Decison-Aid Methodologies in Transportation Optimization Exercise 3 - - PowerPoint PPT Presentation

decison aid methodologies in transportation
SMART_READER_LITE
LIVE PREVIEW

Decison-Aid Methodologies in Transportation Optimization Exercise 3 - - PowerPoint PPT Presentation

Decison-Aid Methodologies in Transportation Optimization Exercise 3 Tom Robenek April 30, 2013 1 / 9 Agenda 1 Branch and Price 2 OPL Functions 2 / 9 1 Branch and Price Master Problem Initial Solution Sub-Problem 2 OPL Functions 3 / 9


slide-1
SLIDE 1

Decison-Aid Methodologies in Transportation

Optimization Exercise 3 Tomáš Robenek April 30, 2013

1 / 9

slide-2
SLIDE 2

Agenda

1 Branch and Price 2 OPL Functions

2 / 9

slide-3
SLIDE 3

1 Branch and Price

Master Problem Initial Solution Sub-Problem

2 OPL Functions

3 / 9

slide-4
SLIDE 4

Branch and Price

Start Solve Master Problem Initial Columns Pricing Solver Dual Variables New Columns with Negative Reduced Cost Integer Solution? No Columns with Negative Reduced Cost Branch & Bound No Output Solution yes End

  • Initial Solution
  • Column Generation – Lower

Bound

  • Branch and Bound –

Optimal Integer Solution

4 / 9

slide-5
SLIDE 5

Master Problem

Start Solve Master Problem Initial Columns Pricing Solver Dual Variables New Columns with Negative Reduced Cost Integer Solution? No Columns with Negative Reduced Cost Branch & Bound No Output Solution yes End

Idea

  • Cutting Stock Problem
  • relaxation of decision variables
  • solution space significantly reduced, e.g.

faster to solve

5 / 9

slide-6
SLIDE 6

Initial Solution

Start Solve Master Problem Initial Columns Pricing Solver Dual Variables New Columns with Negative Reduced Cost Integer Solution? No Columns with Negative Reduced Cost Branch & Bound No Output Solution yes End

  • has to be feasible!
  • in your case already given

6 / 9

slide-7
SLIDE 7

Sub-Problem

Start Solve Master Problem Initial Columns Pricing Solver Dual Variables New Columns with Negative Reduced Cost Integer Solution? No Columns with Negative Reduced Cost Branch & Bound No Output Solution yes End

Idea

  • generate better columns based on the

reduced cost

  • need of dual variables
  • finish, when no negative reduced cost

columns

7 / 9

slide-8
SLIDE 8

1 Branch and Price 2 OPL Functions

8 / 9

slide-9
SLIDE 9
  • in the master problem use float decision variable(s); duals do

not exist for integers

  • duals are per constraint, e.g.:
  • forall(i in Items)

duals: x[i] <= 5;

  • to get the dual value then:
  • duals[i].dual
  • you can use the execute function to print the values in

scripting log

  • no need for general code, solve it manually and copy paste the

values

  • don’t forget to increase the size of the range columns after

each iteration!

  • use one project with 2 configurations

9 / 9