND-SET. This dollar pays for visiting the root and its

child, and for the path compression of these two nodes, during the FIND-SET. All

other nodes on the nd path use their stored dollar to pay for their visitation and path

compression. As mentioned, after the FIND-SET, all nodes on the nd path become

children of a root (except for the root itself), and so whenever they are visited during a

subsequent FIND-SET, the FIND-SET operation itself will pay for them.

operationsis charged at most 2m dollars, and so the total time is O(m). Observe that

nothing in the above argument requires union by rank. Therefore, weget an O(m) time

bound regardless of whether we use union by rank.

=====Using (every node has rank at most lg n.) give a simple proof that operations

on a disjoint-set forest with union by rank but without path compression run in O(m lg

n) time.

Proof: Clearly, each MAKE-SET and LINK operation takes O(1) time. Because the

rank of a node is an upper bound on its height, each find path has length O(lg n),

which in turn implies that each FIND-SET takes O(lg n) time. Thus, any sequence of

m MAKE-SET, LINK, and FIND-SET operations on n elements takes O(m lg n) time. It

is easy to prove an analogue of Lemma 21.7 to show that if we convert a sequence of

m_ MAKE-SET, UNION, and FIND-SET operations into a sequence of m MAKE-SET,

LINK, and FIND-SET operations that take O(m lg n) time, then the sequence of m_

MAKE-SET, UNION, and FIND-SET operations takes O(m_ lg n) time.

=====Professor Dante reasons that because node ranks increase strictly along a

path to the root, node levels must monotonically increase along the path. In other

words, if rank(x) > 0 and p[x] is not a root, then level(x) level(p[x]). Is the professor

correct?

Proof: Professor Dante is mistaken. Take the following scenario. Let n = 16, and

make16 separate singleton sets using MAKE-SET. Then do 8 UNION operations to

link the sets into 8 pairs, where each pair has a root with rank 0 and a child with rank

1. Now do 4 UNIONs to link pairs of these trees, so that there are 4 trees, each with a

root of rank 2, children of the root of ranks 1 and 0, and a node of rank 0 that is the

child of the rank-1 node. Now link pairs of these trees together, so that there are two

resulting trees, each with a root of rank 3 and each containing a path from a leaf to the

root with ranks 0, 1, and 3. Finally, link these two trees together, so that there is a path

from a leaf to the root with ranks 0, 1, 3, and 4. Let x and y be the nodes on this path

with ranks 1 and 3, respectively. Since A1(1) = 3, level(x) = 1, and since A0(3) = 4,

level(y) = 0. Yet y follows x on the nd path.

=====Consider the function '(n) = min {k : Ak (1) lg(n + 1)}. Show that '(n) 3 for

all practical values of n and, using (every node has rank at most lg n) show how to

modify the potential-function argument to prove that a sequence of m MAKE-SET,

UNION, and FIND-SET operations, n of which are MAKE-SET operations, can be

performed on a disjoint-set forest with union by rank and path compression in worstcase time O(m '(n)).

Proof: First, _(22047 1) = min {k : Ak (1) 2047} = 3, and 22047 1 _ 1080.

Second, we need that 0 level(x) _(n) for all nonroots x with rank[x] 1.

With this denition of _(n), we have A_(n)(rank[x]) A_(n)(1) lg(n + 1) >

lg n rank(p[x]). The rest of the proof goes through with _(n) replacing (n).

===== When an adjacency-matrix representation is used, most graph algorithms

require time (V2), but there are some exceptions. Show that determining whether a

directed graph G contains a universal sink-a vertex with in-degree |V| - 1 and outdegree 0-can be determined in time O(V), given an adjacency matrix for G.

Proof: We start by observing that if ai j = 1, so that (i, j ) E, then vertex i cannot be a

universal sink, for it has an outgoing edge. Thus, if row i contains a 1, then vertex i

cannot be a universal sink. This observation also means that if there is a self-loop (i,

i ), then vertex i is not a universal sink. Now suppose that ai j = 0, so that (i, j ) _ E,

and also that i _= j . Then vertex j cannot be a universal sink, for either its in-degree

must be strictly less than |V| 1 or it has a self-loop. Thus if column j contains a 0 in

any position other than the diagonal entry ( j, j ), then vertex j cannot be a universal

sink.

Using the above observations, the following procedure returns TRUE if vertex k is a

universal sink, and FALSE otherwise. It takes as input a |V| |V| adjacency matrix A =

(ai j ).

IS-SINK(A, k)

let A be |V| |V|

for j 1 to |V| _ Check for a 1 in row k

do if akj = 1

then return FALSE

for i 1 to |V| _ Check for an off-diagonal 0 in column k

do if aik = 0 and i _= k

then return FALSE

return TRUE

Because this procedure runs in O(V) time, we may call it only O(1) times in order to

achieve our O(V)-time bound for determining whether directed graph G contains a

universal sink. Observe also that a directed graph can have at most one universal

sink. This property holds because if vertex j is a universal sink, then we would have (i,

j ) E for all i _= j and so no other vertex i could be a universal sink.

The following procedure takes an adjacency matrix A as input and returns either a

message that there is no universal sink or a message containing the identity of the

universal sink. It works by eliminating all but one vertex as a potential universal

sink and then checking the remaining candidate vertex by a single call to IS-SINK.

UNIVERSAL-SINK(A)

let A be |V| |V|

ij1

while i |V| and j |V|

do if ai j = 1

then i i + 1

else j j + 1

s0

if i > |V| then return .there is no universal sink.

elseif IS-SINK(A, i ) = FALSE

then return .there is no universal sink.

else return i .is a universal sink.

UNIVERSAL-SINK walks through the adjacency matrix, starting at the upper left

corner and always moving either right or down by one position, depending on whether

the current entry ai j it is examining is 0 or 1. It stops once either i or j exceeds |V|.

To understand why UNIVERSAL-SINK works, we need to show that after the while

loop terminates, the only vertex that might be a universal sink is vertex i . The call to

IS-SINK then determines whether vertex i is indeed a universal sink.

Let us x i and j to be values of these variables at the termination of the while loop.

We claim that every vertex k such that 1 k < i cannot be a universal sink.

That is because the way that i achieved its nal value at loop termination was by

finding a 1 in each row k for which 1 k < i . As we observed above, any vertex k

whose row contains a 1 cannot be a universal sink.

