- Using the current IPv4 Internet Protocol, the IP address of a
computer connected to the Internet consists of 4 bytes (32 bits).
Realizing that we may soon run out of IP addresses to give to all
possible computers out there, the IPv6 protocol allows for 16-byte
IP addresses.
-
Under IPv4, how many computers with IP addresses could
co-exist on the Internet at any point in time?
-
Potentially, under IPv6, how many computers with IP
addresses could co-exist on the Internet at any point in time?
- Using the IP protocol, information communicated over the
Internet is broken down into packets, each of which carrying 1.5KB
of data. How many packets would it take to send each of the
following on the Internet:
- A 140-character tweet on Twitter. Recall that a character
can be represented in a single byte.
- A 1,600x1,200 bitmap picture with 24-bit color depth. Recall
that each pixel in a bitmap picture is represented by three
bytes for RGB (hence the 24-bit color depth).
- A 3GB movie.
- A friend of yours who has not taken MA/CS-109 is wondering how
is it that signals carried by fiber-optic cables, Wi-Fi, wires,
satellites, and cell phones are all part of the same Internet. How
do you explain to him/her that one does not need to re-engineer the
Internet any time a new way to carry signals is invented?
- A friend of yours who has not taken MA/CS-109 is wondering how
is it that a network designed to carry text is now being used for
music, telephony, photos, video, etc. After all, the good old
telegraph wires could not be used for telephone communication, and
telephone wires did not help when TV was invented. How do you
explain to him/her that you do not need to re-engineer the Internet
whenever a new application is invented?
- Researchers at Georgia Tech, at UC Berkeley, and at Boston
University (among other places) are looking into making the Internet
available at very remote/under-developed parts of the world. The
main challenge in "connecting" such rural areas is that it is not
economically feasible to run wires (coaxial or fiber) or to use
satellite communication to remote villages with very sparse
population. To solve this problem, these researchers are using buses
(yes buses) to move the IP packets between a remote rural area and
larger neighboring towns where Internet is available. The bus is
literally acting as a wire between the router in the remote village
and the router in the neighboring town! Upon hearing about this, a
friend of yours who did not take MA/CS-109 wondered about the
usefulness of such an approach since in his/her mind, one must
re-engineer the whole Internet to allow for buses to ferry packets
around!
How do you explain to your friend that you do not need to
re-engineer the Internet whenever a new method of moving packets
around is invented?
- Consider the following switching network made up of a 16-to-1
multiplexer connected to a 1-to-16 de-multiplexer.
-
What bits would you provide as the sender
address and what bits would you provide as the receiver address to
enable Lady W to speak to Lady X?
Sender Address: _______________
Receiver Address:
_______________
-
With lady W and lady X chatting away, when lady
Y tried to speak to lady Z, she got the "all lines are busy" signal.
Why did that happen?
-
With phone networks (which are designed using
switches as illustrated above), especially around busy periods
(e.g., Mother's day), we sometimes run into the same "all lines are
busy" problem (even though both ends of the call are not themselves
busy).The same is not true with the Internet in the sense that even
when loads are high, calls on the Internet are still able to go
through (even if the quality of the connection is not that great).
What design feature of the Internet (as opposed to the phone
network) makes that problem (of "all lines are busy") disappear?
-
As discussed in class and explained in
the notes, IP addresses of computers on the Internet consist of
4 bytes. If all computers in the computer science department
research lab start with the same sequence of bits, namely:
1000000011000101000010. How many different computers could the
CS department have on the Internet?
-
Label each statement as either "always true",
"always false", or "it depends" (i.e., it could go either way
depending on other details).
- The number of steps of Dijkstra's
shortest path algorithm when applied to an arbitrary graph is
linear in the number of nodes in the graph.
- For graph nodes A, B, and C: the cost
of the shortest path from A to B is less than or equal to the
sum of the costs of the shortest paths from A to C and from C to
B.
- A graph with N nodes can always be
colored using N or less colors.
- If the rate with which cars arrive at a
toll booth doubles, then the length of the queue of cars waiting
to go through the tolls will also double.
- If the utilization of a service is 80%
and the rate with which requests are made is 40 requests per
hour, then the capacity of the service is 50.
