Q HOHJHUHD GHSOLQ  D FXUVXOXL SUHVXSXQH

DD GXS

 FXP V D DU WDW vQF  GH OD

nceput-LFXQRWLQ HFRQH[HQWUHDFHVWHDFHOH PDWHPDWLFHRFXS

XQORFLPSRUWDQW

$1(;$ FXSULQGH R FROHF LH GH HQXQ XUL PDWHPDWLFH FDUH DMXW  SH FLWLWRU V

clarifice unHOH GLQ SUREOHPHOH vQWkOQLWH I

-L

U  D SXWHD V  VH VXEVWLWXLH SDUFXUJHULL

OXFU ULORUPDWHPDWLFHGHED] 

1. 0XO
S-a folosit adesea termenul de

LPL

PXO LPH PXO LPHD DJHQ LORU GLQ HFRQRPLH

PXO LPHD EXQXULORU PXO LPHD GH FRQVXP D FRQVXPDWRUXOXL PXO LPHD GH SURGXF LH

etc.).
&RQFHSWXO GH PXO LPH HVWH XQ FRQFHSW SULPDU 2 FROHF LH GH RELHFWH GH RULFH IHO
HVWH R PXO LPH 2ELHFWHOH FDUH IRUPHD]  PXO LPHD VH QXPHVF

elemente ale acesteia.

2PXO LPHHVWHILQLW GDF HDFRQ LQHPHOHPHQWHPILLQGXQQXP URDUHFDUH([LVW 

LPXO LPLLQILQLWH3HQWUXDSXWHDVSXQHGHVSUHXQHOHPHQWF HVWHVDXQXvQPXO LPH

QHILLQG SRVLELO vQ DFHVW FD] V

 VH HQXPHUH WRDWH HOHPHQWHOH PXO LPLL  WUHEXLH GDW R

UHJXO FDUHGHILQHWHRSURSULHWDWHSHFDUHRDXHOHPHQWHOHPXO LPLLUHVSHFWLYHLQXR

DXHOHPHQWHOHFHQXIDFSDUWHGLQHD)LLQGGDWHGRX PXO LPLUDSRUWXOGLQWUHHOHHVWH
GHWHUPLQDW vQ vQWUHJLPH GH SURSULHWDWHD GH DSDUWHQHQ

 6H SRW GHILQL DVWIHO RSHUD LLOH

IXQGDPHQWDOHSULYLQGPXO LPLOHUHXQLXQHLQWHUVHF LHFRPSOHPHQWDU H

tc.).

'H IRDUWH PXOWH RUL HFRQRPLVWXO DUH QHYRLH V  RUGRQH]H HOHPHQWHOHXQHL PXO LPL
'H H[HPSOX SUHIHULQ HOH FRQVXPDWRUXOXL vQWUH GLYHUVHOH FRPSOH[H GH EXQXUL FDUH
FRPSXQ PXO LPHD VD GH FRQVXP SURGXF R RUGRQDUH D DFHVWRUD SHQWUX D vQ HOHJH FXP

alege consXPDWRUXO FDQWLW

LOH GH EXQXUL SH FDUH GRUHWH V  OH FRQVXPH" 7UHEXLH

LQWURGXV QR LXQHDGHUHOD LHvQWUHHOHPHQWHOHXQHLPXO LPL

1.1.

5HOD LHELQDU

Fie X R PXO LPH FDUH FXSULQGH HOHPHQWHOH x1, x2, x3  HWF L P R UHOD LH vQWUH
aceste elemente. P este o reOD LH ELQDU
xi , x j X

 GDF  SHQWUX RULFH SHUHFKH GH HOHPHQWH

 RUGLQHD HOHPHQWHORU ILLQG GHFLVLY 

xi HVWH SULPXO HOHPHQW DILUPD LD xiPxj

HVWHILHDGHY UDW ILHIDOV 

Cvasiordonare (preordonare)
2 UHOD LH ELQDU 
SURSULHW

P produce o cvasiordine (SUHRUGLQH SH PXO LPHD X GDF

 H[LVW 

LOH

a) reflexivitate: xiPxi pentru orice xi X ;

b) tranzitivitate: xiPxjLxjPxkLPSOLF xiPxk.
Ordonare

1 .1 .

2FYDVLRUGLQHHVWHRRUGLQHGDF 

reale eVWHRUGRQDW

GHUHOD LDELQDU 

xiPxjLxjPxiLPSOLF xi=xj0XO LPHDQXPHUHORU

, deoarece x yLy xLPSOLF x = y.

3UHIHULQ HOH FRQVXPDWRULORUSURGXFQXPDL R FYDVLRUGLQHSH PXO LPHDGH FRQVXP

i x1 L x2 F

3XWHP VSXQH GHVSUH GRX  FRPSOH[H GH EXQXU

SUHIHUDWFDL

x2LF

x1 HVWH FHO SX LQ OD IHO GH

x2HVWHFHOSX LQOHIHOGHSUHIHUDWF

L

x1I

U FD

x1L x2V

fie identice.
Cnd xiPxj L xjPxi VXQW DPEHOH DGHY

UDWH GDU

xi x j VH VSXQH F

xj

xi sunt

echivalente n cadrul preordinii date de P pe X.

