lagrange multipliers erasmus mobility program (24apr2012) pollack mihály engineering faculty (pmmk)...

26
Lagrange Multipliers Erasmus Mobility Program (24Apr2012) Pollack Mihály Engineering Faculty (PMMK) University of Pécs João Miranda [email protected]

Upload: ada-watts

Post on 05-Jan-2016

223 views

Category:

Documents


3 download

TRANSCRIPT

Page 1: Lagrange Multipliers Erasmus Mobility Program (24Apr2012) Pollack Mihály Engineering Faculty (PMMK) University of Pécs João Miranda jlmiranda@estgp.pt

Lagrange Multipliers

Erasmus Mobility Program (24Apr2012)Pollack Mihály Engineering Faculty (PMMK)

University of Pécs

João Miranda

[email protected]

http://tiny.cc/jlm_estg ; http://tiny.cc/jlm_ist

Page 2: Lagrange Multipliers Erasmus Mobility Program (24Apr2012) Pollack Mihály Engineering Faculty (PMMK) University of Pécs João Miranda jlmiranda@estgp.pt

24-Apr-12 Lagrange Multipliers 2

Lagrange Multipliers

• Introduction: maximization of 2-variable function subject to one restriction;

• Generalization: maximization of n-variable function subject to m restrictions

• Tables & Chairs problem.

Page 3: Lagrange Multipliers Erasmus Mobility Program (24Apr2012) Pollack Mihály Engineering Faculty (PMMK) University of Pécs João Miranda jlmiranda@estgp.pt

24-Apr-12 Lagrange Multipliers 3

Lagrange Multipliers

Optima of multi-variable functions when submitted to several constraints.

• Problem:

– Obtain the maxima/minima of the function,

– But the variables x and y have to satisfy the equation:

),( yxfu

0),( yxg

Page 4: Lagrange Multipliers Erasmus Mobility Program (24Apr2012) Pollack Mihály Engineering Faculty (PMMK) University of Pécs João Miranda jlmiranda@estgp.pt

24-Apr-12 Lagrange Multipliers 4

• Assuming y(x) as implicit function of x, the derivative of the composite function u(x,y),

(necessary condition)

• Or,

Lagrange Multipliers

dx

du

0

0.

dx

dy

y

f

x

f

x

f

dx

dy

y

f

Page 5: Lagrange Multipliers Erasmus Mobility Program (24Apr2012) Pollack Mihály Engineering Faculty (PMMK) University of Pécs João Miranda jlmiranda@estgp.pt

24-Apr-12 Lagrange Multipliers 5

• The derivation of the implicit function, y(x):

then,

Lagrange Multipliers

yg

xg

dx

dy

.y

g

x

g

x

g

dx

dy

dx

dy

y

g. 0

Page 6: Lagrange Multipliers Erasmus Mobility Program (24Apr2012) Pollack Mihály Engineering Faculty (PMMK) University of Pécs João Miranda jlmiranda@estgp.pt

24-Apr-12 Lagrange Multipliers 6

• Conjugating the two zero equations:

where is known as Lagrange multiplier (1736-1813).

(http://en.wikipedia.org/wiki/Joseph_Louis_Lagrange) • Re-allocating terms,

Lagrange Multipliers

dx

dy

y

f

x

f.

x

g

x

f. 0..

dx

dy

y

g

y

f

dx

dy

y

g

x

g.. 0

Page 7: Lagrange Multipliers Erasmus Mobility Program (24Apr2012) Pollack Mihály Engineering Faculty (PMMK) University of Pécs João Miranda jlmiranda@estgp.pt

24-Apr-12 Lagrange Multipliers 7

• The multiplier is selected in conformity with,

and it is also necessary that:

(http://en.wikipedia.org/wiki/Lagrange_multiplier)

Lagrange Multipliers

0.

y

g

y

f

0.

x

g

x

f

Page 8: Lagrange Multipliers Erasmus Mobility Program (24Apr2012) Pollack Mihály Engineering Faculty (PMMK) University of Pécs João Miranda jlmiranda@estgp.pt

24-Apr-12 Lagrange Multipliers 8

• Thus, in optimal points, the three following equations are simultaneously verified (necessary conditions to the existence

of constrained optima):

• Notice that they represent the three partial derivatives in (x, y, ) of the function:

Verify it!

Lagrange Multipliers

yxgyxfyxL ,.,),,(

0.

x

g

x

f

0.

y

g

y

f

0),( yxg

Page 9: Lagrange Multipliers Erasmus Mobility Program (24Apr2012) Pollack Mihály Engineering Faculty (PMMK) University of Pécs João Miranda jlmiranda@estgp.pt

24-Apr-12 Lagrange Multipliers 9

• Procedure:1. Build the Lagrangean function, L(x,y,) ;

2. Set to zero the related first order partial derivatives;

3. Obtain the values (x, y, ) that are satisfying the system of equations.

Lagrange Multipliers

Page 10: Lagrange Multipliers Erasmus Mobility Program (24Apr2012) Pollack Mihály Engineering Faculty (PMMK) University of Pécs João Miranda jlmiranda@estgp.pt

24-Apr-12 Lagrange Multipliers 10

• Problem (general):

– Obtain the maxima/minima of the function,

– But the n variables have to simultaneously satisfy the set of m (m<n) equations:

Lagrange Multipliers

),...,,( 21 nxxxfu ),...,,( 21 nxxx

0),...,,(

(...)