- The Internet protocol guarantees the
delivery of every packet it handles.
- The Internet protocol deals with
packets differently depending on whether these packets carry
text, audio, video, etc.
- It is easy to check whether a proposed
solution to an NP problem is a correct solution.
- According to Google’s PageRank
algorithm, a web page’s rank is based solely on the total number
of links from other pages to that page.
- Earlier this month (April 2010), The Library of Congress (LoC)
decided that it will archive all the tweets that were ever tweeted
(and will continue to do so for the foreseeable future) – note: A
tweet is a message consisting of at most 140 characters.
Upon hearing this development, a friend of
yours who is quite concerned about government spending commented
that “this is crazy; the LoC will have to spend a ton of money to
store all these tweets.” Follow the steps below to develop a
quantitative argument about this issue:
-
Assuming that an average Twitter user
produces 10 tweets per day, how much storage will be needed to
preserve the tweets of one account for one year?
-
Assuming that an individual will tweet for
an average of 60 years, how much storage will be needed to
preserve the lifetime tweets of all 300 million US citizens
(assuming that they all have Twitter accounts)?
-
Today, it costs around $1 to store 10GB.
How much will it cost the LoC to preserve the tweets of all US
citizens?
-
Looking to the future, the LoC will have to
preserve the tweets of all future accounts. Considering that
there are about 15 births per 1,000 people in the US per year,
the number of accounts kept by the LoC will grow by about 1.5%
per year. Storage costs, on the other hand are decreasing at a
fixed rate per year (about 40% per year for hard disks). Do
these rates bode well for the LoC or should they worry about the
escalating costs of maintaining all these tweets?
- Would your thinking change if you consider the entire world
population (as opposed to only US citizens)? Note: In fact, the
LoC is archiving all Twitter account tweets (not just those by
US citizens).
- If government spending is
not a concern of yours (or is a non-issue considering your
quantitative assessment above), are there other concerns that
you should be considered?
- Answer the following questions about the graph abstraction of
the DC Metro shown below, where the red dots represent a subset of
the stations and the numbers shown represent the number of metro
stops between two pairs of stations.
-
How many nodes are there in this graph?
-
How many edges are there in this graph?
-
Are the edges in this graph directed or not?
-
What is the maximum degree of any node in
this graph? (Hint: The degree of a node is the number of edges
from/to that node)
-
What is the diameter of this graph? (Hint:
The diameter of a graph is the length of the longest shortest
path between any two nodes)
- Prove that the shortest path between two nodes in a graph with N
nodes cannot be more than N-1 in length.
- Consider applying Dijkstra's shortest-path algorithm to the
following graph starting with node A.
Recall that Dijkstra's algorithm works iteratively. In each
iteration, it augments by one the set of nodes for which a shortest
path is already known. For example, in the first iteration (for the
above graph), the algorithm will add node "C" since the path to C
has the minimum cost (of 3) among all nodes it is yet to reach (B,
C, D, and E). In the next iteration, the algorithm will add "B"
since the path to B (through C) has the minimum cost (of 4) among
all nodes it is yet to reach (B, D, and E). This process goes on
until the shortest path to all reachable nodes is found.
In what order will Dijkstra's algorithm discover the shortest
paths between A and the various destination nodes in the graph?
- You are planning a car trip from city A to city B, and you are
interested in stopping along the way to visit a friend in city C.
You are told that the shortest path between A and B is 135 miles and
that the shortest path between A and C is 80 miles. What is your
estimate of the shortest path between C and B? Your answer should be
in the form: “The shortest path between C and B is (at least, at
most, or exactly) … miles long”. In one or two sentences explain
your logic.
- Your friend who is a CS major wrote a program to compute the
shortest path between a given node in a graph and all other nodes in
that graph using Disjkstra’s algorithm. When she ran her program on
a graph with 100 nodes, the program took 1 second to run. Can you
guess how long her program will take if it is run on a 1,000-node
graph? Why?
- Consider three stops A, B, and C on the T subway map. The
shortest path between A and B takes 15 minutes and the shortest path
between A and C takes 25 minutes. For each of the statements below,
state which one is correct (i.e., you can prove it), incorrect
(i.e., you can disprove it), or neither.
