Computer Science Exam Help

Computer Science Exam Help
Programming Exam Help
For any Programming Exam related queries,Call us at : - +1 678 648 4277
You can mail us at : - support@programmingexamhelp.com or
reach us at : - https://www.programmingexamhelp.com

Problems
Both exercises and problems should be solved, but only the problems should be turned in.
Exercises are intended to help you master the course material. Even though you should not
turn in the exercise solutions, you are responsible for material covered by the exercises.
Mark the top of the first page of your solution with your name, the course number, the
problem number, your recitation section, the date, and the names of any students with whom
you collaborated.The homework template (LATEX) is available on the course website.
You will often be called upon to “give an algorithm” to solve a certain problem.Your write-up
should take the form of a short essay. A topic paragraph should summarize the problem you are
solving and what your results are.The body of the essay should provide the following:
1. A description of the algorithm in English and, if helpful, pseudo-code.
2. At least one worked example or diagram to show more precisely how your algorithm
works.
3. A proof (or indication) of the correctness of the algorithm.
4. An analysis of the running time of the algorithm. Remember, your goal is to communicate.
Full credit will be given only to correct solutions that are described clearly. Convoluted and
opaque descriptions will receive lower marks.

Exercise 1-1. Do Exercise 2.3-3 in CLRS.
Variations of Median
Select Recall from lecture the worst case linear time algorithm for selecting the ith smallest
element from a set S of n distinct elements. In class we organized the elements into groups
of 5, in this problem we will analyze the behavior of the algorithm by picking different sizes
for the groups.

(b)What happens if we choose the group size to be 4? Since the median of 4 elements is
somewhat ambiguous, assume that we always choose the *lower median* from every group,
that is, the 2nd smallest element. Again, write down the recurrence relation forT(n), solve it
using the substitution method and prove its correctness using induction.Again stating your
assumptions.
Land for sale in one dimension
(a) In 1-D land, everybody lives in a one dimensional region.You own a stretch of land in the
shape of a circle, and would like to sell it to n customers. Each of these customers is interested
in a contiguous segment on the perimeter (ie. an arc), which can be represented in polar
coordinates. For example, customer i wants to have θi1 to θi2 on the circle. Unfortunately,
many of these arcs overlap, and you cannot sell a portion of a customer’s request. Devise an
algorithm to maximize the number of customers’ requests you can fulfill.
(a) If we choose the group sizes to be 7, and picking the median element from each group of 7,
repeat the analysis done in CLRS and write down the recurrence relation forT(n). Solve the
recurrence using the substitution method and prove its correctness using induction, in
particular, state the assumptions you make about the base cases (for example, for groups of
5, we assumedT(n) ≤ cn for n ≤ 140).

(b)You are a firm believer in the free market and competition. So, in addition to each customer
giving you an arc that they would like, he also offers a price. However, you still cannot sell a
portion of any request. Devise an algorithm to maximize your profit.
(c)You have acquired a new area in the shape of a cross. Once again, there are n customers
wanting to pay a price for a contiguous 1-D region on the cross (which could be a line segment,
a “T” segment, or a “cross” segment). Like before, these requested regions overlap and can
only be sold in its entirety. Devise an algorithm to maximize your profit.

Solutions
Variations of Median Select Recall from lecture the worst case linear time algorithm for
selecting the ith smallest element from a set S of n distinct elements. In class we organized the
elements into groups of 5, in this problem we wil analyze the behavior of the algorithm by
picking different sizes for the groups.
(a) If we choose the group sizes to be 7, and picking the median element from each group of 7,
repeat the analysis done in CLRS and write down the recurrence relation forT(n). Solve the
recurrence using the substitution method and prove its correctness using induction, in
particular, state the assumptions you make about the base cases (for example, for groups of 5,
we assumedT(n) ≤ cn for n ≤ 140).
Solution:
We divide the n elements of the input array into ⌊n/7⌋ groups of 7 elements each and at most
one other group containing the remaining elements.At least half of the medians for these
groups are greater than or equal to the medianof-medians x.With the exception of the one
group that contains fewer than 7 elements and the group containing x, each of the ⌈n/7⌉ groups
has at least 4 elements greater than x.
So, the number of elements greater than x is at least

