COS 226 Algorithms and Data Structures Fall 2014

Final

This test has 14 questions worth a total of 100 points. You have 180 minutes. The exam is closed book, with the
exception of a one page cheatsheet. No calculators or other electronic devices are permitted. Write out and sign
the Honor Code pledge just before turning in the test.
This exam is preprocessed by computer. Please use a pen; if you use a pencil, be sure to write
darkly. Do not write any answers outside of the designated frames. And do not write on the corners.

“I pledge my honor that I have not violated the Honor Code during this examination.”

Name:

netID:

Room:

P01 P02 P03 P03A P04 P04A

Precept:

Problem Score Problem Score P01 F 9 Andy Guna

0 7 P02 F 10 Jérémie Lumbroso
1 8 P03 F 11 Josh Wetzel
2 9 P03A F 11 Jérémie Lumbroso
3 10 P04 F 12:30 Robert MacDavid
4 11 P04A F 13:30 Shivam Agarwal
5 12
6 13
Sub 1 Sub 2

Total
0. Initialization (2 points)
In the space provided on the front of the exam, write your name and Princeton netID; fill in your precept number;
write the name of the room in which you are taking the exam; and write and sign the honor code.

1. Digraph Traversal (6 points)

Consider the following digraph. Assume the adjacency lists are in sorted order: for example, when iterating
through the edges pointing from vertex 5, consider the edge 5 → 3 before the others.

8
1 7

4 5
0
6

3
2

(a) Starting from vertex 0, run a depth-first search of the digraph, and list the vertices in reverse postorder.

(b) Starting from vertex 0, run a depth-first search of the digraph, and list the vertices in preorder.
2. Analysis of Algorithms (5 points)
For each code fragment on the left, check the best matching order of growth of the running time. You may use an
answer more than once or not at all.

N log N N log N R+N RN N + R2 (N + R) log N N (N + R)

int x = 1 , i ;
for ( i = 0; i < N ; i ++)
x ++;

public static int f2 ( int N ) {

int x = 1;
while ( x < N )
x = x * 2;
return x ;
}

int x = 0 , i ;
for ( i = 0; i < N ; i ++)
x += f2 ( N );

int x = 1, i, j;
for ( i = 0; i < N ; i ++)
for ( j = 1; j < R ; j ++)
x = x * j;

int x = 0 , i , j ;
for ( i = 1; i <= N ; i ++)
for ( j = 1; j <= N + R ; j += i )
x += j ;
3. String Sorting Algorithms (7 points)
The column on the left is the original input of 24 strings to be sorted; the column on the right are the strings
in sorted order; the other 7 columns are the contents at some intermediate step during one of the 3 radix sorting
algorithms listed below.
Match up each column with the corresponding sorting algorithm. You may use a number more than once.
Hint: think about algorithm invariants; do not trace code.

leaf cost hash edge rank load find cost cost

size edge edge cost hash leaf load edge edge
null flow cost fifo edge heap size find fifo
type find heap flow leaf swap type fifo find
cost fifo fifo find heap node trie flow flow
sink heap flow heap less fifo node heap hash
heap hash find hash next edge edge hash heap
trie leaf leaf leaf fifo trie time leaf leaf
loop loop loop loop time swim leaf loop less
flow less load load find null push less list
less load less less sink time hash load load
node list list list list find sink list loop
find null next next size sink rank null next
next node node node flow rank null node node
fifo next push push load loop swim next null
push push rank rank node flow fifo push push
rank rank trie trie loop type heap rank rank
load size sink sink cost push loop size sink
edge sink type type trie hash swap sink size
hash swap time time null less less swap swap
time swim swap swap push cost cost swim swim
swap type null null swap list next type time
list trie swim swim swim next list trie trie
swim time size size type size flow time type

0 4

(0) Original input (2) MSD radix sort

(1) LSD radix sort (3) 3-way radix quicksort (no shuffle)

(4) Sorted
4. Substring Search (8 points)
(a) Consider the Knuth-Morris-Pratt DFA for the following string of length 8:
C A C A C B C B
Complete the first row of the table.