- The shortest path between B and C is less than 10 minutes
- The shortest path between B and C is exactly 10 minutes
- The shortest path between B and C is more than 10 minutes
- The shortest path between B and C is at most 10 minutes
- The shortest path between B and C is at least 10 minutes
- Consider three stops A, B, and C on the T subway map. The
shortest path between A and B takes 15 minutes and the shortest path
between B and C takes 25 minutes. For each of the statements below,
state which one is correct (i.e., you can prove it), incorrect
(i.e., you can disprove it), or neither.
- The shortest path between A and C is less than 40 minutes
- The shortest path between A and C is exactly 40 minutes
- The shortest path between A and C is more than 40 minutes
- The shortest path between A and C is at most 40 minutes
- The shortest path between A and C is at least 40 minutes
- The Federal Communications Commission (FCC) prevents
interference between radio stations by assigning appropriate
frequencies to each station. Two stations cannot use the same
channel when they are within 150 miles of each other. Use graph
coloring to find out how many different frequencies are needed for
the six stations located at the distances shown in the table below
by following the steps below.
-
Model the conflict relationships between the
above stations with a graph. What do the nodes of the graph
represent? What does an edge in that graph represent?
-
Identify the minimum number of colors needed to
color the nodes of the graph you obtained in step (a) such that
no two adjacent nodes are assigned the same color.
-
Use your answer in step (b) to determine the
minimum number of frequencies needed for the six stations.
- Consider the following data collected using Traceroute
experiments between 3,600 pairs of computers. Answer the following
questions:
-
What proportion of the sample had between 11 and
14 hops, inclusive?
-
Can you give a margin of error and a confidence
interval for your answer in part a?
-
What is the chance (probability) that the
confidence interval you provided in part b. will not catch the
"real" proportion of pairs of computers with a distance between
11 and 14 hops?
-
You are starting a new company "AcmeCorp.com" and
want to make sure that your competition will not "hijack" your web
presence by creating web sites for misspelled versions of "AcmeCorp".
Since web site names are case-insensitive, you have to account for a
typo in each position of the word "acmecorp". How many web site
names would you have to register with the Internet domain name
authority to protect yourself from a single typo?
-
Using contradiction, make an argument for why the
same node X cannot appear twice on the shortest path between any
pair of nodes, say A and B. What assumption is necessary for your
argument to hold?
-
Using induction, prove the following statement: The
shortest path between any two nodes in a connected graph with n
nodes cannot contain more than n-1 edges. What assumption is
necessary for your argument to hold?
-
You have been hired as an intern at a
consulting firm who is retained to look at where would be the best
places to install Internet kiosks at Disney's Magic Kingdom (or you
favorite park). Each one of the k (e.g., k=3) kiosks to be installed
needs to be right next to the entrance to a landmark (e.g., a ride
or a food court, etc.) The prevailing wisdom in the firm is that the
best place for the kiosks would be next to the landmarks that have
the most neighboring landmarks, since these are the most “central”.
Two landmarks are “neighbors” if there is a way to walk directly
from one to the other. However, having taken MA/CS-109, your
intuition is that it is far better to identify the landmarks that
are likely to be visited the most, whether or not they are central
by virtue of having many other landmarks next to them. You were
given the set of landmarks (and the walkways between them) and the
percentage of people going from one landmark to every one of its
neighbors.
Answer the following questions:
-
Show how you could model the flow of people
between park landmarks as a graph. In particular, specify what
constitutes the nodes of the graph, the edges of the graph,
whether the edges are directed or not, and the labels on the
edges.
-
In your own words, explain how you would make
the case for a different approach -- in particular an approach
that mirrors how Google ranks the various web pages to measure
their popularity.
-
As part of the recovery act, the small town of Wanderland
is slated to receive $1M to upgrade their three town intersections
to relieve traffic congestion. To decide how much money to allocate
to the upgrade of each intersection, Wanderland collected traffic
data, which they represented as a graph in which intersections are
nodes and directed edges are streets. The data they collected
allowed them to label each edge (street) out of a node
(intersection) with the proportion of cars at the intersection that
would take that edge. The results they obtained are shown below.
