Design and Analysis of Algorithms Assignment Help

For any Assignment related queries, call us at : - +1 678 648 4277
visit : - https://www.programminghomeworkhelp.com/, or
Email : - support@programminghomeworkhelp.com

Problem 1. True or False.
Circle T or F for each of the following statements to indicate whether the statement is true or false,
respectively, and briefly explain why (one or two sentences). Your justification is worth more points
than your true-or-false designation. Careful: Some problems are straightforward, but some are
tricky!
(a) T F Suppose that every operation on a data structure runs in O(1) amortized time. Then
the running time for performing a sequence of n operations on an initially empty data structure is
O(n) in the worst case.
Solution: True:
(b) T F Suppose that a Las Vegas algorithm has expected running time Θ(n) on inputs of size n.
Then there may still be an input on which it always runs in time Ω(n lg n).
Solution: False.
(c) T F If there is a randomized algorithm that solves a decision problem in time t and outputs the
correct answer with probability 0.5, then there is a randomized algorithm for the problem that runs
in time Θ(t) and outputs the correct answer with probability at least 0.99.
Solution: False. Every decision problem has an algorithm that produces the correct answer with
probability 0.5 just by flipping a coin to determine the answer.
n
(d) T F Let H : {0, 1} → {1, 2, . . . , k} be a universal family of hash functions, and let
programminghomeworkhelp.com

S ⊆ {0, 1} be a set of |S| = k elements. For h chosen at random from H, let E
be the event that for all y ∈ {1, 2, . . . , k}, the number of elements in S hashed to y is at most 100,
that is, |h (y) ∩S| ≤ 100. Then we have Pr {E } ≥ 3/4. −1−1
Solution: False. |h (y) ∩ S| is likely to be Θ(log k/ log log k) for some y. Only its expectation is
O(1).
(e) T F Let Σ = {a, b, c, . . . , z} be a 26-letter alphabet, and let s ∈ Σ and p ∈ Σ n b e m
strings of length n and m < n respectively. Then there is a Θ(n)-time algorithm to check whether p
is a substring of s.
Solution: True. E.g., using suffix trees.
(f) T F If an iteration of the Ford-Fulkerson algorithm on a network places flow 1 through an edge
(u, v), then in every later iteration, the flow through (u, v) is at least 1.
Solution: False: A later augmenting path may pass through (v, u), causing the flow on (u, v) to be
decreased.
(g) T F There exists a minimization problem such that (i) assuming P /= NP , there is no
polynomial-time 1-approximation algorithm for the problem; and (ii) for any constant ǫ >
0, there is a polynomial-time (1 + ǫ)-approximation algorithm for the problem.
Solution: True. There are NP-hard optimization problems with a PTAS, such as
PARTITION, as we saw in class.
Use the substitution method to show that the recurrence

T (n) = √
n T (√
n) + n
has solution T (n) = O(n lg lg n).
Solution: First, prove a base case. For 4 ≤ n ≤ 16, let T (n) = O(1) ≤ k for some k > 0. For
some c ≥ k/4, we have that T (n) ≤ cn lg lg n.
Now, prove the inductive case. Assume T (n) ≤ cn lg lg n for some c > 0. Then:
T (n) = √
n T (√
n) + n ≤ √
n c√
n lg lg √
n + n
We now find c > 0 so that
√
n c√
n lg lg √
n + n ≤ cn lg lg n nc(lg lg n − lg 2) + n ≤
cn lg lg n −c lg 2 + 1 ≤ 0 1 ≤ c
Hence, the inductive case holds for any c ≥ 1. Setting c = max{k/4, 1}, we have T (n) ≤
cn lg lg n for all 4 ≤ n.
Problem 3. Updating a Flow Network
Let G = (V, E) be a flow network with source s and sink t, and suppose that each edge e ∈ E has
capacity c(e) = 1. Assume also, for convenience, that |E| = Ω(V ).
(a) Suppose that we implement the Ford-Fulkerson maximum-flow algorithm by using depth-
first search to find augmenting paths in the residual graph. What is the worst- case running
time of this algorithm on G?
Solution: Since the capacity out of the source is |V |− 1, a mincut has value at most
|V |− 1. Thus the running time is O(V E ).

