• Nu S-Au Găsit Rezultate

View of On a Newton type method

N/A
N/A
Protected

Academic year: 2022

Share "View of On a Newton type method"

Copied!
5
0
0

Text complet

(1)

ll1ì\'f iì r)';rN.\t.tstì ¡-uNÌìnIQUIì Iì,I rltì

].iltioIìlIì

I)rt L',\t'ì,ltoxlìt.\1.ro\

'l-cnre 23, No P, lf)94, ltp, l$7- 1221

ON A NE\ /TON TYPII. I{äTHOI)

IO]IN r..\zÃtì

(Cìu j-N:r¡toca)

I

.

IN'l.ltoì)u(;T l0N

rype :*:'åiä.il: 3d;,ä:îi*îi"ï,^"1î3il"iîå"å ",Ì"ìffiì

also

eiarnlrle.

the

r,oots

of a

single

nonlincar

equation .f(n)

:

0,

Moser [2]

has proposed

the

follorving- it-eratió;"s

fin+r: r" - !/nÍ(rò, lJn+t:'!J" - l/n(f'(ròy* -7),

t¿

:0,I,...

the first iteration is

sirnilar:

to

Ne eclna,l

Lo 7lf'(n"). 'Ihe

seconcl

iter

that :I$2... rhe

orcrcr

or

con\¡er.sc,ìc"

r""";lÏJiil:,:"1ä:i"t1-'?î Ì'¡;;T;":

.An

irnproved scrrc're \\'as proposecr

by u|rn [4]

ancl

Hafcl f1l.

tr'or' 1,he cquation

(1.1)

F(;n)

:

o,

*.ltete

n' : D c -\'_- I, -L a'tl y

a,re tn,o B¿r,n:r,ch spaces ancl

/) c -l a'

open set, the a'thors ha'e

consiclerecl

ilre

iteratioî,s

(1

.2)' r;n+t:

e:,,

-

A,,F(ur),,

(1.3)

An+t

: an

- A^(n,(c,n*t)An- 1) : a,,(2I _

x,'(e:,,nr)A,),

n:0, 1,..., n,ilìr

LIt.c'

jnifial

qrr(rsses

"roe

f)

arrr-l _zl

oe

/,(

I,,

_\.), qlrer,c

!-(y,.\') Y into

,ìí.rlcrrolcs

llre

.Barraelr

*iir.o ui 1l;"31*"ìì;ìi iiiLit;

o¡cr,aror,s ¡r,orr

. ,Jterations (1'2).antl (1.8)

rreep

the

propertics

of

n{oser's iterations ancl

thc

seqnencc (ø)_conr-e'gj"*

q.ä,l"atiäall'i'to a

snr..tion

of (1.1)

(scc Zehncler.

[5]

ancl

rráia p

11.

(2)

,

Oll a Ncrv[o¡r '-l')'pc l\.I0tììod

168 I. Lazirr 169

Since,

by

(2.1)

I]ncler l(a'tor.or.iCh t¡'pc

zìsisurnlltions)

si.rilarl¡',.rvit'h

l,hose 01

givc a

cónvelgence theorem

for

the bouncls are sharper t'han t'hose

from

th:rrr in ì{en'tonts

llllocess.

2. (lONYlìIìGlÌ\i;lì 1'lIIìolllìll

[)oncorningccluatiorr(1.1)ancltlreiter.ations(1.2)_arrcì11.3)l'ehar'e llrtnolturt

2.1.

II It is þ'róclta 'tttfeì'ltti"iìL''"" 51'o' ì'; - !Y-t 'Y'

l,

r--a,orl < r'l c n',i,l,t'lni'"

r'o,1,,íi'"ì,'r"

tí, ír o

ttttd r¡ >"tJ

thc 'fdlouing

c¡tn, tl'í,t i,c¡ns h,oltl

(2.

la) Ao is

inuet't'ible utttl

-4t'

e

t

(-\

'

Y) ;

(2.

1b)

llr{oLi(øo))ll

< I

;

(2,

1c)

ll

Á0(/¡'(r)-lr"1v¡¡ll < I

ll

n-all,

V

n'v eS(ro'

r') ;

2.

1rt)

li

r -

Àozr"(rco)

ll <

q ;

I

¿¿,

(2.