The Wanderland board of selectmen, upon looking
at the results noted that since each intersection has the same
number of streets into it, decided to split the $1M equally.
-
Explain why just counting the number of roads
going into an intersection is not a good measure of whether the
intersection may (or may not) be busy.
-
What process may be used to model how cars go
through Wanderland's intersections?
-
Following the process you adopted in part (b)
write down three relationships that would allow you to compute how
relatively busy the intersections are.
-
To explain your ideas to the board of selectmen
you decided to simulate the process you adopted in part (b). So, you
started with 100 cars in each intersection and proceeded to compute
the number of cars in successive one minute intervals (assuming it
takes one minute to travel between any two intersections). Show the
number of cars in the first minute and in the 1001st
minute of the simulation below.
Time |
@ A |
@B |
@C |
0 |
100 |
100 |
100 |
1 |
|
|
|
… |
|
|
|
1000 |
72.687 |
121.586 |
105.727 |
1001 |
|
|
|
-
What criticisms do you expect from the board of
selectmen regarding the process you adopted in part (b)? How would
you answer them?
-
You were hired by a marketing firm and were asked to
review the rates that the firm is charging for three special
displays at an amusement park. One of these displays is located at
the main entrance of the park (location A); the second is located
next to the food court (location B), and the third is located next
to main ride (location C). Studies of how visitors of the park move
from one location to another in a period of 15 minutes suggest that:
-
Of all
people in location A: 20% end up going to location B, 50% end up
going to location C, and 30% remain in location A.
-
Of all
people in location B: 60% end up going to location C, and 40%
end up going to location A.
-
Of all
people in location C: 30% end up going to location A, and 30%
end up going to location A, and 40% remain in location C.
Given this information, you proposed that a sensible
approach to setting the pricing for the special displays is to make
the price proportional to the number of people that are expected to
be at each one of the three locations.
-
Show how
you could model the flow of people as a graph. In particular,
specify what constitutes the nodes of the graph, the edges of the
graph, whether the edges are directed or not, and the labels on the
edges.
-
Draw the
graph corresponding to the above observations.
-
If P(A),
P(B), and P(C) denote the proportion of the park visitors expected
at each one of the three locations (long after the park opens in the
morning). Write three relationships that would allow you to figure
out these proportions.
-
Prove that adding a single edge between two distinct
nodes in a Eulerian graph will result in a new graph that is not
Eulerian. Hint: Recall that for a Eulerian circuit to exist in a
graph, all the nodes of the graph must have an even degree.
-
Consider three stops A, B, and C on the T
subway map. You are told that:
- The shortest path between A and B takes 15
minutes
- The shortest path between B and C takes 25
minutes
Complete the table below by specifying for each
statement whether the statement is correct (i.e., you can prove it),
incorrect (i.e., you can disprove it), or neither.
Statement |
Correct? (Yes/No/Maybe) |
The shortest path
between A and C is less than 40 minutes |
|
The shortest path
between A and C is exactly 40 minutes |
|
The shortest path
between A and C is more than 40 minutes |
|
The shortest path
between A and C is at most 40 minutes |
|
The shortest path
between A and C is at least 40 minutes |
|
-
Facebook is a social networking application
that allows individuals to befriend each other. Twitter is a social
networking application that allows individuals to “follow” the news
(tweets) of one another. One can model each of these applications
with a graph. Complete the table below:
Question |
Facebook |
Twitter |
What do graph nodes
represent? |
|
|
What do graph edges
represent? |
|
|
Are edges directed or
undirected? |
|
|
What would be a good use
of the solution to the shortest-path algorithm on the graph? |
|
|
What would be a good use
of the solution to the PageRank algorithm on the graph? |
|
|
-
Consider the following word game (called Doublet and
proposed by Lewis Carroll in 1879).
You are given two English words of the same length, and you are
asked to come up with a sequence of words of the same length
starting with the first and ending with the last, such that every
word in the sequence is a correct English word (e.g., from the
Webster dictionary), and any two consecutive words in the sequence
differ in exactly one letter. Such a sequence of words is said to be
a “valid sequence”.