If i > |V| at loop termination, then we have eliminated all vertices from consideration,

and so there is no universal sink. If, on the other hand, i |V| at loop termination, we

need to show that every vertex k such that i < k |V| cannot be a universal sink. If i |

V| at loop termination, then the while loop terminated because j > |V|. That means that

we found a 0 in every column. Recall our earlier observation that if column k contains

a 0 in an off-diagonal position, then vertex k cannot be a universal sink. Since we

found a 0 in every column, we found a 0 in every column k such that i < k |V|.

Moreover, we never examined any matrix entries in rows greater than i , and so we

never examined the diagonal entry in any column k such that i < k |V|. Therefore, all

the 0s that we found in columns k such that i < k |V| were off-diagonal. We conclude

that every vertex k such that i < k |V| cannot be a universal sink.

Thus, we have shown that every vertex less than i and every vertex greater than I

cannot be a universal sink. The only remaining possibility is that vertex i might be a

universal sink, and the call to IS-SINK checks whether it is.

To see that UNIVERSAL-SINK runs in O(V) time, observe that either i or j is

incremented in each iteration of the while loop. Thus, the while loop makes at most 2 |

V| 1 iterations. Each iteration takes O(1) time, for a total while loop time of O(V) and,

combined with the O(V )-time call to IS-SINK, we get a total running time of O(V).

=====The incidence matrix of a directed graph G = (V, E) is a |V| |E| matrix B = (bij)

such that: bij =-1 if edge j leaves vertex I; 1 if edge j enters vertex I; 0 otherwise

Describe what the entries of the matrix product B BT represent, where BT is the

transpose of B

Proof: BBT (i, j ) = aaaaaa

If i = j , then biebje = 1 (it is 1 1 or (1) (1)) whenever e enters or leaves

vertex i , and 0 otherwise.

If i != j , then biebje = 1 when e = (i, j ) or e = ( j, i ), and 0 otherwise.

Thus,

BBT (i, j ) =

degree of i = in-degree + out-degree if i = j ,

(# of edges connecting i and j ) if i != j .

===== Argue that in a breadth-first search, the value d[u] assigned to a vertex u is

independent of the order in which the vertices in each adjacency list are given. Using

Figure 22.3 as an example, show that the breadth-first tree computed by BFS can

depend on the ordering within adjacency lists.

Proof: The correctness proof for the BFS algorithm shows that d[u] = (s, u), and the

algorithm doesn.t assume that the adjacency lists are in any particular order.

In Figure 22.3, if t precedes x in Adj[w], we can get the breadth-rst tree shown in the

gure. But if x precedes t in Adj[w] and u precedes y in Adj[x], we can get edge (x, u)

in the breadth-rst tree.

=====Give an example of a directed graph G = (V, E), a source vertex s V , and a

set of tree edges E E such that for each vertex v V, the unique path in the graph

(V, E) from s to v is a shortest path in G, yet the set of edges E cannot be produced

by running BFS on G, no matter how the vertices are ordered in each adjacency list.

Proof: The edges in E are shaded in the following graph:

s

uw

vx

To see that E cannot be a breadth-rst tree, let.s suppose that Adj[s] contains u

before v. BFS adds edges (s, u) and (s, v) to the breadth-first tree. Since u is

enqueued before v, BFS then adds edges (u,w) and (u, x). (The order of w and x in

Adj[u] doesn.t matter.) Symmetrically, if Adj[s] contains v before u, then BFS adds

edges (s, v) and (s, u) to the breadth-rst tree, v is enqueued before u, and BFS adds

edges (v,w) and (v, x). (Again, the order of w and x in Adj[v] doesn.t matter.) BFS will

never put both edges (u,w) and (v, x) into the breadth-rst tree.

In fact, it will also never put both edges (u, x) and (v,w) into the breadth-rst tree .

====There are two types of professional wrestlers: "good guys" and "bad guys." Between

any pair of professional wrestlers, there may or may not be a rivalry. Suppose we have n

professional wrestlers and we have a list of r pairs of wrestlers for which there are rivalries.

Give an O(n + r)-time algorithm that determines whether it is possible to designate some of

the wrestlers as good guys and the remainder as bad guys such that each rivalry is between

a good guy and a bad guy. If is it possible to perform such a designation, your algorithm

should produce it. Proof: Create a graph G where each vertex represents a wrestler and

each edge represents a rivalry. The graph will contain n vertices and r edges.

Perform as many BFS.s as needed to visit all vertices. Assign all wrestlers whose distance

is even to be good guys and all wrestlers whose distance is odd to be bad guys. Then check

each edge to verify that it goes between a good guy and a bad guy. This solution would take

O(n + r ) time for the BFS, O(n) time to designate each wrestler as a good guy or bad guy,

and O(r ) time to check edges, which is O(n + r ) time overall.

====Show that edge (u, v) is

1.

a tree edge or forward edge if and only if d[u] < d[v] < f[v] < f[u],

2.

a back edge if and only if d[v] < d[u] < f[u] < f[v], and

3.

a cross edge if and only if d[v] < f[v] < d[u] < f[u].

a. Edge (u, v) is a tree edge or forward edge if and only if v is a descendant of u in the

depth-rst forest. (If (u, v) is a back edge, then u is a descendant of v, and if (u, v) is a cross

edge, then neither of u or v is a descendant of the other.)

By Corollary 22.8(Vertex v is a proper descendant of vertex u in the depth-first forest for a

(directed or undirected) graph G if and only if d[u] < d[v] < f[v] < f[u].), therefore, (u, v) is a

tree edge or forward edge if and only if d[u] < d[v] < f [v] < f [u].

b. First, suppose that (u, v) is a back edge. A self-loop is by denition a back edge. If (u, v)

is a self-loop, then clearly d[v] = d[u] < f [u] = f [v]. If (u, v) is not a self-loop, then u is a

descendant of v in the depth-rst forest, and by Corollary 22.8, d[v] < d[u] < f [u] < f [v].