0 1 2 3 4 5 6 7
A
B 0 0 0 0 0 6 0 8
C 1 1 3 1 5 1 7 1

(b) Suppose that you run the Boyer-Moore algorithm (the basic version considered in the textbook and lecture)
to search for the pattern
M Y F A T H E
in the text
Y B R O T H E R TFinal,
H A T FFall
A T H2014
E R W A S M Y F A T H E R T
Give the trace of the algorithm in the grid below, circling the characters in the pattern that get compared
with characters in the text.

Y B R O T H E R T H A T F A T H E R W A S M Y F A T H E R T
M Y F A T H E

Y B R O T H E R T H A T F A T H E R W A S M Y F A T H E R T
M Y F A T H E
5. Minimum Spanning Tree Algorithms (6 points)
Each of the figures below represents a partial spanning tree. Determine whether it could possibly be obtained from
(a prematurely stopped) Prim’s algorithm, (a prematurely stopped) Kruskal’s algorithm, both or neither.

Prim Kruskal Both Neither

15 14 9 13
10 8
3 6
3 1
4 1 2
7 12 11

13 17 6 1

15 14 9 13
10 8
3 6
3 1
4 1 2
7 12 11

13 17 6 1

15 14 9 13
10 8
3 6
3 1
4 1 2
7 12 11

13 17 6 1

15 14 9 13
10 8
3 6
3 1
4 1 2
7 12 11

13 17 6 1

15 14 9 13
10 8
3 6
3 1
4 1 2
7 12 11

13 17 6 1

15 14 9 13
10 8
3 6
3 1
4 1 2
7 12 11

13 17 6 1
 Final, Fall 2014

6. Maximum Flow (7 points)

Consider the following flow network and feasible flow f from from the source vertex A to the sink vertex J.

flow capacity

A 7/7 B 0/5 C ?/3 D 1 / 14 E

2 3 ?/
6/6 /
16 7/7 13 0/7 /
12 0 / 10 7 1 / 10
/
0

F 6 / 10 G 15 / 15 H 15 / 19 I 12 / 17 J

(a) Check the value of the flow on edge C → D?

augmenting path: A-G-B-C-I-J
0 1 2 bottleneck
3 4 capacity: 3
min cut: { A, B, C, F, G }
max flow value = 18
C->D
(b) Check the value of the flowhas
f . flow 3
D->J has flow 1
0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15

16 17 18 19 20 21 22 23 24 25 26 27 28 29 30

(c) Starting from the flow f , perform one iteration of the Ford-Fulkerson algorithm. List the sequence of vertices
on the augmenting path.

(d) Check the value of the maximum flow?

0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15

16 17 18 19 20 21 22 23 24 25 26 27 28 29 30

(e) Check the vertices on the source side of a minimum cut.

A B C D E F G H I J
7. Properties of Algorithms (9 points)
Check whether each of the following statements are True or False.

(a) Shortest paths. Consider an edge-weighted digraph G with distinct and positive edge weights, a source
vertex s, and a destination vertex t. Assume that G contains at least 3 vertices, has no parallel edges or self
loops, and that every vertex is reachable from s.
True False
Any shortest s → t path must include the lightest edge.
Any shortest s → t path must include the second lightest edge.
Any shortest s → t path must exclude the heaviest edge.
The shortest s → t path is unique.

(b) Minimum spanning trees. Consider an edge-weighted graph G with distinct and positive edge weights.
Assume that G contains at least 3 vertices, has no parallel edges or self loops, and is connected.
True False
Any MST must include the lightest edge.
Any MST must include the second lightest edge.
Any MST must exclude the heaviest edge.
The MST is unique.

(c) Burrows-Wheeler transform.

True False
Any input x consisting of an integer (between 0 and N − 1) followed by N characters is the
Burrows-Wheeler transform of some string s of length N .

If the Burrows-Wheeler transforms of s and t are equal, then s = t.

If the Burrows-Wheeler inverse transforms of x and y are equal, then x = y.

In practice, applying the Burrows-Wheeler transform is significantly faster than applying

the Burrows-Wheeler inverse transform.
8. Huffman Trees (4 points)
Consider the string “DATA-STRUCTURES-AND-ALGORITHMS”: which of the following trees is an optimal prefix-free
code for this input string?