2 RUGLQHVDXRSUHRUGLQHGHILQLW SH R PXO LPH ; HVWH FRPSOHW

GDF DYHP

xiPxj

sau xjPxi sau ambele, pentru orice xi , x j X . n caz contrar, ordinea sau preordinea
este

SDU LDO

 3XWHP GHILQL R UHOD LH GH SUHIHULQ

 VWULFW  QRWDW 

PS SH PXO LPHD GH

FRQVXPDXQXLFRQVXPDWRUFDUHHYLGHQWHVWHPDLSXWHUQLF GHFkWSUHIHULQ DQHVWULFW 

FHO SX LQ OD IHO GH SUHIHUDW FD QRWDW 

PN GDU UHOD LD PS nu are proprietatea de

UHIOH[LYLWDWH L GHFL QX SXWHP FRQVWUXL RUGRQ UL VDX SUHRUGRQ UL 6  REVHUY P F 

x1PSx2DUHORFGDF

LQXPDLGDF 

x1PNx2HVWHDGHY

UDW LDU

x2PNx1HVWHIDOV

5HOD LDGHHFKLYDOHQ

FDUHvQVHDPQ F VXQWDGHY UDWHDWkW

xiPxjFkWLxjPxi) este

R UHOD LH ELQDU  GDU QX FRQGXFH GHFkW OD R SUHRUGLQH SDU LDO  (VWH DYDQWDMRV SULQ
XUPDUH DWXQFL FkQG GLVFXW P SUHIHULQ HOH FRQVXPDWRULORU V  IRORVLP UHOD LD
UHOD LD GH HFKLYDOHQ
HVWH DGHY UDW F 

 QRWDW   LQGLIHUHQW 6FULLQG F 

x1PSx2, atunci PN este o rela LH ELQDU

PXO LPHD GH FRQVXP L GHILQHWH UHOD LD

H[LVWHQ DVLPXOWDQ D

x1PNx2 SHQWUX D DU


PS L

WD F  QX

PN produce o preordine pe

I FD R UHOD LH GH HFKLYDOHQ

 LPSOLFDW  GH

x1PNx2Lx2PNx1.

1 .1 .1 . Produsul cartezian
3URGXVXO FDUWH]LDQ DO GRX  PXO LPL

X L Y  VH QRWHD]

X Y

L HVWH PXO LPHD

perechilor (x,y) astfel nct x X L y Y . Putem forma produsul cartezian al unei

IDPLOLL GH PXO LPL 1RW P

X 1 X 2 ... X r = X i  (OHPHQWHOH XQHL DVWIHO GH PXO

LPL

i =1

(x1, x2,,xr) se numesc r-tupliLVXQWDQDORDJHSHUHFKLORUGDUFXSULQGUHOHPHQWH$O

i-lea membru dintr-un r-tuplu este un element al celei de-a i-DPXO LPHLVHQXPHWH
component sau

FRRUGRQDW

 (OHPHQWHOH PXO LPLL SURGXV FDUWH]LDQ VH QXPHVF

puncte sau vectori.

Adesea se foloseWHQR LXQHD GHVSD LX. Ea are mai multe sensuri. Atunci cnd
R PXO LPH GHYLQH DWkW GH FXSULQ] WRDUH vQFkW FRQ LQH WRDWH HOHPHQWHOH GDWH vQWU

-un

DQXPLW FRQWH[W HD VH QXPHWH PXO LPH XQLYHUVDO  &kQG R PXO LPH XQLYHUVDO  DUH
XQHOH SURSULHW

L JHRPHWULFH R QXPLP VSD LX 6SD LXO vQWkOQLW IUHFYHQW vQ FDGUXO

FXUVXOXL HVWH VSD LXO UHDO Q

PXO LPL
SURSULHW

-dimensional, notat n , care este produsul cartezian a n

... , unde HVWH PXO

LPHD QXPHUHORU UHDOH 'DF 

L  SULYLQG UHOD LLOH GH GLVWDQ

GLPHQVLRQDO 8Q HOHPHQW DO VSD LXOXL

 XQHOH

 HO HVWH GHQXPLW VSD LXO HXFOLGLDQ Q

n este un n-tuplu numit punct sau vector,

fiecare dintre coordonatele sale fiind uQQXP

1 .1 .2 .

n  SRVHG

UUHDO

0XO LPHFRQYH[

0XO LPHD

X n  VH QXPHWH FRQYH[

 GDF  RGDW  FX GRX  SXQFWH HD FRQ LQH L

VHJPHQWXOFDUHOHXQHWH6HQXPHWHVHJPHQWFHXQHWHSXQFWHOH

puncte x de forma {x n x1 + (1 )x 2 ,0 1}.

x1Lx2PXO LPHDGH

'DF 

X HVWH R PXO LPH FRQYH[

0 1 . Fie x3

0 1 

XQ DOW SXQFW GLQ ; &RQIRUP GHILQL LHL

1 = ,

2 = (1 ) 

t + (1 )x3 X pentru

x1 + (1 )x2 + (1 )x3 X

QORFXLQG SH W UH]XOW 

Notnd

x1 , x2 X , atunci t = x1 + (1 )x2 X pentru

 L

L

3 = (1 ) 

0 1 , 2 , 3 1 

RE LQHP

1 + 2 + 3 = + (1 ) + (1 ) = 1 , deci 1 x1 + 2 x2 + 3 x3 X 
x1 , x 2 ,..., x r VXQW

pentru 0 , 1 .