Now, suppose that d[v] d[u] < f [u] f [v]. If u and v are the same vertex, then d[v] = d[u] < f

[u] = f [v], and (u, v) is a self-loop and hence a back edge. If u and v are distinct, then d[v] <

d[u] < f [u] < f [v]. By Theorem 22.7, interval [d[u], f [u]] is contained entirely within the

interval [d[v], f [v]], and u is a descendant of v in a depth-rst tree. Thus, (u, v) is a back

edge.

c. First, suppose that (u, v) is a cross edge. Since neither u nor v is an ancestor of the other,

Theorem 22.7 says that the intervals [d[u], f [u]] and [d[v], f [v]] are entirely disjoint. Thus, we

must have either d[u] < f [u] < d[v] < f [v] or d[v] < f [v] < d[u] < f [u]. We claim that we cannot

have d[u] < d[v] if (u, v) is a cross edge. Why? If d[u] < d[v], then v is white at time d[u]. By

Theorem 22.9(In a depth-first forest of a (directed or undirected) graph G = (V, E), vertex v is

a descendant of vertex u if and only if at the time d[u] that the search discovers u, vertex v

can be reached from u along a path consisting entirely of white vertices), v is a descendant

of u, which contradicts (u, v) being a cross edge. Thus, we must have d[v] < f [v] < d[u] < f

[u].

Now suppose that d[v] < f [v] < d[u] < f [u]. By Theorem 22.7, neither u nor v is a descendant

of the other, which means that (u, v) must be a cross edge.

=====Give a counterexample to the conjecture that if there is a path from u to v in a

directed graph G, and if d[u] < d[v] in a depth-first search of G, then v is a descendant of u in

the depth-first forest produced.Let us consider the example graph and depth-rst search

below.

df

w16

u23

Clearly, there is a path from u to v in G. The bold edges are in the depth-rst

forest produced. We can see that d[u] < d[v] in the depth-rst search but v is not

a descendant of u in the forest.

=====Give a counterexample to the conjecture that if there is a path from u to v in a

directed graph G, then any depth-first search must result in d[v] f[u].

Proof: Let us consider the example graph and depth-rst search below.

df

w16

u23

v45

uv

w

Clearly, there is a path from u to v in G. The bold edges of G are in the depth-rst

forest produced by the search. However, d[v] > f [u] and the conjecture is false.

=====Explain how a vertex u of a directed graph can end up in a depth-first tree containing

only u, even though u has both incoming and outgoing edges in G

Proof: Let us consider the example graph and depth-rst search below.

df

w12

u34

v56

wuv

Clearly u has both incoming and outgoing edges in G but a depth-rst search of G

produced a depth-rst forest where u is in a tree by itself.

====Let (u, v) be a minimum-weight edge in a graph G. Show that (u, v) belongs to some

minimum spanning tree of G

Proof: Theorem 23.1(Let G = (V, E) be a connected, undirected graph with a real-valued

weight function w defined on E. Let A be a subset of E that is included in some minimum

spanning tree for G, let (S, V - S) be any cut of G that respects A, and let (u, v) be a light

edge crossing (S, V - S). Then, edge (u, v) is safe for A) shows this.

Let A be the empty set and S be any set containing u but not v.

=====Give a simple example of a graph such that the set of edges {(u, v) : there exists a cut

(S, V - S) such that (u, v) is a light edge crossing (S, V - S)} does not form a minimum

spanning tree.

Proof: A triangle whose edge weights are all equal is a graph in which every edge is a light

edge crossing some cut. But the triangle is cyclic, so it is not a minimum spanning tree.

=====Show that a graph has a unique minimum spanning tree if, for every cut of the graph,

there is a unique light edge crossing the cut. Show that the converse is not true by giving a

counterexample.

Proof: Suppose that for every cut of G, there is a unique light edge crossing the cut. Let us

consider two minimum spanning trees, T and T_, of G. We will show that every edge of T is

also in T_, which means that T and T_ are the same tree and hence there is a unique

minimum spanning tree.

Consider any edge (u, v) T . If we remove (u, v) from T , then T becomes disconnected,

resulting in a cut (S, V S). The edge (u, v) is a light edge crossing the cut (S, V S) (by

Exercise 23.1-3). Now consider the edge (x, y) T_ that crosses (S, V S). It, too, is a light

edge crossing this cut. Since the light edge crossing (S, V S) is unique, the edges (u, v)

and (x, y) are the same edge. Thus, (u, v) T _. Since we chose (u, v) arbitrarily, every edge

in T is also in T_. Here.s a counterexample for the converse:

x

y

z

1

Here, the graph is its own minimum spanning tree, and so the minimum spanning tree is

unique. Consider the cut ({x} , {y, z}). Both of the edges (x, y) and (x, z) are light edges

crossing the cut, and they are both light edges.

=====Given a graph G and a minimum spanning tree T, suppose that we decrease the

weight of one of the edges in T. Show that T is still a minimum spanning tree for G. More

formally, let T be a minimum spanning tree for G with edge weights given by weight function

w. Choose one edge (x, y) T and a positive number k, and define the weight function w by

w'(u, v) = w(u,v) if (u,v) != (x,y); w(x,y) k if (u,v) != (x,y) Show that T is a minimum

spanning tree for G with edge weights given by w.

Proof: Let w(T ) = (x, y) T w(x, y). We have w(T ) = w(T ) k. Consider any other spanning

tree T, so that w(T ) w(T).

If (x, y) ! T, then w(T ) = w(T) w(T) > w(T ).

If (x, y) T, then w(T ) = w(T ) k w(T ) k = w(T ).

Either way, w(T) w(T ), and so T is a minimum spanning tree for weight function w.

=====Suppose that all edge weights in a graph are integers in the range from 1 to |V|.

How fast can you make Kruskal's algorithm run? What if the edge weights are

integers in the range from 1 to W for some constant W?

Proof: We know that Kruskal.s algorithm takes O(V) time for initialization, O(E lg E)

time to sort the edges, and O(E (V)) time for the disjoint-set operations, for a total

running time of O(V + E lg E + E (V)) = O(E lg E).

If we knew that all of the edge weights in the graph were integers in the range from 1

to |V|, then we could sort the edges in O(V + E) time using counting sort. Since the

graph is connected, V = O(E), and so the sorting time is reduced to O(E). This would

yield a total running time of O(V + E + E (V)) = O(E (V)), again since V = O(E), and

since E = O(E (V)). The time to process the edges, not the time to sort them, is now

the dominant term. Knowledge about the weights won.t help speed up any other part

of the algorithm, since nothing besides the sort uses the weight values.