Optimal Prefix-Free Code Not Optimal Prefix-Free Code

R A - T

N D U S

C E G I H M L O

T A S R -

U D

L O H M N G I

C E

T - R A

S U H M

N D C E G I L O

R A T

C E - S N D U

H M G I L O
 9. LZW Compression (5 points)
What is the result of compressing the following string of length 15 using LZW compression?

B B B B B B C A B B C B B B C

Assume the original encoding table consists of all 7-bit ASCII characters and uses 8-bit codewords. Recall that
codeword 80 is reserved to signify end of file.

42 80

5.5 Data Compression 815

For reference, below is the hexadecimal-to-ASCII conversion table from the textbook:

When you HexDump a bit- 0 1 2 3 4 5 6 7 8 9 A B C D E F

ns ASCII-encoded charac- 0 NUL SOH STX ETX EOT ENQ ACK BEL BS HT LF VT FF CR SO SI

ght is useful for reference. 1 DLE DC1 DC2 DC3 DC4 NAK SYN ETB CAN EM SUB ESC FS GS RS US

hex number, use the first 2 SP ! " # $ % & ‘ ( ) * + , - . /

index and the second hex 3 0 1 2 3 4 5 6 7 8 9 : ; < = > ?
ndex to find the character 4 @ A B C D E F G H I J K L M N O
r example, 31 encodes the 5 P Q R S T U V W X Y Z [ \ ] ^ _
the letter J, and so forth. 6 ` a b c d e f g h i j k l m n o
bit ASCII, so the first hex 7 p q r s t u v w x y z { | } ~ DEL
ess. Hex numbers starting
Hexadecimal-to-ASCII conversion table
the numbers 20 and 7F)
n-printing control charac-
control characters are left over from the days when physical devices
s were controlled by ASCII input; the table highlights a few that you
 10. Burrows-Wheeler Transform (6 points)
(a) What is the Burrows-Wheeler transform of the following?

A D D B D B C A

(b) What is the Burrows-Wheeler inverse transform of the following?

COS 226 FINAL, FALL 2014 2 COS 226 FINAL, FALL 2014 7
C A D D A B C C

5. Burrows-Wheeler transform. (8 points)

5. Burrows-Wheeler transform. (8 points)

(a) What is the Burrows-Wheeler transform (a)

of the
What
following?
is the Burrows-Wheeler transform of the
A D D B D B C A D D B D B C

Feel free to use both of these grids for scratch work.

Feel free to use this grid for scratch work. Feel free to use this grid for scratch work.

(b) What is the Burrows-Wheeler inverse transform

(b) What
of is
thethe
following?
Burrows-Wheeler inverse transform
2 2
 Collection of DNA fragments (Final, Fall 2014)

11. Algorithm and Data Structure Design (13 points)

Design a data type to store a collection of gene fragments over the DNA alphabet {A, C, T, G}, according to the
following API:

public class FragmentCollection

public FragmentCollection() create an empty collection of DNA fragments

public void add(String fragment) add the DNA fragment to the collection

public int prefixCount(String p) number of DNA fragments that start with prefix p

Here is an example:

Fr ag me ntC ol le cti on fc = new F ra gme nt Co lle ct io n ();

fc . add ( " AC " );
fc . add ( " TACG " );
fc . add ( " TCGAA " );
fc . add ( " CGA " );
fc . add ( " AGCT " );
fc . add ( " TCGG " );
fc . add ( " TCGG " ); // added twice , will be counted twice
fc . prefixCount ( " " ); // returns 7 ( number of adds )
fc . prefixCount ( " T " ); // returns 4 ( TACG , TCGAA , TCGG , TCGG )
fc . prefixCount ( " TC " ); // returns 3 ( TCGAA , TCGG , TCGG )
fc . prefixCount ( " G " ); // returns 0

Give a crisp and concise English description of your data structure. Your answer will be graded on correctness,96
efficiency, and clarity.
(a) Declare the instance variables for your FragmentCollection data type. You may use nested data types.
public class FragmentCollection {