For example, if you are given HEAD and TAIL, then
HEAD-HEAL-TEAL-TELL-TALL-TAIL is an example of a valid sequence.
Clearly there could be many valid sequences from HEAD to TAIL. For
example, HEAD-HEAL-DEAL-TEAL-TELL-TALL-TAIL is another one.
You win the game if you can come up with the shortest valid sequence
of words.
It was suggested that one strategy to win this game is to use a
graph to explore all possible valid sequences from any word in the
English dictionary to any other word in the English dictionary.
-
What should the nodes of
the graph represent?
-
What should the edges of
the graph represent?
-
If you are given two
English words, what algorithm would you use on the graph to come up
with the winning word sequence?
-
Vehicular traffic around the BU Bridge
inexplicably grinds to a halt every time there is a red-sox game
or there is an event at the Agganis Arena. The root of the
problem seems to be that the traffic lights are set on an
automatic timer, which in some cases lets too many cars go
through one traffic light at an intersection, resulting in the
blocking of the intersection, which means that even if a second
traffic light at the intersection turns green, vehicles can’t
move. This in turn may result in other blockages at other
intersections, which (not surprisingly) end up contributing to
the initial blockage. To resolve the blockage at one
intersection requires the resolution of the blockage at another;
yet, to resolve the blockage at that second intersection
requires the resolution of the blockage at the first! Situations
like this are called “deadlocks”.
Deadlocks occur when there is a “cycle”
of blockages, and this cycle could be of any length (not just
two as described above) – i.e., blockage at intersection 1
causes blockage at intersection 2 which causes blockage at
intersection 3, …, which causes blockage at intersection m,
which causes blockage at intersection 1!
By programming traffic lights at various
intersections in a city center (such as around BU), one can
determine if it is ever the case that traffic through one
intersection will cause blockage of traffic in another
intersection. For a particular city center with 7 intersections,
and for a particular setting of the traffic light programming,
the following relationships were observed:
. |
1 |
2 |
3 |
4 |
5 |
6 |
7 |
1 |
. |
* |
* |
* |
- |
* |
- |
2 |
- |
. |
* |
- |
* |
- |
- |
3 |
- |
- |
. |
* |
- |
- |
- |
4 |
- |
- |
- |
. |
- |
- |
- |
5 |
- |
- |
- |
- |
. |
- |
* |
6 |
- |
- |
* |
- |
* |
. |
- |
7 |
- |
* |
- |
- |
- |
- |
. |
In the table above a star in entry row i
and column j means that the intersection i could cause blockage
at intersection j.
To visualize these relationships, you
decided to use a graph where vertices represent intersections
and edges represent blockage relationships.
-
Draw the graph. Is the graph
directed or not?
-
Show that traffic in the city center
could potentially be deadlocked by finding a cycle of
blockage dependencies.
-
Identify the set of intersections
whose traffic lights must be reprogrammed to alleviate this
deadlock potential.
Note: You can read more about deadlocks
(a classical problem in computer science) at
http://mcs109.bu.edu/site/?p=deadlock
- Given the 5-node graph shown, answer the following questions:
- Write down the degree of each node in the graph. What is the
average degree?
- What is the probability that a new node (F) will attach
itself to node A under the “Random Attachment” growth model?
- What is the probability that a new node (F) will attach
itself to node A under the “Preferential Attachment” growth
model?
-
Facebook is a social networking application that
allows individuals to befriend each other. Twitter is a social
networking application that allows individuals to follow the news
(tweets) of one another. One can model each of these applications
with a graph.
Answer the following questions for each one of the
above applications:
-
What would constitute a node in the graph?
-
What would constitute an edge in the graph?
-
Are the edges of the graph directed or
undirected?
-
Can you think of a good use of the solution to
the shortest path problem between two nodes in the graph? What
kind of "social networking" question does it solve?
-
Can you think of a good use of the solution to
the PageRank algorithm on the graph? What kind of "social
networking" question does it solve?
-
The “popularity contest” between CNN and Ashton Kutcher on Twitter has focused on who of the two is able to gather
more followers. How would you explain to a friend of yours who has
not taken MA/CS-109 that a simple count of Twitter followers is not
the best way to settle this popularity context?