If the edge weights were integers in the range from 1 to W for some constant W, then

we could again use counting sort to sort the edges more quickly. This time, sorting

would take O(E + W) = O(E) time, since W is a constant. As in the rst part, we get a

total running time of O(E (V)).

=====Given a weighted, directed graph G = (V, E) with no negative-weight cycles, let m be

the maximum over all pairs of vertices u, v V of the minimum number of edges in a

shortest path from u to v. (Here, the shortest path is by weight, not the number of edges.)

Suggest a simple change to the Bellman-Ford algorithm that allows it to terminate in m + 1

passes.

The proof of the convergence property shows that for every vertex v, the shortestpath

estimate d[v] has attained its nal value after length (any shortest-weight path to v)

iterations of BELLMAN-FORD. Thus after m passes, BELLMAN-FORD can terminate. We

don.t know m in advance, so we can.t make the algorithm loop exactly m times and then

terminate. But if we just make the algorithm stop when nothing changes any more, it will

stop after m+1 iterations (i.e., after one iteration that makes no changes).

BELLMAN-FORD-(M+1) (G,w, s)

INITIALIZE-SINGLE-SOURCE(G, s)

changes TRUE

while changes = TRUE

do changes FALSE

for each edge (u, v) E[G]

do RELAX-M(u, v,w)

RELAX-M(u, v,w)

if d[v] > d[u] + w(u, v)

then d[v] d[u] + w(u, v)

[v] u

changes TRUE

The test for a negative-weight cycle (based on there being a d that would change if another

relaxation step was done) has been removed above, because this version of the algorithm

will never get out of the while loop unless all d.s stop changing.

=====Suppose we change line 4 of Dijkstra's algorithm to the following: 4 while |Q| > 1 This

change causes the while loop to execute |V | - 1 times instead of |V | times. Is this proposed

algorithm correct?

Proof: Yes, the algorithm still works. Let u be the leftover vertex that does not get extracted

from the priority queue Q. If u is not reachable from s, then d[u] = (s, u) = . If u is

reachable from s, there is a shortest path p = s _ x u. When the node x was extracted,

d[x] = (s, x) and then the edge (x, u) was relaxed; thus, d[u] = (s, u).

=====We are given a directed graph G = (V, E) on which each edge (u, v) E has an

associated value r(u, v), which is a real number in the range 0 r(u, v) 1 that represents

the reliability of a communication channel from vertex u to vertex v. We interpret r(u, v) as

the probability that the channel from u to v will not fail, and we assume that these

probabilities are independent. Give an efficient algorithm to find the most reliable path

between two given vertices.

Proof: To nd the most reliable path between s and t, run Dijkstra.s algorithm with edge

weights w(u, v) = lg r (u, v) to nd shortest paths from s in O(E+V lg V) time. The most

reliable path is the shortest path from s to t, and that path.s reliability is the product of the

reliabilities of its edges. Here.s why this method works. Because the probabilities are

independent, the probability that a path will not fail is the product of the probabilities that its

edges will not fail. We want to nd a path s t such that (u,v) p r (u, v) is maximized. This

is equivalent to maximizing lg((u,v)pr(u, v))=(u,v)p lg r (u, v), which is in turn

equivalent to minimizing (u,v)p lg r(u, v). (Note: r (u, v) can be 0, and lg 0 is undefined.

So in this algorithm, define lg 0 = .) Thus if we assign weights w(u, v) = lg r (u, v), we

have a shortest-path problem.

Since lg 1 = 0, lg x < 0 for 0 < x < 1, and we have dened lg 0 = , all the weights w are

nonnegative, and we can use Dijkstra.s algorithm to nd the shortest paths from s in O(E +

V lg V) time.

Alternate answer

You can also work with the original probabilities by running a modied version of

Dijkstra.s algorithm that maximizes the product of reliabilities along a path instead of

minimizing the sum of weights along a path.

In Dijkstra.s algorithm, use the reliabilities as edge weights and substitute

max (and EXTRACT-MAX) for min (and EXTRACT-MIN) in relaxation and the queue,

for + in relaxation,

1 (identity for ) for 0 (identity for +) and (identity for min) for(identity for max).

For example, the following is used instead of the usual RELAX procedure:

RELAX-RELIABILITY(u, v, r )

if d[v] < d[u] r (u, v)

then d[v] d[u] r (u, v)

[v] u

This algorithm is isomorphic to the one above: It performs the same operations except that it

is working with the original probabilities instead of the transformed ones.

=====Let G = (V, E) be a weighted, directed graph with weight function w : E {0, 1, ...,

W } for some nonnegative integer W . Modify Dijkstra's algorithm to compute the shortest

paths from a given source vertex s in O(W V + E) time

Proof: Observe that if a shortest-path estimate is not , then it.s at most (|V| 1)W.

Why? In order to have d[v] < , we must have relaxed an edge (u, v) with d[u] < . By

induction, we can show that if we relax (u, v), then d[v] is at most the number of edges on a

path from s to v times the maximum edge weight. Since any acyclic path has at most |V| 1

edges and the maximum edge weight is W, we see that d[v] (|V|1)W. Note also that d[v]

must also be an integer, unless it is. We also observe that in Dijkstra.s algorithm, the

values returned by the EXTRACTMIN calls are monotonically increasing over time. Why?

After we do our initial |V| INSERT operations, we never do another. The only other way that

a key value can change is by a DECREASE-KEY operation. Since edge weights are

nonnegative, when we relax an edge (u, v), we have that d[u] d[v]. Since u is the minimum

vertex that we just extracted, we know that any other vertex we extract later has a key value

that is at least d[u]. When keys are known to be integers in the range 0 to k and the key

values extracted are monotonically increasing over time, we can implement a min-priority

queue so that any sequence of m INSERT, EXTRACT-MIN, and DECREASE-KEY

operations takes O(m + k) time. Here.s how. We use an array, say A[0 . . k], where A[ j] is a

linked list of each element whose key is j . Think of A[ j ] as a bucket for all elements with

key j . We implement each bucket by a circular, doubly linked list with a sentinel, so that we

can insert into or delete from each bucket in O(1) time.

We perform the min-priority queue operations as follows:

INSERT: To insert an element with key j , just insert it into the linked list in A[ j ]. Time: O(1)

per INSERT.