This means that the number of elements less than x is at most:
By symmetry, the number of elements exceeding x is at most 5n 7 + 8.The cost of dividing the
n elements into groups, finding the median of each, and partitioning the input array is O(n)
The cost of recursively calling SELECT to compute the median of ⌈n/7⌉ medians isT( n 7 ).The
cost of running SELECT recursively on the elements below or above the pivot isT( 5n 7 + 8).
Hence, we obtain the following recurrence relation:
In the derivation below it will become clear that it is sufficient to have the cutoff be any
integer strictly greater than 63, but it is mathematically convenient to make it 126.To solve
the recurrence using the substitution method first assume that the running time is linear.
Then we show thatT(n) ≤ cn for sufficiently large c and all n ≥ 126:

The latter is at most cn if −cn 7 + 9c + an ≤ 0. Since n ≥ 126, this is equivalent to:
Base Case

So, choosing c ≥ a + b 63 will satisfy the base case. Hence, for c ≥ 14a + b and n ≥ 126:T(n) ≤ cn,
satisfying both the base and inductive cases.
(b)What happens if we choose the group size to be 4? Since the median of 4 elements is
somewhat ambiguous, assume that we always choose the *lower median* from every group,
that is, the 2nd smallest element. Again, write down the recurrence relation forT(n), solve it
using the substitution method and prove its correctness using induction.Again stating your
assumptions.
Solution:With groups of size 4, at least half of the medians are greater than or equal to the
median-of-medians x. With the exception of the one group that contains fewer than 4
elements and the group containing x, each of the ⌈n/4⌉ groups has at least 3 elements greater
than x.The number of elements greater than x is at least
This means that the number of elements less than x is at most:

Each of the ⌈n/4⌉ groups except for the group containing x and the residual group, has at least 2
elements less than x. So, the number of elements less than x is at least
This means that the number of elements greater than x is at most:
In the worst case scenario, the algorithm will keep recursing on the bigger partition. So, the
recurrence is as follows:
The solution to this recurrence is no longer linear.We will prove that it is O(n lg(n)) by the
substitution method.We show that for n ≥ 512 and sufficiently large c,T(n) ≤ nc lg(n):

The latter term is at most cn lg(n) if c(5 lg n − n 4 − 1) + an ≤ 0. Making use of the fact that for n ≥
512, n + 4 − 20 lg n > 0, we can rewrite the inequality as follows

Also, note that for n ≥ 512, n + 8 − 40 lg n > 0. Hence, now if we choose c ≥ 8a, then
Base Case
So, choosing c ≥ a + b 256 will satisfy the base case.This proves that for n ≥ 512 and c ≥ 8a+b,
T(n) ≤ cn lg(n) satisfying both the base and inductive cases.

Problem 1-2. Land for sale in one dimension (a) In 1-D land, everybody lives in a one
dimensional region.You own a stretch of land in the shape of a circle, and would like to sell it to
n customers. Each of these customers
Problem Set 1 Solutions
is interested in a contiguous segment on the perimeter (ie. an arc), which can be represented in
polar coordinates. For example, customer i wants to have θi1 to θi2 on the circle. Unfortunately,
many of these arcs overlap, and you cannot sell a portion of a customer’s request. Devise an
algorithm to maximize the number of customers’ requests you can fulfill.
Solution:The challenge of the circular-arc scheduling problem lies in being able to “linearize” the
input and exploit the unweighted greedy activity scheduling problem covered in lecture to obtain
an optimal solution.
To “linearize” the input we need to determine an appropriate breakpoint at which to cut the circle
without affecting the optimality of the solution to the modified problem.Choosing an arbitrary
cutpoint on the circle may invalidate the scheduling of optimal segments crossing that boundary.
Now, suppose we are provided with extra information – we are given αk2, the endpoint of an
activity in an optimal schedule A⋆ = (α11, α12). . .(αk1, αk2).Then we can conclude that any
segments crossing αk2 are not part of an optimal solution, and we can safely choose αk2 as the
initial condition in our linear scheduling algorithm, discarding any overlapping segments from the
input.We don’t actually need αk2 if we run the linear scheduling algorithm for every possible
choice of segment endpoint. One of the instantiations of the linear scheduling problem is
guaranteed to have αk2 as an intial condition, yielding an optimal solution.

