Exercise 1.6 (Rocket control)

Provide linear programming formulations of the two variants of the rocket control problem discussed at the end of Section 1.3.

Answers

First model. The first option described in the Section 1.3 is to minimize the total fuel consumption:

minimize t = 0 T 1 | a t | subject to x 0 = 0 , v 0 = 0 , x t + 1 = x t + v t v t + 1 = v t + a t , t = 0 , , T 1 x T = 1 , v T = 0
(1)

As shown in the previous exercise, for each t = 0 , , T , we can equivalently introduce nonnegative dummies a t + , a t and replace every occurrence of a t with a t + + a t and every occurrence of | a t | with a t + a t . The resulting linear optimization problem is then the following:

minimize t = 0 T 1 a t + + a t subject to x 0 = 0 , v 0 = 0 , x t + 1 = x t + v t v t + 1 = v t + a t + a t , t = 0 , , T 1 x T = 1 , v T = 0 , a t + , a t 0 , t = 0 , , T 1

Second model. The second option suggested in the book is to minimize the maximum thrust that is required for the rocket:

minimize max { t = 0 , , T 1 } | a t | subject to x 0 = 0 , v 0 = 0 , x t + 1 = x t + v t v t + 1 = v t + a t , t = 0 , , T 1 x T = 1 , v T = 0
(2)

We can move the maximum function to the constraints section by writing

minimize z subject to x 0 = 0 , v 0 = 0 , x t + 1 = x t + v t v t + 1 = v t + a t , t = 0 , , T 1 x T = 1 , v T = 0 z | a t | , t = 0 , , T 1

In a manner similar to the previous exercise, we replace z | a t | by the double condition z a t and z a t :

minimize z subject to x 0 = 0 , v 0 = 0 , x t + 1 = x t + v t v t + 1 = v t + a t , t = 0 , , T 1 x T = 1 , v T = 0 z a t z a t , t = 0 , , T 1 .
User profile picture
2022-02-09 20:08
Comments