Design and Analysis of Algorithms, Assignment 1 Solution

Each part of this assignment will be submitted as a separate le. As you nish up the various parts, submit your le using WolfWare Classic, at https://submit.ncsu.edu/. If you improve part of your solution, you can re-submit the same le, and we’ll automatically grade the latest version of the le you submit.

This assignment includes a programming problem and some problems where you get to write up a short answer, an argument or an explanation. For the non-programming problems, you’ll need to turn in a PDF le for your solution. You can prepare the solution in any software package you like (e.g., Microsoft Word, Open O ce, LaTeX), but you’ll need to use correct mathematical notation in whatever package you choose, and you’ll need to create a PDF version of your work when you turn it in.

For the programming problems, be sure your solution compiles and executes successfully on one of the EOS Linux systems. You can access these machines from the public computing labs, and they’re available remotely. The university has a page at http://www.eos.ncsu.edu/remoteaccess/ with brief instructions about accessing and using these systems. I’ll post some additional instructions on our course homepage.

Depending on your choice of programming language, here are some things you’ll need to consider to make sure your programming solutions work as well for us when we’re grading them as they do for you when you’re writing them.

Programming in C

If you’re programming in C, we’re going to build your executable using a command like the following. The behavior of C and C++ programs can vary from platform to platform, so you’ll de nitely want to give yourself enough time try out your work on an EOS Linux before you turn it in.

gcc -Wall -std=c99 -g -O2 -o program program.c

Programming in C++

If you’re programming in C++, we’re going to build your executable using a command like the following. The behavior of C and C++ programs can vary from platform to platform, so you’ll de nitely want to give yourself enough time try out your work on an EOS Linux before you turn it in.

g++ -Wall -g -O2 -o program program.cpp

Java

For a Java program, don’t put your classes in a package; just leave them in the default package. This will make it easy for us to compile and test your program without having to match your directory structure.

For Java, note that the default compiler on the EOS Linux machines is kind of old, version 1.6. Most of the features we’re likely to need will be there in this version of the language, but there are two missing things that sometimes catch people by surprise. In a switch statement, you can’t use a string as the switch value (added in 1.7), so you’ll have to do this kind of thing with nested if statements. Also, when instantiating a generic class, you won’t get the left-to-right type inference to help shorten your code (also added in 1.7). So, for example, you’ll have to use:

ArrayList< Integer > list = new ArrayList< Integer >();

instead of:

ArrayList< Integer > list = new ArrayList< >();


1




When we compile and run your Java programs, we’ll use commands like the following (this is why you shouldn’t put your code inside a package). Here, you can see we expect your main class to be lower-case. This isn’t how you normally name classes in Java, but it makes the java lenames more consistent with the other languages:

javac program.java

java program

Python

Right now, it looks like Python 3 isn’t installed on the EOS Linux systems, so we’ll have to stick to Python 2 for our programming assignments. When we run your python programs, we’ll do it like this:

python2 program.py

C#

It looks like we don’t have a C# compiler installed on the EOS Linux machines. I’m asking about this, but we’ll have to get by without C# on homework assignment 1.

    1. For this problem, you’ll be comparing functions that describe the running times of two di erent algo-rithms (well, at least we’ll be pretending they do). For each question, you’ll need to do two things. Report which of the two algorithms is asymptotically faster (i.e., which function has a slower rate of growth). Also, report the trade-o point between the two functions, the smallest value n0 such that, for all n n0, the faster-growing function has a larger value at n than the slower-growing function. Give this answer rounded to one fractional digit of precision (e.g., a number line 25.2 or 1040.0). You can nd the (approximate) trade-o point, n0 however you want. If you can do it with algebra, that’s great, or you could sample the two functions at various point or even look closely at a plot of the the functions. Any of these techniques is ne. Turn in your answer as a PDF le named compare.pdf. (5 pts each)

        (a) Consider the two functions, f1(n) and g1(n).

f1(n)    =    0:5    n2    10    n
g1(n)    =    3    n


(b) Consider the two functions, f2(n) and g2(n).

