0% found this document useful (0 votes)
40 views6 pages

Assignment 3

This document contains solutions to two problems related to linear programming. For problem 1, it is proven by contradiction that the optimal value of the objective function is unbounded. It is shown that for any finite optimal value, a better feasible solution can be constructed, resulting in a contradiction. For problem 2, the feasible set and conditions for a subgradient are defined. It is shown that the subgradient takes the form of the dual variables from the KKT conditions of the dual problem.

Uploaded by

Vijyan Mahajan
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
0% found this document useful (0 votes)
40 views6 pages

Assignment 3

This document contains solutions to two problems related to linear programming. For problem 1, it is proven by contradiction that the optimal value of the objective function is unbounded. It is shown that for any finite optimal value, a better feasible solution can be constructed, resulting in a contradiction. For problem 2, the feasible set and conditions for a subgradient are defined. It is shown that the subgradient takes the form of the dual variables from the KKT conditions of the dual problem.

Uploaded by

Vijyan Mahajan
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/ 6

r Prr P

sst
r

Pr

r ts trt t t t t t t R rs

x1 = x1 x2 = x2 f (x) = x1 x1 = (x1 , x2 ) s s

2 x2



2
sr > 0 2 x2 ( ) x2 + 2 2 rr x1 = x2 = x2 + 2 2 s
s s st r t t t t s > 0 < r x1 =
x2 = x2 + 2 2 t t t t s ss t s s t trt t
2

(x1 ) x2 0

t t t t

t t s t x = [x1 , x2 ]T t s rt d = [d1 , d2 ]T r trst s tt


t t t r

min

st t

x,d () := x1 + d1
2

(x1 + d1 ) (x2 + d2 ) 0
0

s t r ts t s t s tt s t rst str t tt r rtt


s


() = d21 d22 2 + 2 (x1 d1 x2 d2 ) + x21 x22 0

( r1 ) ( r2 ) 0

t t rts t qrt qt () = 0 = r1 = r2 t tt s x1 x2 d1 d2 r t r1
r2 r t qt rtt s
0 st s sts qt s s
max (0 , min (r1 , r2 )) max (0 , max (r1 , r2 ))
|
|
{z
}
{z
}

sr tt l u qt s s s s l u r t s r1 r2 r t
s t s s t r

Pr
t s st t tt rst r s
M = {x Rn : Ax = b}

r Rn t srt f (x) t x = x t s t sts t qt


f (x) f (x ) + T (x x ) ; x Rn

r x = x t t t st s t sts t qt

T (x x ) 0 ; x M

r t t r s r tt sts qt s t strtr

= AT z

r z Rm = AT y r s y Rm t qt sts t s t sts
qt sr tt
AT y s + w = 0

AT y = (s w )

qt rsts r t rst str t r z = y t = (s w ) t


s t r t

= (s w ) s srt
Pr s tr sss

f (x) t x = x

sT (x + t )

(x + t )

sr t
f (x ) + T (x x )

eT t + T (x x )
T

e t + (s w ) (x x )

eT t + (s + w ) x + (s w ) x

eT t sT t wT t + (s w ) x

(e s w ) t + (s w ) x

(s w ) x

qt rsts r t t tt Pr Pr r qt f (x ) = eT t qt
s t s qts qt rsts r t s str t r
t r s
f (x) =

n
X

|xi | (s w ) x

i=1

t t ith t (s w ) i i = si wi si + wi = 1 t str t r
si , wi 0 tt 1 i = (si wi ) 1 qt s qt t
|x1 | + + |xi | + + |xn | 1 x1 + + i xi + + n xn

1 i 1 |xi | i xi rr qt qt s tr ts s
t r

Pr
P t r t r s t ts sss sst t r t
s s s

r st t r t
tt t t rst s t ts sss sst t rst s s s

r st t rst
sr t rst r t t t t s 10.2222 rrs t st

T

T
s x = 4.88889 0 0 5.33333 rqr srt s = s w = 1 0.148148 0.333333 1

Pr
tt r sr ts r s
min

st t

f (x) := 2x1 x2 + x3 x4 + x21


x21 x3
x22 x4

Ps t tt r t ttr r tt min f (x) > 1 t t t


min f (x) f (x) > 1 t r tt min f (x) > 1 r t t st
x3 = x21 x4 = x22 sttt ts t t f (x) s t

f (x)