SXQFWH DOH PXO LPLL FRQYH[H ; SXQFWXO

*HQHUDOL]kQG

L

GDF 

x = i x i , cu 0 i 1 ,
i =1

i =1

= 1  HVWH WRW vQ ; 5H]XOW

 F  R FRPELQD LH FRQYH[  D RULF UXL QXP U GH SXQFWH

din n HVWHRPXO LPHFRQYH[

LDULQWHUVHF LDPXO LPLORUFRQYH[HHVWHWRWRPXO LPH

FRQYH[ 

0XO LPLvQFKLVH
6HVSXQHF XQVSD LX6HVWH
XQQXP UUHDOQHQHJDWLYQRWDW

metricGDF

SHQWUXILHFDUHGRX SXQFWH

x, y S H[LVW

d(x,y) numit PHWULF sau IXQF LHGHGLVWDQ astfel

nct:
a) d(x,y)=0GDF

LQXPDLGDF 

x = y;

b) d(x,y)= d(y,x);
c) d(x,y)+d(y,z) d(x,z), x,y,z S (inegalitatea triunghiului).
1XPLP YHFLQ WDWH D XQXL SXQFW

d (x, y ) < , unde HVWHXQQXP

2 PXO LPH

LPHD WXWXURU S

unctelor y pentru care

USR]LWLYFRQVLGHUDWPLF

X S  HVWHGHVFKLV

un VXILFLHQWGHPLFFXSULQV

x S  PXO

GDF  ILHFDUHSXQFW

vQvQWUHJLPH

x X

 DUH R YHFLQ WDWH SHQWUX

n X. Un punct z X

HVWHXQSXQFWOLPLW 

VDX SXQFW GH DFXPXODUH DO PXO LPLL ; GDF  ILHFDUH YHFLQ WDWH D OXL ] FXSULQGH FHO
SX LQXQSXQFW FDUH DSDU LQH PXO LPLL ; DOWXOGHFkW]QRW P FX

punctHORUOLPLW
2PXO LPH

DOHPXO LPLL;0XO LPHD

X S este vQFKLV

XX

 PXO LPHD WXWXURU

VHQXPHWHvQFKLGHUHDOXL;

GDF FXSULQGHWRDWHSXQFWHOHVDOHOLPLW 

2PXO LPHFXSURSULHWDWHDF WRDWHSXQFWHOHHLVXQWLSXQFWHOLPLW HVWHR

FRPSDFW

mXO LPH

2 PXO LPH HVWH P

UJLQLW

 GDF  GLVWDQ D GLQWUH RULFDUH SHUHFKL GH SXQFWH QX HVWH

PDLPDUHGHFkWXQQXP UILQLWRDUHFDUH

)RUPHS WUDWLFH

Un polinom omogen de gradul 2 n n variabile poate fi reprezentat sub forma

n

f (x ) = aij xi x j , unde x = (x1 , x2 ,..., xn )  'DF

i =1 j =1

vQFkW FRHILFLHQ LL WHUPHQLORU

aij a ji , atunci aij = a ji =

1
2

(a

ij

+ a ji ) astfel

xixj L xjxi sunt identici. Expresia f(x) HVWH R IRUP

S WUDWLF  3ULQ WUDQVIRUPDUHD DQWHULRDU  PDWULFHD FRHILFLHQ LORU IRUPHL S

QRWDW  $ HVWH R PDWULFH VLPHWULF  L IRUPD S WUDWLF 

unde xTHVWHYHFWRUXOFRORDQ

)RUPD S WUDWLF 

tratice,

f(x) se poate scrie: f (x ) = xAxT ,

x transpus.

f(x) HVWH SR]LWLY QHJDWLY GHILQLW

 GDF 

f (x ) > 0 (< 0 ) pentru orice

x 0.
)RUPD S WUDWLF 

f(x HVWH QHQHJDWLY QHSR]LWLY GHILQLW

 GDF 

f(x) 0 ( 0) pentru

orice x 0 .
)RUPD S WUDWLF 

f(x) QHQHJDWLY QHSR]LWLY GHILQLW

f (x ) > 0 (< 0)LDUSHQWUXDO

LL

f(x)=0VHVSXQHF

 FDUH SHQWUX XQLL

x 0 este

HVWHSR]LWLYQHJDWLYVHPLGHILQLW 

&XPRULFHPDWULFHVLPHWULF UHDO SRDWHILPDWULFHDXQHLIRUPHS WUDWLFHWHUPHQLL

SR]LWLY QHJDWLY GHILQLW  L FHLODO L SRW IL WUDQVIHUD L DVXSUD PDWULFLL IRUPHL
S WUDWLFH
3UREOHPD GHWHUPLQ ULL VHPQXOXL XQHL PDWULFL LQWHUYLQH vQ VWDELOLUHD FRQGL LLORU
GH RSWLPDOLWDWH GH RUGLQXO ,, SHQWUX R SUREOHP  GH H[WUHPXP DO IXQF LLORU GH PDL
PXOWH YDULDELOH 5HFXQRDWHUHD WLSXOXL GH H[WUHPXP PD[L

m sau minim) se face n

IXQF LH GH VHPQXO PLQRULORU SULQFLSDOL DL XQHL PDWULFH (VWH XWLO  GHFL XUP WRDUHD
UHJXO 

2 PDWULFH HVWH SR]LWLY QHJDWLY GHILQLW  GDF  L QXPDL GDF  PLQRULL SULQFLSDOL
SH P VXUD FUHWHULL VXFFHVLYH D RUGLQXOXL DX DFHOHDL V

emne (au semne alternative).