}
(b) Briefly describe how to implement each of the operations, using either prose or code.
• public void add(String fragment):

• public int prefixCount(String p):

(c) What is the order of growth of prefixCount(p) as a function of the number N of keys added, the length W
of the prefix p, the alphabet size R, and the number M of fragments that match the given prefix p?
NR
1 log N N W W + log N W +M N + W log R WR
WR
12. Reductions (13 points)
Consider the following two graph-processing problems:
• Shortest-Path. Given an edge-weighted digraph G with nonnegative edge weights, a source vertex s and
a destination vertex t, find a shortest path from s to t.

• Shortest-Teleport-Path. Given an edge-weighted digraph G with nonnegative edge weights, a source

Final, Fall
vertex s and a destination vertex t, find a shortest path2014
from s to t where you are permitted to teleport
across one edge for free. That is, the weight of a path is the sum of the weights of all of the edges in the
path, excluding the largest one.
For example, in the edge-weighted digraph below, the shortest path from s to t is s → w → t (with weight 11) but
the the shortest teleport path is s → u → v → t (with weight 3).

u 2 v

weight
8

99
1

source s 5 w 6 t destination

(a) Design a linear-time reduction from Shortest-Path to Shortest-Teleport-Path. To demonstrate your

reduction, draw the edge-weighted digraph (and label the source and destination vertices) that you would
construct to solve the Shortest-Path problem on the digraph above. You may additionally explain your
construction with a few concise sentences.
(b) Design a linear-time reduction from Shortest-Teleport-Path to Shortest-Path. To demonstrate your
reduction, draw the edge-weighted digraph (and label the source and destination vertices) that you would
construct to solve the Shortest-Teleport-Path problem on the digraph given in the previous page. You
may additionally explain your construction with a few concise sentences.

(c) Determine whether each of following statements can be infered from the fact that Shortest-Path and
Shortest-Teleport-Path linear-time reduces to one another. For simplicity, assume E ≥ V .

Yes No
If there exists an E log log E algorithm for Shortest-Teleport-Path, then there exists an
E log log E algorithm for Shortest-Path.

If there exists an E log log E algorithm for Shortest-Path, then there exists an E log log E
algorithm for Shortest-Teleport-Path.

If there does not exist a linear-time algorithm for Shortest-Path, then there does not exists
a linear-time algorithm for Shortest-Teleport-Path.

If there does not exist a linear-time algorithm for Shortest-Teleport-Path, then there
does not exists a linear-time algorithm for Shortest-Path.
13. Problem Identification (9 points)
You are applying for a job at a new software technology company. Your interviewer asks you to identify the
following tasks as either possible (with algorithms and data structures introduced in this course), impossible, or an
open research problem.

Possible Impossible Open

Given a digraph, find a directed cycle that is simple (if one exists) in time
proportional to E + V . A simple cycle is a cycle that has no repeated vertices
other than the requisite repetition of the first and last vertex.

Given an edge-weighted digraph in which all edge weights are either 1 or 2 and
two vertices s and t, find a shortest path from s to t in time proportional to
E+V.

Given an edge-weighted DAG with positive edge weights and two vertices s and
t, find a path from s to t that maximizes the product of the weights of the edges
participating in the path in time proportional to E + V .

Given an edge-weighted graph with positive edge weights, find a spanning tree
that maximizes the product of the weights of the edges participating in the span-
ning tree in time proportional to E + V .

Given an edge-weighted graph with positive edge weights and two distinguished
vertices s and t, find a simple path (no repeated vertices) between s and t that
maximizes the sum of the weights of the edges participating in the path in time
proportional to E V .

Given a flow network and a mincut in that flow network, find a maxflow in time
proportional to E + V .

Given an array of N strings over the DNA alphabet {A, C, T, G}, determine
whether all N strings are distinct in time linear in the number of characters in
the input.

Given an array a of N 64-bit integers, determine whether there are two indices
i and j such that ai + aj = 0 in time proportional to N .

Given an array of N integers between 0 and R2 − 1, stably sort them in time

proportional to N + R.