(b) Suppose that a maximum flow for G has been computed using Ford-Fulkerson, and a new edge
with unit capacity is added to E. Describe how the maximum flow can be efficiently updated.
(Note: It is not the value of the flow that must be updated, but the flow itself.) Analyze your
algorithm.
Solution: Simply run one more BFS to find one augmenting path in the residual flow network.
This costs O(E) time. One path suffices because all edges have capacity 1 so any augmenting
path will have capacity 1 as well.
We could also precompute in O(E) time for each vertex in the graph an edge on a path either to
sink or from source in the residual flow network. (Can’t have both if the flow
is maximum!) Then, given the new edge we can update the flow in O(V ) time.
(c) Suppose that a maximum flow for G has been computed using Ford-Fulkerson, but an edge is
now removed from E. Describe how the maximum flow can be efficiently updated. Analyze your
algorithm.
Solution: In the residual flow network, find a path from sink to source via the edge to be
removed. To do so, find a path from sink to the starting endpoint of the directed edge in the
residual flow network and similarly the second segment of the path. Then diminish the flow by one
unit along the path. Since all edges have capacity 1, this removes the flow from the edge to be
removed. After removing the edge, search for augmenting path in the residual flow network. This
is needed in case the edge was not in the minimum cut hence the flow can be redirected. The total
cost is O(E).
Recall the following problem abstracted from the take-home quiz, which we shall refer to as
Eleanor’s optimization problem. Consider a directed graph G = (V, E) in which every edge e ∈ E
has a length ℓ(e) ∈ Z and a cost c(e) ∈ Z. Given a source s ∈ V , a destination t ∈ V , and a cost x,

find the shortest path in G from s to t whose total cost is at most x.
We can reformulate Eleanor’s optimization problem as a decision problem. Eleanor’s decision
problem has the same inputs as Eleanor’s optimization problem, as well as a distance d. The problem
is to determine if there exists a path in G from s to t whose total cost is at most x and whose length is
at most d.
(a) Argue that Eleanor’s decision problem has a polynomial-time algorithm if and only if Eleanor’s
optimization problem has a polynomial-time algorithm.
Solution: Reduce decision to optimization: solve optimization, output “YES” iff length of found
path is less than d.
Reduce optimization to decision: Compute sum of all edge lengths which gives dmax.
Recall the NP-complete PARTITION problem: Given an array S[1 . . n] of natural numbers and a
value r ∈N, determine whether there exists a set A ⊆ {1, 2, . . . , n} of indices such that
(b) Prove that the Eleanor’s decision problem is NP-hard. (Hint: Consider the graph il- lustrated
below, where the label of each edge e is “c(e), ℓ(e)”.)
Solution: Label the left-hand vertex in the hint graph as s and the right-hand vertex
t. Run the algo to the decision problem with this graph and x = c and d = c to finish the
reduction. Any s − t path in the graph corresponds to a set A of the indices of Σthose ”widgets”
where the path follows thΣeupper branch. The length of this path is i∈A S[i], while the cost of this

path is i∈/A S[i]. The decision algo says YES iff there exists a path of both length and cost less
than c, that is, the answer to the PARTITION problem is YES.
If the decision algo returns “NO“ for dmax then there is no solution. Now binary search for the
smallest d in the range [0, dmax] on which the decision algo returns “YES”. Note log dmax is
polynomial in input size.
(c) Argue on the basis of part (b) that Eleanor’s decision problem is NP-complete.
Solution: The witness is the path which can be checked in time linear in the number of its
edges, which is less than |V |. Since the decision problem belongs to NP and is NP-hard, it is
NP-complete.
(d) The solutions to the take-home quiz showed that there is an efficient O(xE +xV lg V ) algorithm
for Eleanor’s optimization problem. Why doesn’t this fact contradict the NP-completeness of
Eleanor’s decision problem?
Solution: An algorithm with runtime O(xE + xV lg V ) has runtime polynomial in the value of
x. In other words, the runtime is actually pseudo-polynomial, rather than truly polynomial. We
showed that Eleanor’s optimization problem was hard by a reduction from Partition, which is
weakly NP-hard. As a result, we have only shown that Eleanor’s Optimization Problem is weakly
NP-hard, and so it’s not a contradiction to have a pseudo-polynomial algorithm.

Design and Analysis of Algorithms Assignment Help

Recommended

More Related Content

Similar to Design and Analysis of Algorithms Assignment Help (20)

More from Programming Homework Help (20)

Recently uploaded (20)

Design and Analysis of Algorithms Assignment Help