Scribd
Upload
75 views28 pages

Dr. Joon-Yeoul Oh: IEEN 5335 Principles of Optimization

The lecture discusses linear programming principles using a chair and table production example. It introduces the LP model, showing constraints for material limits. The example is solved graphically, finding the optimal production point that maximizes profit within feasible regions defined by the constraints. Key assumptions of LP like proportionality, additivity and divisibility are also covered. The review emphasizes that LP problems have linear objectives and constraints with known coefficients.

Uploaded by

Soumojit Kumar
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
75 views28 pages

Dr. Joon-Yeoul Oh: IEEN 5335 Principles of Optimization

The lecture discusses linear programming principles using a chair and table production example. It introduces the LP model, showing constraints for material limits. The example is solved graphically, finding the optimal production point that maximizes profit within feasible regions defined by the constraints. Key assumptions of LP like proportionality, additivity and divisibility are also covered. The review emphasizes that LP problems have linear objectives and constraints with known coefficients.

Uploaded by

Soumojit Kumar
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 28

IEEN 5335

Principles of Optimization

Lecture 2

Dr. Joon-Yeoul Oh

Dept. of Mechanical & Industrial Engineering


Texas A&M University – Kingsville
Chair and Table example
 Q: How many chairs and/or tables to make for max
profit?
 Each require pine, oak & time
 Limits on pine, oak and time.
 Profit: $6/chair, $12/table.
 Data:
Item Chair Table Limit
Profit $6 $12 ---
Pine 3 bf 12 bf 120 bf
Oak 8 bf 4 bf 160 bf
Time 6 hr 10 hr 150 hr
2
Chair and Table example - model
 LINDO format

! C = # of chairs to produce
! T = # of tables to produce

Max 6 C + 12 T
s.t.
pine) 3 C + 12 T <= 120
Oak) 8 C + 4 T <= 160
Time) 6 C + 10 T <= 150
end
3
Chair and Table example - model
 LINDO output

OBJECTIVE FUNCTION VALUE


1) 162.8571
VARIABLE VALUE REDUCED COST
C 14.285714 0.000000
T 6.428571 0.000000

ROW SLACK OR SURPLUS DUAL PRICES


PINE) 0.000000 0.285714
OAK) 20.000000 0.000000
TIME) 0.000000 0.857143
4
C & T example – graphical method

 Usable if the number of decision variables is 2


 Constraint: Half plane
 Bounded by gi = b
 One side is Feasible
 Feasible area: Overlap of all constraints
 Object function: set of straight lines

5
C & T example – graphing constraints plot

 Two points determine a line


 Any two different points
 e.g. when one DV = 0
 All points on same side same
 Any point
 e.g. (0,0)

6
Plotting example (1)

7
Plotting example (2)

8
Plotting example (3)

9
C & T example – graphing constraints

10
C & T example – graphing constraints

11
C & T example – feasible area

12
C & T example – optimal point

13
C & T example – optimal direction

14
C & T example – optimal solution

15
C & T example – optimal solution

 Intersection of Pine & Time


 Pine: 3C + 12T = 120
 Time: 6C + 10T = 150
 Opt. Point: ( 100/7, 45/7 )
 z = 6(100/7) + 12(45/7)
= $162.86

16
Another example
– solve it using graphical method

Max 2 A + 3 B
s.t.
limit1) A + B <= 5
limit2) 8 A + 2 B <= 8
limit3) 3 A + 6 B <= 12
end

17
LP models – possible results

 LP model has Solution then,


 x*: Feasible & Optimal
 LP model has No Solution then,
 No feasible region
 z value - unlimited favored direction

18
LP assumptions

 1. Proportionality
 2. Additivity
 3. Divisibility
 4. Certainty

19
LP assumptions - proportionality
 If aij units of the ith item are consumed (or produced) in
carrying out activity j at unit level, then aijxj units of this
item are consumed (or produced).
 For each and every d.v. xj,
 z(xj) = cj xj
 gi(xj) = aij xj
 e.g. Cannot have:
 #Price breaks, powers of xj

 Must have linear functions!

20
LP assumptions - additivity

 Total consumption (or production) of an item is


equal to the sum of the various quantities of the
item consumed (or produced).
 For all values of all d.v. xj,
 z = c1x1 + c2x2 + ··· + cnxn
 And, for every constraint,
 gi = ai1x1 + ai2x2 + ··· + ainxn
 e.g. Cannot have:
 Interaction, cooperation, etc.
 Must have linear functions!

21
LP assumptions - divisibility

 For all d.v. xj,


 Any real value is allowed
 e.g. Cannot have:
 Integer only values
 Finite number of allowed values
 Must have linear functions of real variables!

22
LP assumptions - certainty

 Assume we know exactly


 All coefficients (cj, aij)
 All constraint limits (bi)
 e.g. all constants are exact
 Must have linear functions with known constants of real
variables!

23
Coping with assumptions

 Reality:
 Situation seldom fits exactly
 Seldom know exactly
 Coping
 “Close enough” techniques
 Sensitivity analysis

24
LP assumption summary

 The optimization function and all constraints are


linear functions of real d.v.s.
 Coefficients are exactly known (but later we use
sensitivity analysis).

25
Lecture review
 Basic Model:
 Max or min z(x)
 Subject to: gi(x) <, >, or = bi
 Non-negativety: xj > 0
 All functions linear
 All constants known
 Graphical Method (only for 2 dimensional)
 Use gi(x) to find boundary
 Find which side is feasible
 Plot z line
 Find optimal direction
26
Comments 1

 This lecture contains one LP modeling example,


and the model is solved by graphical method.
 As you can see on the graphical method, you
always can find the optimal solution of LP on the
boundary ( or corner point feasible solution) of
solution set (feasible area).
 The optimal solution is one of the corner point
feasible solutions, if the model has one optimal
solution.

27
Comments 2

 So, just check the corner point feasible solutions,


and pick the best one, then that is the optimal.
 However, this graphical method is only good for
two dimension (two variables).
 If a model has more than 2 variables, we will use a
different method, which we will discuss later.

28

You might also like