Scribd
Upload
114 views

HW 5

This document provides instructions for homework #5 in course 02-713, due on March 3rd. It describes 5 problems to solve for the homework. Students should work independently but indicate who they worked with, and solutions should be submitted as a PDF. For problems asking for algorithms, students should describe the algorithm, argue why it is correct, and estimate the running time using O notation.

Uploaded by

Tushar Gupta
Copyright
© © All Rights Reserved
Available Formats
Download as PDF, TXT or read online on Scribd
114 views

HW 5

This document provides instructions for homework #5 in course 02-713, due on March 3rd. It describes 5 problems to solve for the homework. Students should work independently but indicate who they worked with, and solutions should be submitted as a PDF. For problems asking for algorithms, students should describe the algorithm, argue why it is correct, and estimate the running time using O notation.

Uploaded by

Tushar Gupta
Copyright
© © All Rights Reserved
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 2

02-713 Homework #5

Due: Mar. 3 by 9:30am


You may discuss these problems with your current classmates, but you must write up your solutions independently, without using common notes or worksheets. You must indicate at the top of your homework
who you worked with. Your write up should be clear and concise. You are trying to convince a skeptical reader that your answers are correct. Your homework should be submitted via Autolab (https:
//autolab.cs.cmu.edu/02713-s14/) as a typeset PDF. A LaTeX tutorial and template are available on
the class website if you choose to use that system to typeset. For problems asking for an algorithm: describe
the algorithm, an argument why it is correct, and an estimation of how fast it will run. Use O notation for
running times.
1. Often we want to compute shortest paths between all pairs of vertices in a graph G = (V, E). There
are several ways to do this. One way to do this is to run Dijkstras algorithm using each vertex as the
start node. This takes O(|V ||E| log |V |) time, but doesnt work if there are negative weights. Here we
explore one fix to that suggested by Edmonds and Karp.
(a) Give an example where Dijkstras algorithm fails if there are edges of negative weight even if there
are no negative cycles.
(b) Let G = (V, E) be a directed graph with possibly-negative weights d(u, v) on the edges (but no
negative cycles). Define a new graph G0 that is created from G by adding a new vertex s and edges
of 0-weight from s to every node in V . Let dists (v) be the shortest path distance from s to v in G0
computed via Bellman-Ford. Define new weights on G as d0 (u, v) = d(u, v) + dists (u) dists (v). Argue
that d0 (u, v) defined in this way is always positive.
(c) Argue that any path is a shortest path in G using weights d0 (u, v) if and only if it is a shortest path
in G using weights d(u, v). (In other words, changing the weights to d0 (u, v) preserves which paths are
shortest paths.)
(d) Describe how to use (b) and (c) to compute all pairwise shortest paths in O(|V ||E| log |V |)-time.
2. Let T = (V, E) be a tree with nodes V and edges E that is rooted at node r, and let w(u) be the weight
of node u. Recall that an independent set is a subset U of V such that no edge exists in between any
two nodes in U . Give a polynomial time algorithm to compute the weight of largest-weight Independent
Set in T .
3. Let x1 , . . . , xn be a list of integers. Give an O(n) divide-and-conquer algorithm to find the largest
possible sum of a subsequence of consecutive items in the list.
Example: 10 -20 3 4 5 -1 -1 12 -3 1
3 + 4 + 5 + -1 + -1 + 12 = 22

4. You have have found n id cards with magnetic strips that each encodes some id number. Due to
security precautions, you do not have any way to read the id number off of a card directly. However,
you do have a tester machine that will read two cards and tell you whether they have the same id
number encoded on them.
Give an O(n log n)-time algorithm to test whether some subset of more than n/2 cards have the same
identifier encoded on them.
5. 02-713 only! You want to draw a two-dimensional skyline like the following:

y4

h5
h4

y3
h6

h1

y5

y1
y2

h2
h3
l1

l 2 r1

l3

r3 l 4

r2

r4

l4

l5

r4

r5

x1

x2

x3

x4

x5

x6

x7

You are given a list of the building x-coordinates and their heights:
(l1 , h1 , r1 ), (l2 , h2 , r2 ), . . . , (ln , hn , rn )
This list will be sorted in increasing order of left-most x-coordinate. Sketch an O(n log n) algorithm
to produce the skyline line to draw in the format:
x1 , y1 , x2 , y2 , x3 . . .
meaning that at x1 we draw a building at height y1 until x2 at which point we draw a line up or down
to high y2 and then continue horizontally until x3 and so on.
Note that for this problem, you need only sketch the main ideas, not provide as detailed a discussion
as typically.

You might also like