- Cars arrive at a car wash at an average rate of 10 cars per
hours and it takes 5 minutes to wash each car.
Answer the following questions:
- What is the maximum rate with which cars can go through the
wash?
- What is the utilization of the car wash?
- On average, how many cars do you expect to find at the car
wash? [Recall that the average size of a queue is given by
U/(1-U), where U is the queue utilization].
-
On a typical day, customers arrive to the post office at an average
rate of 8 customers per hour and it takes 5 minutes for the
post-office employee to serve each such customer, on average. Answer
the following questions:
-
What is the “capacity” of the post-office –
i.e., how many customers would it be possible for the post office
employee to serve per hour?
-
What is the utilization of the post office on a
typical day?
-
What is the likelihood that a customer will not
have to wait in line once they arrive to the post office?
-
What is the likelihood that a customer will
have to wait in line for more than one person ahead of them? [Hint:
You can find this out by calculating the probability that there will
be either nobody in line or exactly one other customer in line].
-
On average, how many customers would be waiting
in line on a typical day?
-
On one of those busy days before Christmas, the
rate with which customers arrive to the post office increased by
45%. Repeat parts b, c, d, and e.
-
Customers arrive at the line for a fast food restaurant at an
average rate of 15 customers per
hour and it takes 3 minutes on average to complete the order for
each such customer.
Answer the following questions:
- What is the maximum rate with which the fast food restaurant
can serve its customers? In other words, how many customers per
hour can the restaurant keep up with?
- What is the utilization of the fast food restaurant?
- What is the probability that a customer arriving at the fast
food restaurant will not have to wait in line for service?
- What is the probability that a customer arriving at the fast
food restaurant will find exactly 2 other customers in the
store?
- On average, how many customers would one expect to find in
the restaurant?
- An advertisement campaign resulted in a 20% increase in the
popularity of the fast-food restaurant. Repeat parts b, c, d,
and e.
-
When you mentioned to your parents that you
won't be able to make it home early for the holidays, they were
quite annoyed and suggested that BU was not doing a good job
scheduling the exams. As an MA/CS-109 graduate, you want to explain
to your parents that the problem of scheduling the exams is not as
simple as it sounds. In particular, two courses cannot have their
exams at the same time if any student is enrolled in both. Moreover,
given enrollment data, figuring out the schedule that minimizes the
total number of exam slots is computationally intensive.
To explain this to your parents, you decided to
map the problem to a graph coloring problem whereby the number of
colors used to color the graph represent the number of distinct exam
slots needed for finals. For example, assuming AM and PM slots for
exams per day, Red = Monday AM, Blue = Monday PM, Green = Tuesday
AM, etc.
-
What do the nodes in your graph represent?
-
What do the edges in your graph represent?
-
For a set of 5 classes, what is the maximum
number of exam slots needed? Draw an example graph that
requires that number of slots.
-
To explain the concept, you used the 6
classes that a group of 5 friends are taking in the current semester
as an example. Friend #1 is in courses A, B, and C. Friend #2 is in
A, B, and D. Friend #3 is in B, C, and E. Friend #4 is in B, C, and
F. Friend #5 is in B, D, and F. Find the minimum number of exam
slots in this example (you need to draw and color a graph model of
this example).
You explained to your parents that, in general,
to figure out the minimum number of exam slots, one would need to
check every possible assignment of classes to slots. For example,
to figure out if 3 classes (A, B, and C) can fit in 2 slots (AM and
PM), one would need to check a total of 8 possible schedules for
conflicts, since A can be either AM or PM, and for each one of these
choices, B can be either AM or PM, and for each one of these
choices, C can be either AM or PM. Thus in total one would have to
check 2*2*2=8 possible schedules for conflicts.
-
You want to convince your parents that doing
the above is just too much work. So, you decide to go for more
realistic numbers. How many schedules would have to be checked for
conflicts if the number of classes is 10? How many would have to be
checked if the number of classes is 30? What is the formula for N
classes? What kind of function is that?