EXTRACT-MIN: We maintain an index min of the value of the smallest key extracted.

Initially, min is 0. To nd the smallest key, look in A[min] and, if this list is nonempty, use any

element in it, removing the element from the list and returning it to the caller. Otherwise, we

rely on the monotonicity property (and that there is no INCREASE-KEY operation) and

increment min until we either nd a list A[min] that is nonempty (using any element in A[min]

as before) or we run off the end of the array A (in which case the min-priority queue is

empty).

Since there are at most m INSERT operations, there are at most m elements in the minpriority queue. We increment min at most k times, and we remove and return some element

at most m times. Thus, the total time over all EXTRACTMIN operations is O(m + k).

DECREASE-KEY: To decrease the key of an element from j to i , rst check whether i j ,

agging an error if not. Otherwise, we remove the element from its list A[ j] in O(1) time and

insert it into the list A[i] in O(1) time.

Time: O(1) per DECREASE-KEY.

To apply this kind of min-priority queue to Dijkstra.s algorithm, we need to let k = (|V| 1)W,

and we also need a separate list for keys with value. The number of operations m is O(V +

E) (since there are |V| INSERT and |V| EXTRACTMIN operations and at most |E|

DECREASE-KEY operations), and so the total time is O(V + E + VW) = O(VW + E).

=====Let G = (V, E) be a weighted, directed graph with source vertex s, and let G be

initialized by INITIALIZE-SINGLE-SOURCE(G, s). Prove that if a sequence of relaxation

steps sets [s] to a non-NIL value, then G contains a negative-weight cycle.

Proof: Whenever RELAX sets for some vertex, it also reduces the vertex.s d value. Thus if

[s] gets set to a non-NIL value, d[s] is reduced from its initial value of 0 to a negative

number. But d[s] is the weight of some path from s to s, which is a cycle including s. Thus,

there is a negative-weight cycle.

=====Let G = (V, E) be a weighted, directed graph that contains no negative-weight cycles.

Let s V be the source vertex, and let G be initialized by INITIALIZE-SINGLE-SOURCE(G,

s). Prove that there exists a sequence of |V | - 1 relaxation steps that produces d[v] = (s, v)

for all v V

Proof: Suppose we have a shortest-paths tree G . Relax edges in G according to the

order in which a BFS would visit them. Then we are guaranteed that the edges along each

shortest path are relaxed in order. By the path-relaxation property, we would then have d[v]

= (s, v) for all v V. Since G contains at most |V| 1 edges, we need to relax only |V| 1

edges to get d[v] = (s, v) for all v V.

=====Let G be an arbitrary weighted, directed graph with a negative-weight cycle reachable

from the source vertex s. Show that an infinite sequence of relaxations of the edges of G

can always be constructed such that every relaxation causes a shortest-path estimate to

change

Proof: Suppose that there is a negative-weight cycle c = _v0, v1, . . . , vk_, where

v0 = vk , that is reachable from the source vertex s; thus, w(c) < 0. Without loss of generality,

c is simple. There must be an acyclic path from s to some vertex of c that uses no other

vertices in c. Without loss of generality let this vertex of c be v0, and let this path from s to v0

be p = _u0, u1, . . . , ul _, where u0 = s and ul = v0 = vk . (It may be the case that ul = s, in

which case path p has no edges.) After the call to INITIALIZE-SINGLE-SOURCE sets d[v] =

for all v V {s}, perform the following sequence of relaxations.

First, relax every edge in path p, in order. Then relax every edge in cycle c, in order, and

repeatedly relax the cycle. That is, we relax the edges (u0, u1), (u1, u2), . . . , (ul1, v0), (v0,

v1), (v1, v2), . . . , (vk1, v0), (v0, v1), (v1, v2), . . . , (vk1, v0), (v0, v1), (v1, v2), . . . , (vk1,

v0), . . ..

We claim that every edge relaxation in this sequence reduces a shortest-path estimate.

Clearly, the rst time we relax an edge (ui1, ui ) or (vj1, vj ), for

i = 1, 2, . . . , l and j = 1, 2, . . . , k 1 (note that we have not yet relaxed the last edge of

cycle c), we reduce d[ui] or d[vj ] from to a nite value. Now consider the relaxation of any

edge (vj1, vj ) after this opening sequence of relaxations.

We use induction on the number of edge relaxations to show that this relaxation reduces d[vj

].

Basis: The next edge relaxed after the opening sequence is (vk1, vk). Before relaxation,

d[vk ] = w(p), and after relaxation, d[vk ] = w(p)+ w(c) < w(p), since w(c) < 0.

Inductive step: Consider the relaxation of edge (vj1, vj ). Since c is a simple cycle, the last

time d[vj ] was updated was by a relaxation of this same edge. By the inductive hypothesis,

d[vj1] has just been reduced. Thus, d[vj1] + w(vj1, vj) < d[vj ], and so the relaxation will

reduce the value of d[vj ].

=====Let Q be an abstract decision problem on an instance set I , and let e1 and e2 be

polynomially related encodings on I . Then, e1(Q) P if and only if e2(Q) P.

Proof We need only prove the forward direction, since the backward direction is symmetric.

Suppose, therefore, that e1(Q) can be solved in time O(nk) for some constant k. Further,

suppose that for any problem instance i, the encoding e1(i) can be computed from the

encoding e2(i) in time O(nc) for some constant c, where n = |e2(i)|. To solve problem e2(Q),

on input e2(i), we first compute e1(i) and then run the algorithm for e1(Q) on e1(i). How long

does this take? The conversion of encodings takes time O(nc), and therefore |e1(i)| = O(nc),

since the output of a serial computer cannot be longer than its running time. Solving the

problem on e1(i) takes time O(|e1(i)|k) = O(nck), which is polynomial since both c and k are

constants

=====P = {L : L is accepted by a polynomial-time algorithm}.

Proof Since the class of languages decided by polynomial-time algorithms is a subset of the

class of languages accepted by polynomial-time algorithms, we need only show that if L is

accepted by a polynomial-time algorithm, it is decided by a polynomial-time algorithm. Let L

be the language accepted by some polynomial-time algorithm A. We shall use a classic

"simulation" argument to construct another polynomial-time algorithm A that decides L.