1. Algorithm DescriptionWe provide an O(n 2 ) algorithm for this problem, that uses the
greedy unweighted activity scheduling algorithm presented in class as a subroutine.
Assume that the input consists of n ordered pairs representing the start and end points of the
arcs in polar coordinates: {(θi1, θi2)}, for i = 1 . . . n. Our algorithm first sorts the list of ordered
pairs by the second coordinate.Then going through each of the sorted list of segments, we run
the greedy unweighted scheduling algorithm as a subroutine on segments from the current
segment end time onwards.The input to the linearized greedy scheduling algorithm is already
sorted, and as part of its execution the subroutine discards segments that overlap the starting
point from which the algorithm is executed.
In order to be able to recover an optimal scheduling, we keep track of the set of segments used
for each of the executions of the linearized scheduling subroutine. At termination we return the
set with the maximum number of segments.
Remarks Note that an O(n lg n) solution has been identified in recent literature, but an O(n 2 )
algorithm is sufficient to receive full credit.
2. Worked Out Example
Correctness Proof The circular scheduling problem can be reduced to the linearized scheduling
problem given that the cut point on the circle corresponds to the end time for one of the
segments in an optimal solution. Since we run the linearized algorithm for all possible cutpoint
choices, one of these will have the appropriate initial condition. So, the correctness proof of the
original algorithm is then reduced to the correctness proof for the greedy linear scheduling
algorithm described in class.

4. Running time analysis The algorithm runs in O(n 2 ) time.We only need to sort the set of
ordered pairs by the endpoint of each segment once, taking O(n lg n) time.Then for each for
the n segment endpoints we discard segments overlapping with the start point and execute
the greedy selection subroutine in linear time.This contributes O(n 2 ) time. Finding the
maximum among the n solutions runs in linear time. So, the overall time complexity is O(n 2
)
(b)You are a firm believer in the free market and competition. So, in addition to each customer
giving you an arc that they would like, he also offers a price. However, you still cannot sell a
portion of any request. Devise an algorithm to maximise your profit.
Solution:
1. Algorithm Description Here the algorithm uses the weighted activity scheduling dynamic
program presented in class as a subroutine in a manner analogous to the previous
problem.The overall complexity for the algorithm is O(n 2 ).The algorithm first sorts all of
the angles corresponding to the start and end points of the segments in ascending order.
Then it iterates through each possible segment end point in ascending order, removes
segments overlapping with that boundary, and executes the linear weighted scheduling
dynamic program on the previously sorted input from the boundary onwards
.
In addition to the optimal cumulative profit for each of the subroutine executions,
segment choice backpointers are kept allowing for the recovery of an optimal solution

2. Worked Out Example
3. Correctness Similarly to 2b, the problem of finding the set of requests to fulfill to obtain
maximum profit for circular arc segment inputs can be reduced to the problem of finding the
‘linearized ”schedule” with the maximum cumulative weight, using the dynamic programming
algorithm presented in lecture.Again, this is due to the fact that one of the executions of the
subroutine will have for an initial starting point, the endpoint that corresponds to the endpoint for
one of the segments in an optimal solution. So, we simply refer to the correctness proof for the
linearized weighted dynamic programming algorithm.
4. Running time analysis Sorting all of the endpoints takes O(n lg n) time.We only need to do this
once. For each of the n boundaries, we remove the overlapping segments in O(n) time and run
the dynamic program on the 2n angles in linear time.This requiresO(n 2 ) time. Finding the
solution with the maximum profit among the various runs of the subroutine takes linear time. the
overall time is O(n 2 ).
Solution:
We will present a Θ (N lg N) algorithm.3
Solution: We will present a Θ (N lg N) algorithm.3 Description of Algorithm First, we look
through the requests, and separate them into five categories:
• requests that contain the center of the cross
• requests that contain only the northern arm of the cross
• requests that contain only the eastern arm of the cross
• requests that contain only the southern arm of the cross
• requests that contain only the western arm of the cross

Then, we do dynamic programming on each of the arms to calculate the maximum profit we can
make by selling tips of the arms.To do this, we sort all of the starting indices si and ending
indices ei of the requests on an arm, and iterate through them starting at the end of the arm. If
we let p[j] represent the maximum profit possible by selling the end of the arm up to index j, and
if we let ci represent the price that customer i is willing to pay, then we have the recurrence
relation:
Using this recurrence relation, we can compute the values of p[j] for each arm, starting with a
base case of p[0] = 0.We will store the values of p[j] for each arm so that we can use them
later.
Then, we iterate through the requests that contain the center of the cross. For each request,
we consider accomodating it and selling the remaining arms for the best profit possible. Out of
all these requests, we find the maximum profit possible, and report it as our answer. Note
however, that we should also consider an empty request so that we will be able to consider
selling only the arms of the cross.
Illustrative Example