0),...,,(

0),...,,(

21

2212

1211

mnm

n

n

bxxxg

bxxxg

bxxxg

Page 11: Lagrange Multipliers Erasmus Mobility Program (24Apr2012) Pollack Mihály Engineering Faculty (PMMK) University of Pécs João Miranda jlmiranda@estgp.pt

24-Apr-12 Lagrange Multipliers 11

• Procedure:1. Build the Lagrangean function, ;

Lagrange Multipliers

),...,,(),...,,,,...,,( 212121 nmn xxxfxxxL

12111 ),...,,(. bxxxg n

),( λxL

22122 ),...,,(. bxxxg n

mnm bxxxg ),...,,(.

...

211

Page 12: Lagrange Multipliers Erasmus Mobility Program (24Apr2012) Pollack Mihály Engineering Faculty (PMMK) University of Pécs João Miranda jlmiranda@estgp.pt

24-Apr-12 Lagrange Multipliers 12

• (...) Procedure:2. Set to zero the n first order partial derivatives in

the n variables ,

and...

Lagrange Multipliers

1x

f

mnmmnn

n

bxxxgbxxxgbxxxg

xxxfL

),...,,(.(...)),...,,(.),...,,(.

),...,,(),(

212212212111

21

λx

1

11. x

g

2

11

2

.x

g

x

f

0......

(...)

22

11

n

mm

nnn x

g

x

g

x

g

x

f

1

22.

x

g

0....1

x

gmm

0.....22

22

x

g

x

g mm

),...,,( 21 nxxx

Page 13: Lagrange Multipliers Erasmus Mobility Program (24Apr2012) Pollack Mihály Engineering Faculty (PMMK) University of Pécs João Miranda jlmiranda@estgp.pt

24-Apr-12 Lagrange Multipliers 13

• (...) Procedure:3. ... Notice that the first order partial derivatives in

order to the m multipliers is driving the m constraints,

Lagrange Multipliers

),...,,( 21 m

),...,,( 211 nxxxg

0),...,,(

(...)

21 mnm bxxxg

0),...,,( 2212 bxxxg n

01 b

mnmmnn

n

bxxxgbxxxgbxxxg

xxxfL

),...,,(.(...)),...,,(.),...,,(.

),...,,(),(

212212212111

21

λx

Page 14: Lagrange Multipliers Erasmus Mobility Program (24Apr2012) Pollack Mihály Engineering Faculty (PMMK) University of Pécs João Miranda jlmiranda@estgp.pt

24-Apr-12 Lagrange Multipliers 14

• (...) Procedure:3. Obtain the values set that simultaneously

satisfy the group of (n + m) equations. That is,

Lagrange Multipliers

),( λx

njx

g

x

f

x

L m

i j

ii

jj

,1,0)(

.)(),(

1

xxx λ

mibgL

iii

,1,0)(),(

xx λ

Page 15: Lagrange Multipliers Erasmus Mobility Program (24Apr2012) Pollack Mihály Engineering Faculty (PMMK) University of Pécs João Miranda jlmiranda@estgp.pt

24-Apr-12 Lagrange Multipliers 15

A furniture factory builds Tables (t) at a profit of 4 Euros per Table, and Chairs (c) at a profit of 3 Euros per Chair).

Suppose that only 8 short (s) pieces and 6 large (l) pieces are available for building purposes, what combination of Tables and Chairs

do you need to build to make the most profit?

.

If the availability of the short pieces is 8008 and the availability of the large pieces is 6007,

how many Tables and Chairs do you need to build to make the most profit?

Tables & Chairs (T&C)

Page 16: Lagrange Multipliers Erasmus Mobility Program (24Apr2012) Pollack Mihály Engineering Faculty (PMMK) University of Pécs João Miranda jlmiranda@estgp.pt

24-Apr-12 Lagrange Multipliers 16

Note: modeling is based in the proportionality, aditivity and divisibility between the produced quantities of t and c

and profit and utilization of l and s components.

.

Tables & Chairs (T&C)

,subject to

ProfitTotalMaximize

tables

large

produce to

used

components

tables

short

produce to

used

components

chairs

large

produce to

used

components

components

ofty Availabili

large

tables

short

produce to

used

components

components

oftyAvailabili

short

Page 17: Lagrange Multipliers Erasmus Mobility Program (24Apr2012) Pollack Mihály Engineering Faculty (PMMK) University of Pécs João Miranda jlmiranda@estgp.pt

24-Apr-12 Lagrange Multipliers 17

Note: modeling is based in the proportionality, aditivity and divisibility between the produced quantities of t and c

and profit and utilization of l and s components.