-
Now, to impress your parents even more (and
convince them that taking MA/CS-109 was worth delaying coming home
late for the holidays), you decided to tell them that graph coloring
is an example of the "NP" class of problems in Computer Science. In
a few sentences, explain what it means for a problem to be labeled
as such. What other problems can you mention to them as belonging to
the same class?
-
You were hired as a consultant to help
look for ways to improve the operation of a tropical fish farm
in Florida. The farm raises six different types of tropical
fish, each identified by a letter: A, B, C, D, E, and F.
Because of predator-prey relationships, water conditions, and
size, some fish can be kept in the same tank, while others
cannot. The following table shows which fish cannot be together
-- i.e., they have to be shipped in different tanks/containers.
For example, fish of type A cannot be in a tank containing fish
of type B or fish of type C.
Fish Type |
A |
B |
C |
D |
E |
F |
Cannot be with |
B,C |
A,C,E |
A,B,D,E |
C,F |
B,C,F |
D,E |
Every week, the farm arranges for a
shipment of fish to a major pet store chain in the Northeast.
The price of shipping a single container is $500 and the farm
current practice is to ship each type of fish in a separate
container, which implies a cost of $3,000 per week (for 6
containers) or $156,000 per year.
Upon reviewing these facts, you realized
that this problem is not very different from many of the
problems you have encountered in MA/CS-109, in which graph
coloring was used to identify the minimum number of groupings of
vertices so that no two vertices with a conflict relationship
are put in the same group. Thus, you decided to use graph
coloring to figure out a more effective shipping strategy.
Answer the following questions:
-
Draw the conflict graph. What do the vertices
represent? What do the edges of the graph represent?
-
Based on the conflict graph you obtained, what is the
minimum number of tanks needed to ship the fish?
-
How much money would this new shipment strategy save the
farm business per year?
-
A friend of yours working as a
work-study in the chemistry department was asked to come up with
a schedule for their wet labs. Seven courses in the department
require the use of the lab once a week and the department would
like to have the lab open for the least number of days possible.
However, to be considerate, the department also wants to avoid
having a student do lab work for more than one course on the
same day. In the table below a star in entry row i and column j
means that course i and course j have at least one student in
common, so labs for these courses should not be scheduled on the
same day.
. |
1 |
2 |
3 |
4 |
5 |
6 |
7 |
1 |
. |
* |
* |
* |
- |
* |
* |
2 |
* |
. |
* |
- |
- |
- |
* |
3 |
* |
* |
. |
* |
- |
- |
- |
4 |
* |
- |
* |
. |
* |
* |
- |
5 |
- |
- |
- |
* |
. |
* |
- |
6 |
* |
- |
- |
* |
* |
. |
* |
7 |
* |
* |
- |
- |
- |
* |
. |
Knowing that you took MA/CS-109, your
friend asked you for help. After thinking about it for a bit,
you realized that this problem is similar to the problem of
minimizing the number of tables at a wedding party, given a list
of pairs of guests who cannot be seated together. In particular,
you recalled that this problem was solved by modeling it as a
graph coloring problem.
-
Model the lab scheduling problem as
a graph. In particular, identify what constitutes a vertex
in the graph and what constitutes an edge.
-
Show how you could come up with a
schedule by coloring the vertices of the graph.
-
Are you certain that the schedule
you obtained will minimize the number of days when the lab
is open? Would you be able to give the same answer for any
arbitrary graph? Explain why or why not.
-
A directed graph is a graph in which
edges are directional, in the sense that you can traverse them
in one way but not the other. A cycle exists in a directed graph
if one can find a path that goes through a vertex twice. A
directed graph is called acyclic if it has no cycles in it.
Many interesting questions related to
real-world problems could be answered by modeling the real-world
as a directed graph and by finding if cycles exist in such a
graph (e.g., finding if deadlocks may materialize across a
number of intersections, detecting if a set of routers on the
internet may end up sending traffic destined to a particular
target computer in endless loop, detecting if a set of programs
on your PC will wait for one another indefinitely causing your
computer to inexplicably “hang”, …)
- Prove or disprove the following conjectures: "The
maximum length of any path in an acyclic graph with N
vertices is N-1."
- Prove or disprove the following conjecture: "In an
acyclic graph, there must exist at least one vertex without
any outgoing edges."