=
=
=

2x1 x2 + x21 x22 + x21



x21 x22 2x1 x2 + 1 + x21 1
2

(x1 x2 1) + x21 1

sr tt r x1 , x2 R (x1 x2 1)2 + x21 > 0 s s s (x1 x2 1)2 + x21 = 0 t x1 = 0 x1 x2 = 1


tt x2 = r x2 = s t t R rr t tt r s t tts ts

t st f (x) = 1 s min f (x) > 1 s s t r r


t t xb sts f (b
x) 1 + 2 r > 0 r ts t s t x
b sts f (b
x) = 1 + 2
2
s qt qt 1 +

(b
x1 x
b2 1) + x
b21 1 = 1 + 2

ss st t qt s xb1 =
> 0 x
bi R , i

xb2 =

b3 = x
b21 =
+ 1 x

(b
x1 x
b2 1) + x
b21 = 2

2
2

xb4 = xb22 =

+1

2

Pr
K K Rn s s
r ts r



K = y Rn : y T x 0 ; x K
K := {x Rn : 0 x1 x2 . . . xn }

1.0

x2
0.5

Line 1

Line 2

Line 4

Line 3

0.0

x1

-0.5

-1.0
-1.0

-0.5

0.0

0.5

1.0

r Pt s t K r t K r r n = 2
strt r t s n = 2 K s t r r trs t rt
t r ts s t s s t rtr r t y y T x 0 r x t
r t t s t r t t rr t
rst rr K s t r r t K
rtr t qt


 
0 1 x1
0
1 1 x2

tt t r n = 2 t

K qt s


K = s Rn : B T s 0

0
0

B = BT = .
..

0
1

0
0
0
..
.

.
..

1
1

0 0 1
0 1 1

1 1 1

.. .. ..
. . .

1 1 1
1 1 1

Pr s qt t r t t s
s Rn s t sts t qt B T s 0 t x K s tt sT x < 0
tt s Rn s t sts t kth qt B T s 0


Pr

n
X

si < 0

i=nk+1

t t x K s tt sT x < 0 sr tt qt qt rtt s

s T xo < 0

r xo =

1
|{z}

1
|{z}

nth t

(nk+1)th t

sr tt xo K ts s t r

s Rn stss qt
B T s 0 t sT x 0 ; x K

r s = s1 sn sts B T s 0 x = x1


Pr

0
|{z}

1st t


xn K t t r tt

s 1 x1 + s 2 x2 + s 3 x3 + + s n xn 0

ts sr tr = 1 n r i 0 : i t t ith r qt B T s 0 t i
t t rs s t t qt
s1 (1 ) + s2 (1 + 2 ) + s3 (1 + 2 + 3 ) + + sn (1 + 2 + 3 + + n ) 0

rr B s 0 i 0 : i t t qt tr r qt
t t t st t t qts
T

x1

1 + 2

x2

1 + 2 + 3
..
.

=
..
.

x3
..
.

1 + 2 + 3 + + n

x3

s t i s r t t r tt qt s tr st t t st r
qts s 1 = x1 2 = x2 x1 3 = x3 x2 n = xn xn1 x K x1 0
xi xi1 ; i = 2, . . . , n i 0 : i s s t r

Pr
tt r sr ts r s
min

st

1
q (x) := cT x + xT Qx
2
Ax b

sr t st ts x + d r 0
x s s t r t tt r Ax b Ad 0 t x + d s s
t r 0
Pr

A (x + d) = Ax + Ad b + 0 = b

s ts t r
dT Qd 0 (c + Qx)T d < 0 t q (x + d) s
sr t

q (x + d)

=
=
=

1
T
(x + d) Q (x + d)
2

1 T
cT x + cT d +
x Qx + 2dT Qx + 2 dT Qd
2


2
1
T
cT x + xT Qx + dT Qd + (c + Qx) d
{z
}
|
2
2
{z
} | nd{z }
|
3rd tr
cT (x + d) +

1st tr

tr

1st tr qt s t dT Qd 0 t 2nd tr 2 dT Qd s t rs t
strt rs r 0 (c + Qx)T d < 0 t 3rd tr (c + Qx)T d s strt t rs r
0 t t s q (x + d) s s ts t r
2

r rqr r ts r s rt sq

You might also like