Because A accepts L in time O(nk) for some constant k, there also exists a constant c such

that A accepts L in at most T = cnk steps. For any input string x, the algorithm A simulates

the action of A for time T. At the end of time T, algorithm A inspects the behavior of A. If A

has accepted x, then A accepts x by outputting a 1. If A has not accepted x, then A rejects x

by outputting a 0. The overhead of A simulating A does not increase the running time by

more than a polynomial factor, and thus A is a polynomial-time algorithm that decides L

+++++++++=====Show how in polynomial time we can transform one instance of the

traveling-salesman problem into another instance whose cost function satisfies the triangle

inequality. The two instances must have the same set of optimal tours. Explain why such a

polynomial-time transformation does not contradict Theorem 35.3, assuming that P NP.

Proof: We transform an instance of the travelling salesman into another instance that sati_es

the triangle equality. Let k be the sum of weights of all edges. By adding k to all edges we

obtain an instance that sati_es the triangle inequality since k dominates the sums. The

optimal tour remains unchanged since all tours contain the same number of edges and we

therefore simply added nk to all tours. This does not contradict theorem 35:3 since the ratio

bound of the transformed problem does not give a constant ratio bound on the original

problem.

===Give an efficient greedy algorithm that finds an optimal vertex cover for a tree in linear

time

Proof: To construct an optimal vertex cover for a tree simply pick a node v such that v has at

least one leaf. Select v and remove v and all vertices and edges incident to v and continue

until no more vertices are left. Since the edges incident to the leafs of v needs to be covered

and we can do this optimally by selecting v the algorithm finds the optimal cover.

=====Give an example of a graph for which APPROX-VERTEX-COVER always yields a

suboptimal solution

Proof: A graph with two vertices and an edge between them has the optimal vertex cover of

consisting of a single vertex. However, APPROX-VERTEX-COVER returns both vertices in

this case.

=====The subgraph-isomorphism problem takes two graphs G1 and G2 and asks whether

G1 is isomorphic to a subgraph of G2. Show that the subgraph-isomorphism problem is NPcomplete

Proof: It is straightforward to show that the problem is in NP. To show that it is NP-hard we

reduce from the hamiltonian cycle problem. Let G be a graph G with n vertices. Clearly, G

has a hamiltonian cycle if and only if the cycle with n vertices Cn is isomorphic to a

subgraph of G.

=====The set-partition problem takes as input a set S of numbers. The question is whether

the numbers can be partitioned into two sets A and = S - A such that x x. Show that

the set-partition problem is NP-complete

Proof: Clearly, using the partion of the set as certificate shows that the problem is in NP. For

the NP-hardness we give a reduction from subsetsum. Given an instance S = {x1,,,,xn} and t

compute r such that r + t = ( xS x+r)/2 (r = ( xS x -2t). The instance for set-partition is then

R = {x1,,,xn}. We claim that S has a subset summing to t if and only if R can be partitioned.

Assume S has a subset S0 summing to t. Then the set S0 {r} sums to r + t and thus

partitions R. Conversely, if R can be partitioned then the set containing the element r sums

to r + t. All other elements in this set sum to t thus providing the subset S0.

34:5 6 Show that the hamiltonian-path problem is NP-complete

By exercise 34:2 - 6 the hamiltonian-path problem is in NP. To show that the problem is NPhard, construct a reduction from the hamilton-cycle problem. Given a graph G pick any

vertex v and make a .copy. of v, u that is connected to the same vertices as v. It can be

shown that this graph has a hamiltonian path from v to u if and only if G has a hamiltoncycle.

34:1 6 Show that the class P, viewed as a set of languages, is closed under union, intersection, concatenation, complement, and Kleene star. That is, if L1, L2 P, then L1 L2

P, etc Proof: Assume that L, L1, L2 P. The following statements hold:

_ L1 U L2 P since we can decide if x L1 U L2 by deciding if x L1 and then if x L2. If

either holds then x L1 U L2 otherwise it is not.

_ L1 L2 P since we can decide if x L1 and then if x L2. If both holds then x L1 L2

otherwise it is not.

_

!L P since x !L x ! L.

_ L1L2 P. Given a string x of length n denote its substring from index i to j by xij. We can

then decide x by deciding x1k 2 L1 and xk+1n 2 L2 for all the n possible values of k.

_ L* P. We can prove this showing that the result holds for Lk for all k and thus for U ki=0Lk.

We will use induction on k. If k = 0 we only consider the empty language and the result is

trivial. Assume that Lk P and consider Lk+1 = LLk. The above result on concatenation gives

us that LLk P.

34:2 3 Show that if HAM-CYCLE P, then the problem of listing the vertices of a

hamiltonian cycle, in order, is polynomial-time solvable

Proof: Assume that HAM-CYCLE P. First notice that for each nodes exactly two incident

edges participate in the cycle. We can find a hamilton cycle as follows. Pick a node v V

and let Ev be the edges incident to v. Compute a pair e 1; e2 Ev such that G = (V, (E - E v) U

{e1, e2}) contains a hamilton cycle. This can be done in polynomial time by trying all

possible pairs. Recursively apply the procedure on another node w for the graph G. This

produces in polynomial time a graph H = (V;C) where C E is a hamiltonian cycle.

34:2 - 9 show that P co - NP. Proof: Assume that L P. Since P is closed under

complement we have that !L P and thus !L NP giving us that L co - NP.

34:3 2 Show that the P relation is a transitive relation on languages. That is, show that if

L1 P L2 and L2 P L3, then L1 P L3

Proof: Then there exists polynomial-time computable reduction functions f1 and f2 such that:

_ x L1 => f1(x) L2

_ x L2 => f2(x) L3

The function f2(f1(x)) is polynomial-time computable and sati_es that x L1 <=> f2(f1(x))

L3 thus giving us that L1 P L3.

34:3 6 A language L is complete for a language class C with respect to polynomial-time

reductions if L C and L P L for all L C. Show that and {0,1}* are the only languages in

P that are not complete for P with respect to polynomial-time reductions.

Proof: Assume that L P. We wish to show that L is P-complete unless it is the empty

language or {0; 1}*.Given L P we can reduce it to L simply by using the polynomial-time

algorithm for deciding L to construct the reduction function f. Given x decide if x L0 and set

f(x) such that f(x) L if x L and f(x) ! L otherwise. Clearly, this is not possible for the two

trivial languages mentioned above.