&RQGL LDWUHEXLHvQGHSOLQLW SHQWUXWRDWHVHULLOHSRVLELOHGHPLQRULSULQFLSDOL

)XQF LL

6H SUHVXSXQ FXQRVFXWH SURSULHW

6XQWGDWHDLFLGRDUXQHOHSURSULHW
IXQF LLFXIRUPDJHQHUDO 

QRW P

LXWLOHSHQWUXIXQF LLGHPDLPXOWHYDULDELOHDGLF 

f(x), unde x este un vector cu n componente, care descriu o

UHJXO  D XQHL DSOLFD LL GLQ

convenabil V

LOH HOHPHQWDUH DOH XQHL IXQF LL GH R YDULDELO 

n n  'DF

 FRQVLGHU P QXPHUHOH

y=F(x) UHOD LH LQWHUSUHWDW

 H[LVW 

m IXQF LL GH IRUPD fi(x), este

yi= fi(x) GUHSW FRPSRQHQWH DOH XQXL YHFWRU L V

 FD R IRUP  SUHVFXUWDW  SHQWUX FHOH P UHOD LL

yi=

fi(x).
F(x) HVWH DVWIHO R IXQF LH YHFWRULDO

 LDU

fi(x) este a I-D FRPSRQHQW

 D YHFWRUXOXL

F(x).
$VHP Q WRU GHILQL LHL GHULYDWHL XQHL IXQF LL GH R YDULDELO  VH GHILQHWH GHULYDWD
SDU LDO  D OXL

aceasta cu

f(x) ID

 GH R FRPSRQHQW  D YHFWRUXOXL

x, de exemplu xk L VH QRWHD]

f (x )
 'DF  QX H[LVW  FRQIX]LL VH IRORVHVF L DOWH QRWD L
x k

i pentru derivata

SDU LDO  'H H[HPSOX

fk

HVWH R DVWIHO GH QRWD LH 2 IXQF LH

f(x) unde x are n

componente, are nGHULYDWHSDU LDOHSXWkQGILHOHvQVHOHFRQVLGHUDWHGUHSWFRPSRQHQWH

ale unui vector, numit vector gradient DOIXQF LHLIFDUHVHQRWHD]
grad f = ( f1 , f 2 ,..., f n ) =
'DF 

f(x)HVWHRIXQF LHFRQWLQX

SXWHP FDOFXOD L SHQWUX

fk

f (x ) f (x )
x1
x 2

,..., f x(nx )

JU

ad f sau f .

LDUHGHULYDWHOHSDU LDOHGHRUGLQXOXQXFRQWLQXH

GHULYDWHOH SDU LDOH ID

 GH RULFDUH GLQ FRPSRQHQWHOH

vectorului x2E LQHPDVWIHOGHULYDWHOHSDU LDOHGHRUGLQXOGRL

f (x ) = 2 f (x ) = f .
rk
xr x
k xr xk
([LVW 

n2 GHULYDWH SDU LDOH GH RUGLQXO GRL L ILHFDUH DYkQG GRL LQGLFL SRDWH IL

FRQVLGHUDW FDHOHPHQWDOXQHLPDWULFL

H(x)=(frk(x))
H(x)HVWHPDWULFHDKHVVLDQ

FX

nOLQLLLn coloane.

'DF 

f(x) HVWHFRQWLQX

HVWHRIXQF LHGHFODV

LDUHGHULYDWHGHRUGLQXOXQXLGRLFRQWLQXHVHVSXQHF 

& QPRGHOHOHHFRQRPLFHIXQF LLOHXWLOL]DWHIUHFYHQWIXQF LD

GHXWLOLWDWHIXQF LDGHSURGXF LHHWFVHSUHVXSXQHF DWLQJDFHVWJUDGGHFRQWLQXLWDWH

f (x ) C 2 , atunci

WHRUHP  LPSRUWDQW  GLQ DQDOL]D PDWHPDWLF  VWDELOHWH F  GDF 

f rk = f kr  $SDUH DVWIHO R SURSULHWDWH LPSRUWDQW

VLPHWULF

 PDWULFHD KHVVLDQ

 + HVWH R PDWULFH

Q PRGHOHOH HFRQRPLFH VH IRORVHWH L GLIHUHQ LDOD XQHL IXQF LL SHQWUX D P VXUD
DSUR[LPDWLYPRGLILFDUHDIXQF LHL

f(x)FDXUPDUHDXQHLPLFLPRGLILF

GLUHF LHD DUJXPHQWHORU IXQF LHL 'LIHUHQ LDODXQHLIXQF LL

G[L

ULvQWU RDQXPLW 

f(x) HVWH R IXQF LHOLQLDU

 vQ

df = f dx = f k dxk .
k =1

Majoritatea modelelor economice conduc, din punct de vedere matematic, la

DIODUHD XQXL H[WUHPXP DO XQHL IXQF LL GH PDL PXOWH YDULDELOH SH R PXO LPH FRQYH[ 
([LVWHQ DXQHLDVWIHOGHVROX LLHVWHGDW GHRWHRUHP IXQGDPHQWDO 

'DF 

f (x ) C 2 , ea are un maxim (miQLP SH PXO

LQWHULRU PXO LPLL

X GDF

 L QXPDL GDF 

SR]LWLYVHPLGHILQLW SH

LPHD FRQYH[ 

( )

f x o = 0  L PDWULFHDKHVVLDQ

X n punctul x0
 + HVWH QHJDWLY

X.

