关键词 > ORIE4154/5154
ORIE 4154/5154 Spring 2023
发布时间:2023-03-25
Hello, dear friend, you can consult us at any time if you have any questions, add WeChat: daixieit
ORIE 4154/5154
Spring 2023
Rules.
· You cannot use your free days for the prelim.
· You are not allowed to collaborate with other students.
· Questions only during lecture allocated times. There is no extra office hours.
· Explain your answers in detail, do not write just the final answer.
Problems below are not related to each other, unless otherwise stated.
Demand response model
(15pts) Problem 1
Consider the following demand function d(p) = 3 - ^p .
Part (a)
Write the region where the demand is non-negative. Compute the inverse demand function q(x).
Part (b)
Write the revenue function as a function of the price, R(p), and check whether is concave in the region you found in (a). Compute the optimal price p* and R(p* ).
Part (c)
Compute the elasticity e(p) and show that |e(p* )| = 1 (| . | indicates absolute value).
(25pts) Problem 2
Suppose you want to sell one item and there are two customers labeled 1 and 2 which arrive sequentially: first 1 and then 2. If the item is sold to 1, then 2 leaves without anything. If the item is not sold to 1, then 2 has a chance to buy it. We assume that both customers have random valuations for the item V1 and V2 which are i.i.d. Uniform[0, 1]. Let p be the uniform price (i.e. the same for both customers) you would like to set for the item.
Part (a)
If p is the item’s price, what is the demand function of customer 1, i.e., what is the probability that customer 1 buys the item? What is the demand function of customer 2, i.e., what is the probability that customer 2 buys the item?
Part (b)
Given the demand functions above, write the total expected revenue as a function of p. Compute the optimal price p* and optimal revenue.
Part (c)
Show that under p* the probability that customer 1 buys the item and the probability that customer 2 buys the item are different.
Part (d)
Now, you would like to impose a fairness condition for the customers’ probabilities of buying the item. For this, suppose that you would like to find a price p1 for customer 1 and a price p2 for customer 2. Write the optimization problem that maximizes the total expected revenue dependent on p1 and p2 , subject to the buying probabilities being equal (also dependent on p1 and p2 ).
Part (e)
Compute the optimal prices p 1(*) and p2(*) of your problem in Part (d); for this part only, you are allowed to use a software, e.g., WolframAlpha. Compare the optimal revenue for this case with the one obtained in Part (b). How do p* , p 1(*) and p2(*) relate (in terms of magnitude)? Comment briefly on the meaning of this.
Multiple fares single-resource capacity allocation
(10pts) Problem 1
Consider the two fare-class allocation problem, with capacity C = 100, fares p1 = 200 and p2 = 100, and demands D1 ~ Uniform{0, 1, . . . , 120}1 and D2 ~ Geometric(1/200). What is the optimal protection-level for fare-class 1?
(10pts) Problem 2
Suppose there are n fare-classes, C = 12 and the values of ∆Vt-1 (s) for a specific t are given in Table 1.
s |
1 |
2 |
3 |
4 |
5 |
6 |
7 |
8 |
9 |
10 |
11 |
12 |
∆Vt-1 (s) |
100 |
80 |
75 |
75 |
60 |
55 |
55 |
50 |
50 |
45 |
40 |
40 |
What would be the optimal protection level y*t-1 if the price of the current class t is pt = 58? What would be the optimal protection level if the price is pt = 38?
(10pts) Problem 3
Suppose you are implementing the policy defined by the optimal protection levels. In stage t, you observed a demand Dt = 25. If the optimal protection level for the future classes is y*t-1 = 10 and the current remaining capacity is st = 30, how many seats do you sell for class t under the optimal policy? Suppose that instead you had a remaining capacity st = 9, how many seats would you sell under the optimal policy?
(30pts) Problem 4: Overbooking
Suppose you are selling tickets for a broadway show. You have C seats to sell, and customers can reserve a seat in advance by paying a price p. However, each customer who has a reservation (independently) shows up with probability q and does not show up with probability 1 - q . We assume that if a customer does not show up, they are NOT refunded their reservation cost. In order to increase revenues, you can o"erbook by admitting b 2 C reservations. However, each customer who shows up, but is denied admission due to lack of seats, is refunded an amount θ .
Part (a)
Let N(b) be the (random) number of people who shows up to the show assuming that b reservations were allowed. What is the distribution of N(b)?
Part (b)
Let R(b) be the expected profit (i.e., expected earnings minus refund costs) you get if you allow b reservations. Write down an expression for R(b) in terms of p, θ, C and N(b).
Part (c)
Assume that we have the following expression for ∆R(b):
∆R(b) = R(b + 1) - R(b) = q . (p - θP[N(b) 2 C])
Suppose that P[N(b) 2 C] is increasing in b. Argue that ∆R(b) is decreasing in b. As we did in class and in HW2, find an expression for the optimal number of reservations b* that maximizes your profit. The expression should be explicit in terms of q, p, θ, N(b) and C .
Network revenue management
(10pts) Problem 1
Assume you are a revenue manager of an airline. You are in charge of the products that relate to 3 cities: A, B and C. Your resources are non-stop flights AB and BC. The products offered are: flight AB (non-stop) at price pAB = 90, flight BC (non-stop) at price pBC = 200 and flying AC through B (i.e., fly AB and then BC) at price pAC = 320. A super computer gave you access to the following table of value functions Vt(sAB , sBC ) for a given t, where sAB and sBC indicate the remaining capacities of each resource. This table allows you to implement the optimal policy we
AB |
0 |
1 |
2 |
3 |
0 |
0 |
500 |
800 |
900 |
1 |
350 |
700 |
800 |
1200 |
2 |
450 |
700 |
1000 |
1200 |
3 |
550 |
800 |
1000 |
1200 |
studied in Network RM. If a customer arrives requesting to fly AB and the remaining capacities are sAB = 3 and sBC = 1, should you accept this request under the optimal policy? If a customer arrives requesting to fly AC and the remaining capacities are sAB = 2 and sBC = 2, should you accept this request under the optimal policy?
(10pts) Problem 2
Assume you are running a hotel and your remaining capacities for Thursday, Friday and Saturday are CTh = 3, CF = 2 and CSa = 1, respectively. At the beginning of the season, you computed the optimal solutions of the dual of the fluid LP and you obtained zT(*)h = 50, zF(*) = 200 and zS(*)a = 210. One customer arrives requesting to stay Thursday, Friday and Saturday at price $470, does the bid- price policy accept this request and why? After this customer, another customer arrives requesting to stay Friday and Saturday at price $415, does the bid-price policy accept this request and why?
Choice models and assortment optimization
Problem 1
Consider a universe of 4 items {1, 2, 3, 4} and a MNL choice model with the following parameters (v0 , v1 , v2 , v3 , v4 ) = (1, 0.5, 2, 4, 1).
(5pts) Part (a)
Compute p1 ({1, 3, 5}), p3 ({1, 2, 3}), p0 ({1, 5}) and p4 ({1, 2}).
(15pts) Part (b)
Find the optimal assortment and the optimal revenue under the above MNL model when the items have revenues (r1 , r2 , r3 , r4 ) = (10, 6, 5, 4).