34:3 7 Show that L is complete for NP if and only if !L is complete for co-NP

Proof: Assume L NPC. Then L NP giving us that !L co - NP. Assume further that L P L

for all L NP. This means that x !L f(x) !L for some polynomial-time reduction function

f. Then we have that x !L f(x) !L which means that !L P !L for all !L co - NP. The

converse can be shown similarly.

33:1 4 Show how to determine in O(n2 lg n) if three point are collinear in a set of n points.

Proof: For each point p0 sort the n - 1 other points according to the polar angle with respect

to p. If two points p1 and p2 have the same polar angle then p0, p1 and p2 are collinear.

This can approach can be implemented in O(n 2 lg n).

33:3 2 Show that a lower bound of computing a convex hull of n points is (n lg n) if the

computation model has the same lower bound for sorting. Proof: The n points can be

transformed into points in the plane by letting a point i map to the point (i; i 2). Computing the

convex hull outputs the points in counterclockwise order and thus sorted.

22:1 1 Given an adjacency-list representation of a directed graph, how long does it take to

compute the out-degree of every vertex? How long does it take to compute the in-degrees

Proof: Given the adjacency-list representation of a graph the out and in-degree of every

node can easily be computed in O(E + V) time.

22:1 3 The transpose of a directed graph G = (V, E) is the graph GT = (V, ET), where ET =

{(v, u) V V : (u, v) E}. Thus, GT is G with all its edges reversed. Describe efficient

algorithms for computing GT from G, for both the adjacency-list and adjacency-matrix

representations of G. Analyze the running times of your algorithms

Proof: The transpose of a directed graph GT can be computed as follows: If the graph is

represented by an adjacency-matrix A simply compute the transpose of the matrix AT in time

O(V2). If the graph is represented by an adjacency-list a single scan through this list is

suf_cient to construct the transpose. The time used is O(E + V).

22:1 5 The square of a directed graph G = (V, E) is the graph G2 = (V, E2) such that (u, w)

E2 if and only if for some v V , both (u, v) E and (v, w) E. That is, G2 contains an

edge between u and w whenever G contains a path with exactly two edges between u and

w. Describe efficient algorithms for computing G2 from G for both the adjacency-list and

adjacency-matrix representations of G. Analyze the running times of your algorithms

Proof: The square of a graph can be computed as follows: If the graph is represented as an

adjacency matrix A simply compute the product A 2 where multiplication and addition is

exchanged by and'ing and or'ing. Using the trivial algorithm yields a running time of O(n 3).

Using Strassens algorithm improves the bound to O(n lg 7) = O(n2;81).

If we are using the adjacency-list representation we can simply append lists eliminating

duplicates. Assuming the lists are sorted we can proceed as follows: For each node v in

some list replace the v with Adj[v] and merge this into the list. Each list can be at most V

long and therefore each merge operation takes at most O(V) time. Thus the total running

time is O((E+V)V) = O(E+V2).

22:1 6 When an adjacency-matrix representation is used, most graph algorithms require

time (V2), but there are some exceptions. Show that determining whether a directed graph

G contains a universal sink-a vertex with in-degree |V| - 1 and out-degree 0-can be

determined in time O(V), given an adjacency matrix for G

Proof: Notice that if edge is present between vertices v and u then v cannot be a sink and if

the edge is not present then u cannot be a sink. Searching the adjancency-list in a linear

fashion enables us to exclude one vertex at a time.

22:2 7The diameter of a tree T =(V, E) is given by max u,v V (u,v) that is, the diameter is

the largest of all shortest-path distances in the tree. Give an efficient algorithm to compute

the diameter of a tree, and analyze the running time of your algorithm

Proof: The diameter of a tree can computed in a bottom-up fashion using a recursive

solution. If x is a node with a depth d(x) in the tree then the diameter D(x) must be:

D(x) = max{maxi{D(x:childi)}; maxij{d(x:childi) + d(x:childj)} + 2g; if x is an internal

node

0 if x is a leaf

Since the diameter must be in one of the subtrees or pass through the root and the longest

path from the root must be the depth. The depth can easily be computed at the same time.

Using dynamic programming we obtain a linear solution.

Actually, the problem can also be solved by computing the longest shortest path from an

arbitrary node. The node farthest away will be the endpoint of a diameter and we can thus

compute the longest shortest path from this node to obtain the diameter.

23:1 3 Show that if an edge (u; v) belongs to some minimum spanning tree then it is a

light edge of some cut.

Proof: Consider a cut with u and v on separate sides. If (u; v) is not a light edge then clearly

the graph would not be a minimum spanning tree.

23:1 5 Let e be a maximum-weight edge on some cycle of G = (V, E). Prove that there is a

minimum spanning tree of G = (V, E -{e}) that is also a minimum spanning tree of G. That is,

there is a minimum spanning tree of G that does not include e

Proof: Show that if e be the maximum-weight edge on some cycle in G = (V; E) then e is not

part of a minimum spanning tree of G. If e was in the minimum spanning tree then replacing

it with any other edge on the cycle would yield a .better. minimum spanning tree.

23:2 1 Kruskal's algorithm can return different spanning trees for the same input graph G,

depending on how ties are broken when the edges are sorted into order. Show that for each

minimum spanning tree T of G, there is a way to sort the edges of G in Kruskal's algorithm

so that the algorithm returns T

Proof: Given a minimum spanning tree T we wish to sort the edges in Kruskal's algorithm

such that it produces T. For each edge e in T simply make sure that it preceeds any other

edge not in T with weight w(e).

23:2 3 Is the Fibonacci-heap implementation of Prim's algorithm asymptotically faster than

the binary-heap implementation for a sparse graph G = (V, E), where |E| = (V)? What

about for a dense graph, where |E| = (V2)? How must |E| and |V| be related for the

Fibonacci-heap implementation to be asymptotically faster than the binary-heap

implementation?