)XQF LLFRQYH[HLFRQFDYH

)XQF LD

f(x)GHILQLW

SHPXO LPHDFRQYH[ 

X n HVWHFRQYH[

GDF SHQWUXRULFH

x1 , x 2 X :

( )

( )

f x1 + (1 )x 2 f x1 + (1 ) f x 2 , cu 0 1 .
'DF  LQHJDOLWDWHD HVWH VWULFW  L DGHY UDW  SHQWUX RULFH

x1 x 2  L 0 < < 1  IXQF

LD

HVWHVWULFWFRQYH[ 

FXQF LD f(x)

HVWH FRQFDY  VWULFW FRQFDY  GDF 

(-f(x))

HVWH FRQYH[  VWULFW

FRQYH[ 
'LQWUHSURSULHW

LOHPDLLPSRUWDQWHDOHIXQF LLORUFRQYH[HLFRQFDYHVXQW

a) RULFHFRPELQD LHOLQLDU

SR]LWLY DXQRUIXQF LLFRQYH[HFRQFDYHHVWHFRQYH[ 

FRQFDY 

b) dac  f(x) HVWH FRQYH[

 PXO LPHD

HVWH FRQFDY  PXO LPHD

S = x f ( x) f ( x o )

 HVWH FRQYH[  'DF 

S = x f ( x) f ( x o )

 HVWH FRQYH[  XQGH

f(x)

xo este orice

punct din S;
c) IXQF LD f(x)

HVWH

( ) ( )

FRQYH[ 

( )(

GDF 

L

QXPDL

GDF 

SHQWUX

RULFH

x1 , x 2 X :

f x1 f x 2 f x 2 x1 x 2 .
,QHJDOLWDWHD LQYHUV  GHILQHWH FRQFDYLWDWHD OXL I[ LDU LQHJDOLW

LOH VWULFWH

GHILQHVFFRQYH[LWDWHDFRQFDYLWDWHDVWULFW 

d) GDF

f(x)

HVWH

FRQYH[ 

VHPLGHILQLW  'DF 

FRQFDY 

KHVVLDQD

f(x) este strict cRQYH[

VD

HVWH

SR]LWLY

QHJDWLY

 VWULFW FRQFDY  KHVVLDQD VD HVWH

SR]LWLYQHJDWLYGHILQLW 

e) GDF

xo este un punct critic al

R IXQF LH FRQYH[ FRQFDY  DUHXQSXQFW FULWLF 

f(x)GDF  f (x o ) = 0 ), ea are un minim (maxim) n acel punct.

IXQF LHL

a.

([WUHPXPDOIXQF LHL

7HUPHQXOUHXQHWHQR LXQLOHGHPD[LPLGHPLQLPDOXQHLIXQF LL)LHIXQF LD

GHILQLW SHPXO LPHD
)XQF LD

X n , x = (x1 , x2 ,..., xn ) , n 1 .

f(x) are un maxim (minim) global n punctul x o X  GDF

( )

( )

f x o f (x ) ( f x o f (x ) HVWHDGHY
)XQF LD

UDW RULFDUHDUIL

( )

f x o f (x ) ( f x o f (x )  HVWH DGHY

 LQHJDOLWDWHD

x X .

f(x) are un maxim (minim) local n punctul x o X  GDF

( )

f(x)

UDW  RULFDUH DU IL

 LQHJDOLWDWHD

x X situat ntr-R YHFLQ

WDWH D

punctului xo.
Q PRGHOHOH HFRQRPLFH HVWH QHFHVDU V  GHWHUPLQ P H[WUHPXPXO XQHL IXQF LL
OXkQG vQ FRQVLGHUDUH QX WRDWH SXQFWHOH PXO LPLL GH GHILQL LH

f(x)

X n , ci numai acele

SXQFWHFDUHVDWLVIDFXQVHWGHFRQGL LLVXSOLPHQWDUHGHVFULVHGHXQVLVWHPGHUHVWULF LL

hi (x ) 0 

 P Q DFHVW FD] SXQFWXO GH H[WUHP VH QXPHWH H[WUHPXP

FRQGL LRQDW0XO LPHDSXQFWHORUGLQ

2,,m (m<nHVWHQRWDW

X care satisfaFVLVWHPXOGHFRQGL LL hi (x ) 0 , i=1,

FX6LVHQXPHWHPXO LPHSRVLELO DGPLVLELO 

$QDOL] PPDLvQWkLFD]XOFkQGUHVWULF LLOHVXQWGDWHVXEIRUP GHHFXD LL'RULPV 

GHWHUPLQ P H[WUHPXPXO IXQF LHL

f(x) supus reVWULF LLORU hi(x)=0, i=1, 2,,m, unde x

este un vector n-GLPHQVLRQDOLm<n.

2 SULP  DERUGDUH FRQVW  vQ D H[SULPD 

n-m YDULDELOH vQ IXQF LH GH FHOHODOWH m

hi(x)=0 LVXEVWLWXLQGX-le n f(x) se reduce problema la una

SULQUH]ROYDUHDUHVWULF LLORU

GH J VLUH D H[WUHPXPXOXL QHFRQGL LRQDW SHQWUX R IXQF LH GH 

n-m YDULDELOH $FHDVW

DERUGDUH SDUH IRDUWH DYDQWDMRDV  GDU RE LQHUHD GH VROX LL H[SOLFLWH DOH HFXD LLORU GH
UHVWULF LHVHSRDWHIDFHQXPDLvQFD]XULSDUWLFXODUHFkQGUHVWULF LLOH

hi(x) sunt liniare.