f2(n)
=
3n log2 n
g2(n)
=
2n1:5

    2. Asymptotic Bound Proofs

Write up a short proof of each of the following asymptotic bounds. Submit your argument as a le, bounds.pdf (10 pts each)

        (a) 2  n2 + 5n log2 n 2 O(n2)

        (b) If f(n) 2 !(g(n)) then f(n) 2  (g(n))


2




    3. Recursion Tree Problems

For each of the following recurrence relations, give a short formula for the cost of the root node, the total cost of the leaves, and the total cost of all the internal nodes (so, you’ll be giving three formulas for each). Be sure to label your formulas so we can tell what you’re showing us. For the internal nodes, you can leave your answer in summation notation; you don’t have to simplify it. Submit your answer as a le named tree.pdf. (10 pts each)

p

(a) T (n) = 4T (n=3) + O(n  n)

        (b) T (n) = 3T (n=3) + O(n)

    4. Master Method

Use the master method to give a solution to each of the following recurrences. Brie y explain your application of the master method (e.g., which case applies to each problem). Write up your answers in a le called master.pdf (5 pts each)

p

(a) T (n) = 2T (n=4) +    (  n)

    (b) T (n) = 9T (n=3) +  (n)
    (c) T (n) = 5T (n=2) +  (n2)
p

(d) T (n) = 2T (n=2) +    (n  n)

    5. Best-case running time

We normally don’t care too much about the best-case running time for an algorithm; it’s just not that useful or informative about the algorithm’s performance. Consider the selection sort algorithm we looked at on the rst day of class (you’ll nd it on the slides). Determine what the best-case running time of this algorithm and describe it asymptotically.

Now, describe a simple modi cation of this selection sort algorithm so its best-case running time becomes (n) for an input sequence of length n. This shows why best-case running time isn’t that useful. It’s usually easy to get an algorithm to handle some special cases quickly, even if this doesn’t change its worst-case or average-case behavior.

Submit your answer as a  le, best.pdf. (5 points)

    6. Common Substring Counting

Consider the two strings below. The rst string has 28 substrings, all of them di erent. For example \abcd" and \cdef" are both four-character substrings of the rst string. The second string has 36 substrings, but some of them are duplicates.

abcdefg cdefabab

Your job is to develop a program (named common.c, common.cpp, common.java or common.py) that counts the number of di erent strings that are substrings of both of two input strings. For example, \f", \ab" and \cde" are all substrings of the two strings above. The \ab" substring occurs twice in the second string, but you should only count it once.

Your program will read its input from standard input, one string per line. Input strings will consist of lower-case letters. As output, your program should simply print the number of di erent strings that are substrings of the two input strings. (15 pts for a working program)




3




Sample Input

abcdefg

cdefabab

Sample Output

13

Test Inputs

On the course homepage, I’ve provided four test inputs for this program. These can help you to make sure your program is doing the right thing. Here’s what I got when I ran my solution on these inputs (implemented in C++):

eos$ ./compare < input_1.txt

13

eos$ ./compare < input_2.txt

20

eos$ ./compare < input_3.txt

23

eos$ ./compare < input_4.txt

556

You can see here, I’m redirecting input from a le, so, even though my program thinks its reading from the user, the shell can trick it into reading from a le whithout any change to the program.

If you’ve implemented your solution in Java or Python, you’ll need to run it di erently. Try something more like:

# for java

eos$ java compare < input_1.txt

13

# for python

eos$ python compare.py < input_1.txt

13

Analysis

In addition to your program, you are to write up an asymptotic analysis of its worst case running time. This will probably take you half a page. You should include enough detail to permit someone else to understand your approach even if they couldn’t see a copy of your source code. Be sure to explain how you measure the input size. You have two input strings, so you’ll probably want to use two di erent variables to describe the input size, one for each string. Describe how the running time of each stage of your algorithm and your data structures are dependent on input size. Combine all of the contributions to running time into a single formula and simplify it using big-O notation. Your argument can be informal, you’re not expected to try to capture the exact run-time of your algorithm (like we did in class) or prove your big-O bound (like you’re doing on Problem 2 above). Submit your analysis in a PDF le called common.pdf. (10 pts for the analysis)


4