Anda di halaman 1dari 3

Design and Analysis of Algorithms October 11, 2018

Massachusetts Institute of Technology 6.046/18.410


Konstantinos Daskalakis, Srinivas Devadas,
Debayan Gupta, Ronitt Rubinfeld Problem Set 5

Problem Set 5
This problem set is due at 10:00pm on Wednesday, October 17, 2018.
Please make note of the following instructions:

• This assignment, like later assignments, consists of exercises and problems. Hand in solu-
tions to the problems only. However, we strongly advise that you work out the exercises
for yourself, since they will help you learn the course material. You are responsible for the
material they cover.

• We require that the solution to the problems is submitted as a PDF file, typeset on LaTeX,
using the template available in the course materials. Each submitted solution should start
with your name, the course number, the problem number, your recitation section, the date,
and the names of any students with whom you collaborated.

• We will often ask you to “give an algorithm” to solve a problem. Your write-up should
take the form of a short essay. Start by defining the problem you are solving and stating
what your results are. Then provide: (a) a description of the algorithm in English and, if
helpful, pseudo-code; (b) a proof (or proof sketch) for the correctness of the algorithm; and
(c) an analysis of the running time. We will give full credit only for correct solutions that are
described clearly and convincingly.
2 Problem Set 5

EXERCISES (NOT TO BE TURNED IN)

Linear Programming

• Do Exercise 29.1-9 in CLRS (pg. 858)

• Do Exercise 29.2-6 in CLRS (pg. 864)

• Do Exercise 29.5-9 in CLRS (pg. 893)


Problem Set 5 3

Problem 5-1. LP Oracle [100 points]


Entrepreneur Melon Usk started companies that make n different types of vehicles (for example,
his company AreaY makes spaceships, while his company Edison makes electric cars). Each
vehicle type uses some of n different types of resources, but in different proportions. Specifically,
suppose one vehicle of type i requires aij units of resource j to make. Further, suppose one vehicle
of type i sells for ci dollars.
Melon Usk has just received a shipment of resources. His shipment contains bj units of resource j.
He would like to know how many of each vehicle type to make to maximize his revenue. Assume
Melon can sell every vehicle he makes. Values do not have to be integral.

(a) [20 points] Formulate this as an LP problem in standard form.


(b) [40 points] You have a very smart friend at MIT who, when given an LP with at
most n + 1 nonnegative variables and n + 1 constraints, can in constant time produce
a feasible solution if it exists, or tell you there is no solution if no solutions exist.
Assume the n + 1 constraints do not have to include nonnegativity conditions; your
friend assumes all variables are nonnegative.
Suppose P ∗ , Melon’s maximum revenue given the resource constraints, is a finite
integer. Show how to find P ∗ using O(log P ∗ ) calls to the friend and O(n + log P ∗ )
time. Hint: Try to bound the value of P ∗ using not too many calls to the friend.
(c) [40 points] Melon thinks his shipment is too expensive, so he wants to renegotiate
resource prices with his supplier. Melon wants the total price of his shipment to be as
small as possible. However, the greedy supplier insists that for every type of vehicle
Melon makes, the cost of the resources in the vehicle must be at least 1/2 of the
sell price of the vehicle. Melon is unwilling to change the vehicle sell prices, so he
asks you to determine a set of prices {yj }, where resource j costs yj per unit, that
minimize his costs and satisfy the supplier. What is the smallest you can negotiate
Melon’s shipment price, in terms of P ∗ ? Describe how to determine an agreeable
set of prices in one call to your friend at MIT, that achieves the shipment price you
determined. Hint: Write out the LP, and compare it to your LP in part (a).

Anda mungkin juga menyukai