Proof: Consider the running times of Prims algorithm implemented with either a binary heap

or a Fibonnacci heap. Suppose |E| = (V) then the running times are:

_ Binary: O(E lg V) = O(V lg V)

_ Fibonnacci: O(E + V lg V) = O(V lg V)

If |E| = (V2) then:

_ Binary: O(E lg V) = O(V2 lg V)

_ Fibonnacci: O(E + V lg V) = O(V2)

The Fibonnacci heap beats the binary heap implementation of Prims algorithm when |E| =

(V) since O(E + V lg V) = O(V lg V) t for |E| = O(V lg V) but O(E lg V) = (V lg V) for |E| =

(V). For |E| = (V lg V) the Fibonnacci version clearly has a better running time than the

ordinary version.

21:3 2 Write a nonrecursive version of FIND-SET with path compression

Proof: The following algorithm does the job using pointer reversal:

Algorithm 11 FIND-SET(x)

Input: The node x.

Output: The root of the tree root to the representative of x.

y p[x]

p[x] x

while x y do

z p[y]

p[y] x

x y

y z

end while

root x

y p[x]

p[x]

x

while x y do

z

p[y]

p[y]

root

x y

y

z

end while

return root

The _rst while loop traverses the path from x to the root of the tree while reversing all parent

pointers along the way. The second while loop returns along this path and sets all parent

pointers along the way to point to the root.

21:4 - 2 every node has rank at most blg nc and therefore the height h of any subtree is at

most blg nc. Clearly, each call to FIND-SET (and thus UNION) takes at most O(h) = O(lg n)

time.

- MELJUN CORTES ALGORITHM Decrease-And-Conquer Algorithm Design Technique IUploaded byMELJUN CORTES, MBA,MPA
- Glasnik RadUploaded bydragance106
- Proceedings O_Sedgewick Robert et al.pdfUploaded byduylinh65
- kernellink.pdfUploaded byoreske
- 50120140506002Uploaded byIAEME Publication
- Further Results on Product Cordial LabelingUploaded byDon Hass
- Brandes U., Erlebach T. eds. Network analysis. Methodological foundations 2005ã. 482ñ. ISBN ISBN10 3-540-24979-6.pdfUploaded bydeisecairo
- Getting Started in Sport of ProgrammingUploaded byShivasheesh Yadav
- On Pathos Total Semitotal and Entire Total Block Graph of a TreeUploaded byDon Hass
- L18.pdfUploaded bySumit Bide
- chordal (1)Uploaded byJJ
- MyL14Uploaded bySimran Singh
- EXL CASE STUDY 2012 SOLUTION-RPR-001Uploaded byAbhishek Ghosh
- DijkstraShortestPath(Jung2 2.0 API)Uploaded bydaniele4ne
- [Marco Slikker, Anne Van Den Nouweland (Auth.)] So(B-ok.org)Uploaded byRogel77
- INFOCOM2009-kanonUploaded byroyamit_1989
- 1108.2357Uploaded byNedeliov Nicolae
- Dijkstra1Uploaded byMohd Iqbal Othman
- s7252 Jayadharini Jaiganesh Efficient Connected ComponentsUploaded byDeevashwer Rathee
- Fundamentals of Data StructuresUploaded byramrammithra
- Cbazgan EnUploaded byYassine Bouigua
- Cs 502 Final Reference Subjective sUploaded bymc190204637 AAFIAH TAJAMMAL
- Paper 3-Measuring Homophily in Social Network Identification of Flow of Inspiring Influence Under New Vistas of Evolutionary DynamicsUploaded byimrana10
- 09-Gargantuan Computing—GRIDs and P2PUploaded byJason
- Robust Local Graph Structure for Texture ClassificationUploaded bySDIWC
- bt1Uploaded byarvchamp
- CS174: note08Uploaded byjuggleninja
- lesson plan 2Uploaded byapi-271875404
- Chapter 4Uploaded byVivek Shrivastava
- Computing and Application on Complex Bus Network SystemUploaded bySEP-Publisher

- docker-tutorial.pdfUploaded byDavid Antonio Carvallo Ponce
- 771 Scen Overview en US OutboundUploaded byleopritam
- DB2_for_Oracle_Pro_HOL_Workbook.pdfUploaded bypiciul2010
- Kumar Cheruku -- Oracle PLSQL Developer , NJ USUploaded byVictor Ray
- D96139GC30_ag.pdfUploaded byrana1812
- Creating Custom Oracle Workflow using AME.pdfUploaded bymfarzadk123
- EFF: spacedraftUploaded byEFF
- thinmanage-3-1-adminguide-rev12Uploaded byjmims17
- OR34_Brochure_LR_C002_105_1Uploaded byiplascak
- Basic Financials Inc Fixed Assets 9.1 TOCUploaded byNaveed Arshad
- CS615Uploaded byshariya12
- Winfiol ScriptUploaded bynaveen_nagesh
- PanelDraw_4.1_Manual.pdfUploaded byAkula Veerraju
- Cmts Cpe BasicoUploaded byMickynegrito Terceros
- au-t2-s-1505-stem-in-year-5-assessment-sheet ver 1Uploaded byapi-374713509
- Harvesting and Analyzing Tweets Using RUploaded byRubila Dwi Adawiyah
- Role of Mobile Banking in Present Scenario IndexUploaded byAditya Sawant
- PI DataLink 2014 User GuideUploaded bySantoso Wijaya
- ScaN InstructorPPT Chapter7 FinalUploaded byLucyana Angel Christine Butarbutar
- realflow_lightwave_connectivity.pdfUploaded byJesus Manuel González
- Android Speech to Text Converter for SMS ApplicationUploaded bySachin
- Panel c.inc. Guia Usuario M-V128fbp-9Uploaded byNelson Ernesto Morales Itas
- PassKit PGUploaded byPanji Agus Muttaqin
- BraumatUploaded byMiguel Angel Benitez Ambrosio
- HADR Multiple Standbys Jn20[1]Uploaded bySantosh Ibrampurkar
- JEB -Nov 9 07 - Dec o6-June 07Uploaded byRakesh Ghimire
- PI TechnicalUploaded byapi-3756014
- Mikrotik - Multiple Hotspot on Single DeviceUploaded byHamami InkaZo
- Introduction to Python and LaTeXt.pdfUploaded byJaime Molina
- Why I Like Virtualization (and Why Hardware Sucks) _ Bob Plankers, The LoneUploaded byEhtesham Opel

## Much more than documents.

Discover everything Scribd has to offer, including books and audiobooks from major publishers.

Cancel anytime.