Homepage › Solution manuals › Dimitris Bertsimas › Introduction to Linear Optimization › Exercise 1.8 (Road lighting)
Exercise 1.8 (Road lighting)
Consider a road divided into segments that is illuminated by lamps. Let be the power of the th lamp. The illumination of the th segment is assumed to be , where are known coefficients. Let be the desired illumination of road .
We are interested in choosing the lamp powers so that the illuminations are close to the desired illuminations . Provide a reasonable linear programming formulation of this problem. Note that the wording of the problem is loose and there is more than one possible formulation.
Answers
Our goal is to minimize the difference between the desired illumination and the actual illumination . We could choose any kind of metric to do so, but the one suitable for the linear optimization is the sum of absolute errors (cf. residual minimization from Section 1.3):
Considering the fact that the powers of the lamps cannot be negative, we formulate the above optimization problem as
We use the procedure from the Section 1.3 (Problems involving absolute values) to get rid of the absolute values:
and to reformulate the above as a linear optimization problem: