Sei sulla pagina 1di 11

Recall

- Example 1
James produces two commodi0es, x1 and x2. These two
products are processed through 2 machines. Maximum
hours available are 120 and 180 for machines 1 and 2,
respec0vely. Profit is P45 for x1 and P55 for x2. Determine
the op0mal x1 and x2 that should be manufactured in order
to maximize profit.

x1 x2
Machine 1 6 4 120
Machine 2 3 10 180
Profit 45 55
Max 45x1 + 55x2
x2 subj. to
6x1 + 4x2 ≤ 120
6x1 + 4x2 = 120 3x1 + 10x2 ≤ 180
x1, x2 ≥ 0

A VOF
(0, 18) 990
(10, 15) 1275
(20, 0) 900
B
(0, 0) 0

3x1 + 10x2 = 180


x1
D C
Lemma 1: If the set {x⎟ Ax = b, x ≥ 0} is feasible, then it has a
basic feasible solution. The basic feasible solution is a vertex in the set

Lemma 2: If x* is a basic feasible solution of the set


{x ⎟ Ax = b, x ≥ 0}, then x* is an extreme point of the
set. The opportunity set contains the basic solution

Lemma 3: Every extreme point of the set {x⎟ Ax = b, x ≥ 0} is a


basic feasible solution. Every extreme point is a basic feasible solution
Computational form of LPP:
Min -cʹx
subject to
Ax=b
x≥0
Example: c’ x
45x1 + 55x2 = [ 45 55 ] [ x1 ]
Max 45x1 + 55x2 [ x2 ]

subject to
6x1 + 4x2 ≤ 120 Max 45x1 + 55x2 = Min -45x1 - 55x2
Max c0 = Min -c0
3x1 + 10x2 ≤ 180
x1, x2 ≥ 0
Standard form: Convert inequality to equality sign by adding a slack so that you now have 4 variables
Max 45x1 + 55x2
subject to
6x1 + 4x2 + x3 = 120
3x1 + 10x2 + x4 = 180
x1, x2, x3, x4 ≥ 0
Computational form n=4
r=2
Min - 45x1 - 55x2 :. There are (n-r) 2 variables I can assign arbitrary values to

subject to
6x1 + 4x2 + x3 = 120
3x1 + 10x2 + x4 = 180
x1, x2, x3, x4 ≥ 0
Def. If the LPP is in computational form, there exists a basis B
such that B = I, cB = 0, then the LPP is said to be in canonical
* Basis of A = Columns of A that form a nonsingular matrix
form with respect to the basis of A. * Nonsingular - Has an inverse & determinant is not = 0
* Canonical - W/c columns of A form an identity matrix
Min -c x = -c0
If you find that the basis of A form an identity matrix & the
s.t. Ax = b coefficients of f(x) are 0[c^B = 0]. then it is canonical

x ≥ 0
Example:
c = [− 45 − 55 0 0]
⎡6 4 1 0⎤ 3rd & 4th

A= ⎢ ⎥
columns
comprise the x B = [x3 x4 ] x N = [x1 x2 ]
⎣ 3 10 0 1 ⎦ basis of A
⎡ xB ⎤
[− c B
− cN ] ⎢ N ⎥ = −c0
⎣x ⎦
x1 & x2 are nonbasic variables :. assign values of zero
x3 & x4 are basic variables

⎡ x B ⎤ ⎡b⎤
[B N ] ⎢ N ⎥ = ⎢ ⎥
⎣ x ⎦ ⎣0⎦
BxB = b
IxB = b
xB = b
x1 = 0, x2 = 0, x3 = 120, x4 = 180 Basic feasible
solution

The solution in the graph would be at D = (0,0)


But you can have a higher profit! :. Machine is at slack
Digression C

C.1 Elementary row operations refer to:


(i)  interchanging two rows of a matrix
(ii)  changing a row by adding to it a multiple of another
row, and
(iii) multiplying each element in a row by the same nonzero
number.

C.2 Elementary row operations do not change the rank


of a matrix.
C.3 Example:

⎡1 4 17 ⎤ ⎡1 4 17 ⎤
A1 = ⎢⎢2 12 46⎥⎥ A2 = ⎢⎢0 4 12 ⎥⎥
⎢⎣3 18 69 ⎥⎦ ⎢⎣0 6 18 ⎥⎦

⎡1 4 17 ⎤
⎡1 4 17 ⎤
A3 = ⎢⎢0 4 12 ⎥⎥ ⎢ ⎥
A4 = ⎢0 4 12 ⎥
⎢⎣0 1 3 ⎥⎦
⎢⎣0 0 0 ⎥⎦

ρ (A1) = ρ (A2) = ρ (A3) = ρ (A4)


Steps in Simplex Algorithm
The LPP is canonical with respect to some basis and the basic
solution associated with this basis is feasible (b ≥ 0).
(i)  Choose the pivot column k such that ck < 0 and |ck| > |cj|
∀ j ≠ k. If there is no ck < 0, and b ≥ 0, ck ≥ 0, the basic
solution associated with the basis is optimal. Otherwise,
continue with the next step.
(ii) Choose the pivot row r such that br/Ark = mini {bi / Aik} > 0.
If there is no such r, STOP. The LP is unbounded.
Otherwise, continue.
(iii) Pivot on Ark. Go to step (i).

Example:
Max 45x1 + 55x2
subject to
6x1 + 4x2 ≤ 120
3x1 + 10x2 ≤ 180
x1, x2 ≥ 0
Standard form:
Max 45x1 + 55x2 Note:
* Pivot - Change pivot element to 1
subject to and change other elements to zero (in
that column)
6x1 + 4x2 + x3 = 120 It is a basic feasible solution that
corresponds to D. Stop only if you
3x1 + 10x2 + x4 = 180 don’t find negative values in Row 0.
1. Pivot the most (-) column -> 55
x1, x2, x3, x4 ≥ 0 2. Pivot the most (-) row (Look at the
RHS)
Computational form Ex: 120/4 = 30 & 180/10 = 18 based
on the key column :. 4 & 10
Min - 45x1 - 55x2 - Quotients are 30 & 18 :. choose the
smaller number
subject to
6x1 + 4x2 + x3 = 120
3x1 + 10x2 + x4 = 180
x1, x2, x3, x4 ≥ 0 Basic variables - x3, x4

Tableau 1: Nonbasic variables - x1, x2


-> Assign 0 values :. x1 & x2 Profit = 0
-> x3, x4 = Slack
x1 x2 x3 x4 RHS
Row 0
Canonical w.r.t.
-cʹx -45 -55 0 0 0 c0 xB = (x3, x4)
6 4 1 0 120 Row 1
Ax b Pivot on a22 = 10
3 10 0 1 180 Row 2
Tableau 2:
From Table 1
x1 x2 x3 x4 RHS 1. Multiply Row 2 by 1/10
2. Multiply by -4/10 & add to Row 1
x1 = 0, x2 = 18 :. = Point A, where VOF = 990
-57/2 0 0 11/2 990
X1 and X4 are not basic, but identity matrix
24/5 0 1 -2/5 48 is basic

3/10 1 0 1/10 18 Use any row:


-57/2 x1 + 11/2 x4
24/5 x1 - 2/5 x3 + 2/5 x4 = 48
Canonical w.r.t. xB = (x3, x2) 3/10 x1 + x2 + 1/10 x4 = 18

Pivot on a11 = 24/5 Basic variables is x2 & x3


X2 = 18, X3 = 48 (Since x1 & x4 = 0)

Tableau 3: Why 990? Bec u have 18 of 55 (55x2) You


choose 48 & 18 bec the identity matrix is
x1 x2 x3 x4 RHS there

0 0 95/16 25/8 1275


1 0 5/24 -1/12 10
0 1 -1/16 11/10 15
Canonical w.r.t. xB = (x1, x2)
x1 = 10; x2 = 15; x3 = 0; x4 = 0

Potrebbero piacerti anche