D GRXD DERUGDUH FRQVW  vQWU R PHWRG  FDUH QX UHGXFH QXP UXO GH YDULDELOH FL

GLPSRWULY OHVSRUHWH$FHDVW PHWRG HVWHFXQRVFXW GUHSW

metoda lagrangeanului.

/DJUDQJHDQXOVDXIXQF LDOXL/DJUDQJHVHGHILQHWHvQDFHVWFD]DVWIHO

L(x, ) = f (x ) i h i (x ),
i =1

unde: = (1 ,..., m ) sunt m variabile arbitrare numite multiplicatori Lagrange.

Q IXQF LH GH VHPQXO DGRSWDW SULQ FRQYHQ LH SHQWUX YDULDELOHOH

i, lagrangeanul

SRDWHILVFULVLVXEIRUPD

L(x, ) = f (x ) + i hi (x ).
i =1

6 REVHUY PF SHQWUX

x S = x hi (x ) = 0 , L(x, ) = f (x )LF

L x, hi (x ), dac x S
.
=
i
0,
dac
x

6  SUHVXSXQHP DFXP F 

(xo,o) HVWH XQ SXQFW GH H[WUHP SHQWUX IXQF LD L(x, ) .

5H]XOW F GHULYDWHOHVDOHSDU LDOHvQDFHVWSXQFWVXQWQXOH&XPGHULYDWHOHSDU LDOHv

raport cu VXQWQXOHUH]XOW  x o S L L(x o , o ) = f ( x o ) , deci xo este un punct de extrem

pentru f(x)GHILQLW
FRQGL LRQDW

SHQWUX

x S DGLF x

HVWHRVROX LHSHQWUXSUREOHPDGHH[WUHPXP

Prin urmare, cnd avHP GH UH]ROYDW R DVWIHO GH SUREOHP

/DJUDQJH L DQXO P GHULYDWHOH SDU LDOH DOH IXQF LHL
SXQFWHOHFULWLF'HULYDWHOHSDU LDOHvQUDSRUWFX

le, ne asigXU

PF 

 VFULHP IXQF LD OXL

L(x, )  SHQWUX

D J VL SXQFWXO

VXQWWRFPDLUHVWULF LLOHLDQXOkQGX-

xS .

$QXODUHD GHULYDWHORU SDU LDOH DOH ODJUDQJHDQXOXL vQ UDSRUW FX

x, conduce la

VLVWHPXOGHHFXD LL

f k i hki = 0 , k=1,2,,n.
i =1

2E LQHUHD SXQFWHORU FULWLFH DOH IXQF LHL

L(x, ) constnd n rezolvarea sistemului

GH HFXD LL UH]XOWDW GLQ DQXODUHD GHULYDWHORU SDU LDOH HVWH FXQRVFXW  VXE GHQXPLUHD GH

FRQGL LLGHRUGLQXO, de extremum:

L x,
=0
x

L x,
=0

i

k = 1,2,..., n

i = 1,2,..., m

sau:
m

i
f k i hk = 0
(I )
i =1
h i (x ) = 0

k = 1,2,..., n

i = 1,2,..., m

Acest sistem de (n+m) HFXD LL FX (n+m) necunoscute constituie, deci, cRQGL LLOH
QHFHVDUH GH H[WUHPXP (OH QX GDX LQIRUPD LL GLQ FDUH V  GLVWLQJHP SHQWUX

f(x)

SXQFWHOH GH PD[LP GH FHOH GH PLQLP 3HQWUX D IDFH DFHVW OXFUX WUHEXLH H[DPLQDWH L

FRQGL LLOHGHRUGLQXO,, de extremum.

$P Y ]XW F  DFHVWH FRQGL LL GH RUGLQXO ,, SHQWUX R SUREOHP  GH H[WUHPXP
QHFRQGL LRQDWVHH[SULP vQWHUPHQLLXQHLKHVVLHQHRPDWULFHDGHULYDWHORUSDU LDOHGH
RUGLQXO ,, L D VHPQHORU PLQRULORU V L SULQFLSDOL 'DF  DP SULYL SUREOHPD FD XQ
H[WUHPXP QHFRQGL LRQDW DO IXQF LHL

L(x, ) s-DU S

UHD F  HVWH VXILFLHQ V  FHUFHW P

KHVVLDQD DVRFLDW  DFHVWHL IXQF LL 1X SXWHP V  SURFHG P DVWIHO GHRDUHFH GDW ILLQG XQ

xo punct de extrem, L(xo, ) = f(xo) L UH]XOW

PRGLILF UL DOH OXL

 'H DVHPHQHD PRGLILF

F 

L(x, ) HVWH VWD LRQDU

ULOH OXL

 SHQWUX RULFH

xo nu pot fi oarecare, ca n cazul

XQXLH[WUHPXPQHFRQGL LRQDWFLHOHWUHEXLHV VDWLVIDF VLVWHPXOGHUHVWULF LL

3HQWUXDGHGXFHFRQGL LLOHGHRUGLQXO,,FRQVLGHU P
L LQkQGFRQWQXPDLGHPRGLILF ULOHFRQGL LRQDWHDOHOXL

L(x, )FDIXQF LHQXPDLGHx

x.