- To decide whether a graph has a cycle, the following
algorithm was proposed:
CheckForCycles(G)
-
For each vertex in the graph
G, count the number of outgoing edges.
-
Find the vertex v with the
minimum number of outgoing edges.
-
If v has any outgoing edges,
then print "Cycle Detected" and stop.
-
Otherwise, obtain a new
graph G' by removing v from G, along with any edges
connecting any other vertex u to v. Now
CheckForCycles(G').
Explain the logic behind the
above algorithm.
- If each of the steps in the above algorithm take at most
1 second, show that for any graph with N vertices, the above
algorithm will stop in no more than 4*N seconds.
- Try the above algorithm on the following graph:
-
A smaller variant of the Sudoku puzzle is
called Shudoku. In Shudoku, you are given a 4x4 square (see below),
and you are asked to fill every one of the 16 cells with a number so
that
-
all numbers in the same row must be
distinct
-
all numbers in the same column must be
distinct
-
all numbers in any 2x2 quadrants must be
distinct
The trick is to find the minimum set of
numbers that make this possible.
The following is an example of a Shudoku
puzzle (in which some of the numbers are already assigned to cells).
To solve the above Shudoku puzzle
(or any other), it was suggested that the puzzle be modeled using
a conflict graph, where nodes are cells and edges (i.e., conflicts)
exist between any two nodes (i.e., cells) that cannot be assigned
the same number.
Answer the following questions:
-
How many total nodes are
there?
-
Are the edges of the graph
directed or not?
-
List all the nodes that will be
adjacent to the top-left cell.
-
What is the degree of each node
in the graph? In other words, how many edges will each node have?
-
In one or two sentences,
explain why a solution to the graph coloring problem for the graph
obtained in step (a) should be a correct solution for the Shudoku
puzzle.
-
Consider a variation of the
Shudoku puzzle, in which in addition to the rules of no repeated
numbers in the same row, column, or quadrant, we add the rule: all
numbers in the 2x2 center square must be distinct. Can you still use
graph coloring to solve this new version? Justify your answer (e.g.,
by showing what modification would be needed to the steps above, or
by explaining why graph coloring cannot be used to solve the new
puzzle).
-
The Audubon society is pursuing a wildlife
preservation project and needs to deploy a team of 100 volunteers to
sample the number of birds from some species. The birds are known to
congregate in three primary locations: A, B, and C.
In preparation for this project, a team of
scientists tagged a small number of birds and determined that the
flying patterns of the tagged birds are as follows:
- 40% of birds in location A were observed
one hour later in location B
- 40% of birds in location A were observed
one hour later in location C
- 50% of birds in location B were observed
one hour later in location A
- 50% of birds in location B were observed
one hour later in location C
- 100% of birds in location C were observed
one hour later in location A
Answer the following questions:
-
One proposal to the Audubon society was to dispatch an equal number
of volunteers to all three locations. Explain why such an approach
may result in a biased sampling of the bird population. How should
the Audubon society dispatch volunteers to mitigate this bias?
[Hint: To ensure that the sampling process is as close as
possible to simple random sampling, it is desirable that the number
of volunteers dispatched to a location be proportional to the number
of birds expected at that location.]
-
Represent the flying patterns of tagged birds as a graph.
What do the nodes in your graph represent? What do the edges
represent? Are the edges directed or not?
-
What process might be used to model the
movement of birds between the three locations? In one or two
sentences explain in plain English what that process means.
-
Explain how the above process can be used to
help the Audubon society decide on the number of volunteers to
deploy to each location.
-
Write down a set of equations that might be
used by the Audubon society to determine the proportion of
volunteers to dispatch to each location.
-
An intern was asked to solve the set of
equations in part (e) and to propose how to dispatch 100 volunteers
accordingly. Her answer was that 50 volunteers should be dispatched
to A, 20 to B, and 30 to C. Do you agree with her conclusion?
Explain why or why not. [Hint: You may solve the equations and use
the solution to figure out the number of volunteers accordingly, or
you may use simulation of the process proposed in part (c).]
-
Coming Soon!
Azer Bestavros (2010-12-16) |