Tables & Chairs (T&C)

Prftmax

0,

,

ct

tosubject

600712 ct

800822 ct

ct 34

Page 18: Lagrange Multipliers Erasmus Mobility Program (24Apr2012) Pollack Mihály Engineering Faculty (PMMK) University of Pécs João Miranda jlmiranda@estgp.pt

24-Apr-12 Lagrange Multipliers 18

T&C: Lagrange Multipliers

Maximize the profit function, Luc(t,c), satisfying the conditions

concerning the availability of the l and s components..

• Problem:

– Obtain the maximum value of profit function,

– But the variables t and c shall satisfy the availability of l and s:

ctPrft 34

800822

600712

ct

ct

Page 19: Lagrange Multipliers Erasmus Mobility Program (24Apr2012) Pollack Mihály Engineering Faculty (PMMK) University of Pécs João Miranda jlmiranda@estgp.pt

24-Apr-12 Lagrange Multipliers 19

• Procedure:1. Build the Lagrangean function, L(t, c, 1, 2) ;

2. Set to zero the 4 first order partial derivatives;

3. Obtain the values (t, c, 1, 2) that are satisfying the system of equations.

T&C: Lagrange Multipliers

Page 20: Lagrange Multipliers Erasmus Mobility Program (24Apr2012) Pollack Mihály Engineering Faculty (PMMK) University of Pécs João Miranda jlmiranda@estgp.pt

24-Apr-12 Lagrange Multipliers 20

• Procedure:1. Build the Lagrangean function, L(t, c, 1, 2) ;

T&C: Lagrange Multipliers

ct 34

600712.1 ct

),(.),(.),(),,,( 2121 ctsctlctPrftctL

800822.2 ct

Page 21: Lagrange Multipliers Erasmus Mobility Program (24Apr2012) Pollack Mihály Engineering Faculty (PMMK) University of Pécs João Miranda jlmiranda@estgp.pt

24-Apr-12 Lagrange Multipliers 21

• (...) Procedure:2. Set to zero the 4 first order partial derivatives, and

notice that those related to the multipliers (1, 2) are driving the original constraints:

and...

T&C: Lagrange Multipliers

0224 21

0

0

0

0

2

1

L

Lc

Lt

L

ct 34

600712.1 ct

),,,( 21 ctL

800822.2 ct

0213 21

0600712 ct

0800822 ct

Page 22: Lagrange Multipliers Erasmus Mobility Program (24Apr2012) Pollack Mihály Engineering Faculty (PMMK) University of Pécs João Miranda jlmiranda@estgp.pt

24-Apr-12 Lagrange Multipliers 22

• (...) Procedure:3. Obtain the values (t, c, 1, 2) that are

simultaneously satisfying the 4 equations:

• Re-allocating the terms of the system of equations,

T&C: Lagrange Multipliers

422 21 321 21

600712 ct

800822 ct

0224 21

0213 21

0600712 ct

0800822 ct

Page 23: Lagrange Multipliers Erasmus Mobility Program (24Apr2012) Pollack Mihály Engineering Faculty (PMMK) University of Pécs João Miranda jlmiranda@estgp.pt

24-Apr-12 Lagrange Multipliers 23

• (...) Procedure:3. Obtain the values (t, c, 1, 2) that are

simultaneously satisfying the 4 equations:

• Applying the Cramer’s Rule to the 1.st subsystem,

T&C: Lagrange Multipliers

21

22

23

24

1

321

422

21

21

2.12.2

2.32.41

2.12.2

1.43.22

12

2

24

681

12

2

24

462

21

22

31

42

2

Page 24: Lagrange Multipliers Erasmus Mobility Program (24Apr2012) Pollack Mihály Engineering Faculty (PMMK) University of Pécs João Miranda jlmiranda@estgp.pt

24-Apr-12 Lagrange Multipliers 24

• (...) Procedure:3. Obtain the values (t, c, 1, 2) that are

simultaneously satisfying the 4 equations:

• Applying the Cramer’s Rule to the 2.nd subsystem,

T&C: Lagrange Multipliers

22

12

28008

16007

t

800822

600712

ct

ct

1.22.2

1).8008(2).6007(t

1.22.2

)6007.(2)8008.(2

c

20032

4006

24

800812014t

20012

4002

24

1201416016

c

22

12

80082

60072

c

Page 25: Lagrange Multipliers Erasmus Mobility Program (24Apr2012) Pollack Mihály Engineering Faculty (PMMK) University of Pécs João Miranda jlmiranda@estgp.pt

24-Apr-12 Lagrange Multipliers 25

• (...) Procedure:3. Obtain the values (t, c, 1, 2) that are

simultaneously satisfying the 4 equations :

• Then,

Optimal Solution!

T&C: Lagrange Multipliers

2001

2003

1

1

2

1

c

t

Edit LINDO:max 4t + 3csubject tol1) 2t + 1c <= 6007l2) 2t + 2c <= 8008END

Page 26: Lagrange Multipliers Erasmus Mobility Program (24Apr2012) Pollack Mihály Engineering Faculty (PMMK) University of Pécs João Miranda jlmiranda@estgp.pt

24-Apr-12 Lagrange Multipliers 26

Lagrange Multipliers (synthesis)

• Introduction: maximization of 2-variable function subject to one restriction;

• Generalization: maximization of n-variable function subject to m restrictions

• Tables & Chairs problem.