Consider the situation above, in which we have a cross that has arms of length 3.We have four
customers, who are willing to pay different prices for different sections of land.
When we separate the requests into categories, we see that request 4 contains the center of
the cross, and all the other requests contain only the western arm. When we run dynamic
programming on each of the arms of the cross, and we get the following results:

Now, when we look at customer 4’s request, we see that the remaining land can be sold for $2,
so we can make $11 if we satisfy customer 4’s request.When we look at the empty request (in
which we don’t sell the center of the cross), we see that the 4 arms of the cross can be sold for
$9.Thus, we see that the optimal solution is to sell to customers 1 and 4 for $11.
Proof of Correctness
Lemma 1 For each arm, the dynamic programming algorithm will correctly compute the values
of p[j] described above. PROOF.To do this, we will prove a loop invariant: p[j] is the maximum
profit possible by selling the tip of the arm up to index j.
During initialization, p[0] = 0, so the loop invariant holds. For each iteration of the dynamic
programming algorithm, we use the recurrence re9 lation
If j is a starting index, then we can’t have enough land to accomodate the request yet (since we
are only selling the land from the end of the arm to j at this time).Thus, we must have p[j] = p[j
− 1].

If j is an ending index, then we have enough land to accomodate the request. If we fulfill the
request, we will make ci profit, and we only have the land from the end of the arm to si −1
remaining, since the land from si to ei was sold.Thus, our maximum profit if we sell to customer
i is p[si − 1] + ci . On the other hand, if we don’t fulfill the request, our maximum profit would be
p[j − 1], as in the previous case.Thus, the maximum profit p[j] we can make is max (p[j − 1], p[si
− 1] + ci). In either case, we see that the loop invariant is maintained, so the computed values of
p[j] are correct.
Now, we will show that the algorithm will compute the optimal profit.We will consider two
cases:
Case 1:The optimal solution satisfies a request that contains the center of the cross. In this
case, our algorithm will have considered the same request (that contained the center of the
cross). Since the algorithm knows how to sell the remaining pieces of land in the optimal way,
the algorithm will return an optimal solution.
Case 2:The optimal solution does not satisfy a request that contains the center of the cross.
In this case, our algorithm will have considered the empty request and considered selling each
arm of the cross in the optimal way.Thus, our algorithm will return an optimal solution. In all
cases, we see that our algorithm will return an optimal solution, so we are done.

RunningTime Analysis
First, we see that separating the requests into five categories takes Θ (N) time. Next, we see that
running dynamic programming on each arm takes Θ (N lg N) time, because we have to sort the
starting and ending indices for each request (which takes Θ (N lg N) time) and applying the
recurrence relation for each index (which takes Θ (N) time). Finally, we see that trying out all the
requests that contain the center of the cross (and the empty request) takes Θ (N) time.Thus, we
see that the total running time is Θ (N lg N)

(d) Finish running the scheduling algorithm described in class. s = {3, 5, 7, 8, 10, 11}, |s| = 6. (e) Now
find a solution with alternative start points. Start with arc 1. s1 = {1}: (f) s1 = {1, 4, 7, 8, 10}, |s1| = 5.
|s| > |s1|, so we need to keep iterating through arcs in R.
Figure 1:Worked out Example for Problem 2a

(b) Iterate through n = 3
arcs. Start with the
highlighted arc.That arc
is now fixed.We want to
use dynamic
programming to find the
maximum weight
solution on the circle
segment from point 2 to
point 0
(counterclockwise).
w(s1) = maximum
weight solution on
segment +w(1), where w
is a weight function.
(c) Ignore this point
because it is part of
an arc that
intersects the fixed
arc i.
d) Add the
green edge to
solution s1.The
algorithm has
finished
considering all
points on the
circle segment.
w(s1) = 2.

(e) Fix the next arc in solution
s2. All points on the circle
segment are part of arcs that
intersect the blue arc, so we
ignore all of them.We’re done
with this solution: w(s2) = 3.
f)We end up with w(s3) = 2. (g) Return s2 because it has
the largest weight of all 3
solutions.

Computer Science Exam Help

More Related Content

Similar to Computer Science Exam Help (20)

More from Programming Exam Help (20)

Recently uploaded (20)

Computer Science Exam Help