Avem:
m
2 L x,
= f rk i hrki .
xr x
i =1
k
(YDOX P DFHVWH GHULYDWH GH RUGLQXO ,, vQ SXQFWXO FULWLF
PDWULFHDVLPHWULF GHRUGLQXO

(xo, o)

L QRW P FX

nn:
2 L xo ,o
A = x x
.
r k

'LIHUHQ LLQGVLVWHPXOGHFRQGL LLUH]XOW 

dhi = hki dxk = 0 , i=1,2,,m.

k =1

1RW PFX

B matricea m n DGHULYDWHORUUHVWULF LLORU

[ ]

B = hki
L

FX

u

YHFWRUXO

FRORDQ 

Q GLPHQVLRQDO

DO

YDULD LLORU

OXL

x

DGLF 

u T = (dx1 , dx2 ,..., dxn ).

&RQGL LLOHGHRUGLQXO,,GHYLQDVWIHO

SHQWUXPLQLPIRUPDS WUDWLF 

SHQWUXPD[LPIRUPDS WUDWLF 

uTAu>0 sub restric LDBu=0;

uTAu<0 sub restric LDBu=0.

&RQVWUXLPDFXPPDWULFHDP UJLQLW 

H, astfel:

A BT
H =
.
B 0

Matricea H are ordinul (n + m ) (n + m ) ILLQG FRPSXV

 GLQ

A de ordin n n , B de

ordin m n , BT de ordin n m LPDWULFHDGHRUGLQ m m &RQGL LLOHGHRSWLPDOLWDWH

GHRUGLQXO,,VHHQXQ

DVWIHO

a) pentru minim, determinantul matricii H L WR L PLQRULL V

mai mare dect (m+1) WUHEXLH V

 DLE  VHPQXO

L SULQFLSDOL GH RUGLQ

(-1)m XQGH P HVWH QXP

UXO

UHVWULF LLORU

b) pentru maxim, determinantul matricii H WUHEXLH V

 DLE  VHPQXO

HVWH QXP UXO YDULDELOHORU LDU PLQRUXO SULQFLSDO GH RUGLXQ

(-1)n, unde n

l (m+n-1) WUHEXLH V

DLE VHPQXORSXVGHWHUPLQDQWXOXLOXL

H. Succesiv, minorii principali mai mici

WUHEXLHV DLE VHPQHDOWHUQDWLYHSkQ ODPLQRUXOSULQFLSDOGHRUGLQXO

Matricea H

FRQ LQH

PXOWLSOLFDWRULL

/DJUDQJH

8WLOL]kQG

m+1.

FRQGL LLOH

GH

optimalitaWH GH RUGLQXO , HL SRW IL HOLPLQD L UH]XOWkQG R PDWULFH FDUH FRQ LQH
QXPDLGHULYDWHOHSDU LDOHGHRUGLQXO,L,,DOHIXQF LHL LDOHUHVWULF LLORU

3HQWUX H[HPSOLILFDUH V  FRQVLGHU P FD]XO

UHVWULF LH

m=1

hi.

DGLF  H[LVW  R VLQJXU 

h(x)=0.
L(x, ) = f (x ) h(x ).

L (x, ) = f h = 0
k
k
x
(I ) k
L (x, ) = h(x ) = 0

k = 1,2,..., n

Matricea A este:
f11 h11
f h
12
A = 12
...

f1n h1n

f12 h12
f 22 h22

f1n h1n
f 2 n h2 n

...

f nn hnn

...
...

...
...
f 2 n h2 n ...

Matricea B este:
B = [h1 h2 ... hn ].

Matricea H este:
f11 h11
f h
12
12
...
H =
f1n h1n

h1

&HDPDLVLPSO IRUP DPDWULFLL

acest caz hk=ctLhrk=0, rezultnd:

f12 h12
f 22 h22
...
f 2 n h2 n

...
...
...
...

f1n h1n
f 2 n h2 n
...
f nn hnn

h2

...

hn

h1
h2
...
.
hn

UHVWULF LD

h(x)=0HVWHOLQLDU

HVHRE LQHGDF

Q

 f11
f
12
...
H =
f1n

h1

f12
f 22

...
...

f1n
f2n

... ...
f 2 n ...

...
f nn

h2

hn

...

h1
h2
... H f
=
hn grad h

(grad h )T

unde Hf HVWHKHVVLDQDIXQF LHLf.

LQkQG FRQW GH FRQGL LLOH GH RUGLQXO , SXWHP vQORFXL SH
RE LQkQGRPDWULFH

hk prin fk, unde = 1 ,

HFDUHFRQ LQHGRDUGHULYDWHOHSDU LDOHDOHIXQF Lei f:

H
H = f
f

(f )T
.
0

$WXQFL FkQG VLVWHPXO GH UHVWULF LL HVWH GDW VXE IRUP  GH LQHJDOLW

h i (x ) 0  L DYHP L UHVWULF

LL GH QHQHJDWLYLWDWH DVXSUD YDULDELOHORU

L GH H[HPSOX

x 0  DD FXP VH

petrec lucrurile cel PDLIUHFYHQWvQPRGHOHOHHFRQRPLFHSXWHPSURFHGDODDG

unor variabile auxiliare ui SHQWUX D WUDQVIRUPD LQHJDOLW

LOH vQ HJDOLW

XJDUHD

L &HD GH D L

-a

UHVWULF LHGHYLQH

h i (x ) + ui = 0 , ui 0 , i=1,2,,m.
)XQF LD/DJUDQJHDWDDW HVWH

i =1

i =1

L(x, u , ) = f (x ) i h i (x )+ ui = f (x ) i h i (x ) i ui
i =1

&RQGL LLOHGHRSWLPDOLWDWHGHRUGLQXO,vQUDSRUWFXYDULDELOHOH

x vor fi identice cu

cele din cazul anterior. n raport cu variabilele ui, avem:

L x,u ,
= i .
ui

Cum i 0 UH]XOW

te sunt negative, implicnd fie i=0, fie ui=0.

F DFHVWHGHULYD

$FHVWHFRQGL LLQXLPSXQUHVWULF LLGLUHFWHDVXSUDYDULDELOHORU

echivalent sub forma: hi (x ) 0 , i=1,2,,m, 0 ,

x. Acestea pot fi redate

h (x ) = 0 . Scriind lagrangeanul
i

i =1

I U D LQHVHDPDGHUHVWULF LLOHVXEIRUP GHLQHJDOLW

LUH]XOW 

L(x, ) = f (x ) i hi (x ), 0 .
i =1

5H]XOW  F WRDWH SXQFWHOH RSWLPDOH vQ UDSRUWFX YDULDELOHOH

u DOH IXQF LHL L(x,u, )

VDWLVIDFUHOD LD

L(x, u, ) = L(x, ).
/XkQGDFXPGHULYDWHOHSDU LDOHvQUDSRUWFX

RE LQHP

L x,
= h i (x ) .
i
'DU FXP UHVWULF LLOH FHU FD

h i (x ) = 0 sau i = 0 UH]XOW
L (x, )
0
i

(sau

h i (x ) 0  LDU FRQGL

LLOH GH RSWLPDOLWDWH VXQW

F 

L (x, )
=0
i

sau i = 0 sau ambele)

FHHD FH DUDW  F  WUHEXLHV  DYHP XQ PLQLP DO IXQF LHL
ILLQGUHVWULF LDGHQHQHJDWLYLWDWH
'HFLGDF 
VXEUHVWULF LLOH

0,

L(x, ) n raport cu  GDW

0.

(x , ) este un punct optimal pentru problema de maximizare a lui f(x)

o

h i (x ) 0 , i=1,2,,m, x 0 HOWUHEXLHV

i =1

i =1

YHULILFHFRQGL LLOH

a) f k i hki 0 (sau f k i hki = 0 sau xk = 0 sau ambele);

b) 0 , hi 0 (sau hi = 0 sau i = 0 sau ambele).
$FHVWHFRQGL LLVXQWFXQRVFXWHVXEQXPHOHGHFRQGL LLOH.XKQ

DuS

 FXP V D REVHUYDW IXQF LD ODJUDQJHDQ

L(x, )  FD IXQF

-Tucker.

LH GH GRX PXO LPL

de variabile (x1 ,..., xn )L (1 ,..., n ) are un maxim n raport cu unele (x)LXQPLQLP
n raport cu celelalte (). Un astfel de punct se nuPHWHSXQFWD.
6ROX LD SUREOHPHL JHQHUDOH GH H[WUHPXP FRQGL LRQDW UHYLQH GHFL OD J VLUHD
XQXLSXQFWDSHQWUXIXQF LDOXL/DJUDQJH

L(x, ) .

8QUH]XOWDWLPSRUWDQWDODQDOL]HLPDWHPDWLFHDUDW F GDF 
RSWLPDOSXQFWDSHQWUX

(x , ) este un punct

L(x, ) , atunci:

L x o , o = max min L(x, ) = min max L(x, ),

x

cunoscut sub numele de teorema minimax.

6 REVHUY PF IXQF LD/DJUDQJHVHSRDWHVFULHLVXEIRUPD

L(x, ) = f (x ) + i h i (x )
i =1

FXUHVWULF LLOH

x 0 , 0, h i (x ) 0 .

Cum L(x, ) HVWHRIXQF LHOLQLDU

HDHVWHLFRQYH[

de x, L(x, )  HVWH R FRPELQD LH OLQLDU

HVWHFRQYH[ 5H]XOW F 

HVWHFRQYH[ 

vQ

 SR]LWLY  D IXQF LLORU

acestea sunt concave, L(x, )  HVWH FRQFDY

[h (x )]
[h (x )]

3ULYLW

vQ

 )XQF LD

L(x, ) HVWHFRQFDY

2E LQHP DVWIHO R WHRUHP  GH H[LVWHQ

FDIXQF LH

f(x) L [ hi (x )] L GDF

[ h (x )]
i

 HVWH FRQFDY  GDF 

vQ GDF 

f(x)HVWHFRQFDY

L

 D VROX LHL SUREOHPHL JHQHUDOH GH

extremum:
3UREOHPD 0D[ I[ VXE UHVWULF LLOH

h i (x ) 0  L

P L

x 0  DUH VROX

LH

GDF 

1) f(x)HVWHFRQFDY
PXO LPHD

2)

LILHFDUH

h i (x ) HVWHFRQYH[

S = x hi (x ) 0, i = 1,2,..., m, x 0

Q DFHVWH FRQGL LL IXQF LD

HVWHP UJLQLW LQHYLG 

i =1

HVWH RSWLPDO SHQWUX SUREOHPD GH H[WUHPXP L

FRQGL LLOH.XKQ
'DF 

(x , )

satisfac

L(x, ) = f (x ) i hi (x ) DUH XQ SXQFW D x o , o , unde x

o 0 . Valorile

-Tucker care sunt suficiente pentru un optim global.

f(x)HVWHVWULFWFRQFDY

SXQFWXO

xo este unic.