1e) d::lt'rt* q < t +Íl

ct'tttl

t" >

'1t

+ hvfFl

rtcr (r',,)

is

wrIL tLe,finett by

(l

'2¡ nrrrl (-l '3)'.rettmin's irr'

S(rn' r)

ttt tt'

stlttliott' t:* oJ

,1ql"ìtlo'i

(l

' l

)'

'l'l¿'is solwl'iot¡

is

tr'n'iqu'e

r), iJ r <(1-q)/tu'

tire ioltotoitìg

u þriori enr¡r

esl'ì'¡t'¡tt'tes

¡ a crÐr*!, , IA, - It''(r'r¡-tll <

cz(2d)'¿"'

n:r'2'"'

tntd the

tr, Ttosteri'ot'i errot' estitt¿u'tes

(2.3)

li e,,

- T'i'li <

(2cl¡'"-t ll"to

- r"-tll'

n'

:2t 3¡"'

r ¡3l I tr)

ll /.11.

wherc

Ct = .'- '-,'

-

'ntttl

C"

-

7,,1

t -ar12) o

I -sdz

ProoJ'

Fot'rr

n'nrl "1,

givcn b¡'('l'2) lnrl

(1'3) rvc shatl

tinrl sitnilat

lelal iotts

to

a*stttt'ptions (2'1)' ,

Irrorn (1.3)

ib

lollorvs i

ltat

Arrt)

: (2I - Ao\''(rr))Aow, lor all

w

eÍ' I -

tlrÞ''(rut)

: (I - Aul'(n'))z

I-A'A;1 -AoE'(n')-I'

(2.7) llI -

Aop,(ør)lJ

< lll _ Aolt,(ur)ll + ll

Ao(I.,(ao)

_ IÌ,(ar))ll

< tul *

e

: it,

u,e

obtain,

using (2.6), ll

I _

A Á_å i/

< d < 1, vhich irnplies that il

îhis .,rere exists

(,4r,4;1) e

L(X) ancl ll(:lr:l;,¡_r il <

f /(1 __ a¡.

(2.8)

-4, is invertible, _4¡r e

L (X, y)

ancl /1,4f,

ll < li-A;, IyÍ _

d,).

}-rom e.4)

ancl

(2.7)

tve obtain

(2.s)

llArî(Ør)ll <

(1

+

r¿)il

Aor- (ur)ll.

and, by

(1.2)

Ao,(nr) : Aoln(rr) _F(ro) _F,(u)(ø1_no)l

+ Aol,(ro) | F,(øo)(rr_nòl

:

A olF ( ur) _ tr, ( u

o) _ ]., (. r o)(

ut*

n ò

I _f ll

_ A oF, ( r o) JA op( n r).

But, fi'orn the

assnmplions

(2.1) it

follorvs

that lhe

is

dirrercntiabte on

Sü;;^;);;ir',i. j',,

Lipschirz,

so operator' ÁoF(.)

ll

AolÛ(nr) _

F(æo)

_ F,(a)(n, _ nòlll

_.

1

ll

nr_rollr, thus, frorn the

above

inequality

and (2.g)

(2.10) lla,r'(æ,)it <

(1

* ,[* n, + tj:t \u (d,_q,).

üre

also

obtain from

(2.4),

(2.7)

and (2.1c)

(2'11)

ll Area'(æ)

- x'(v))li <(1 +

d,)r!ao@,@)

- p,(E)), <Ãi(l *

d)

lrn*y

ri,

;iîlii?íue

8(ø0, r'),

(2'72) lI - Ar?,(#l)il < lll _,qop,(nr)ll,

<

ar.

n"o'r"

uJ$"inIT:?,åtqJ' r0

-_12)

are simitar to

assumptio

i*,,

r,r,i,årï;îîïi;T)113J ,'l"HT;î1i ana (r ãi";;;"#;u'- á|ittl"ä,",,i1î

** ,"rilil_r, let us

consider ttre"sequences (/c,),

(n), (q^) and(d,)

definetl

ho:'k, ïo: I;

Qo

- gt do:

d,

(z rB) r::_':,i;Iî,:^_,,,

!" :

dÊ-r, n^

:-rjþ kti-'

-

etr-t),

n ='r,

2,. . .

(2.4) (2.5) ancl (2.6)

(3)

77 (\

\yhcncc it, folloti's th¿r,t

(2.16) anrl

[, [,az-irl 4

Jr'or' (rln) \vc obatirì 1,he follorving

recrLlrcrt

t'elal,ion

(9.11)

d17

:1(.1 f

1 r/,,-r)2(r[u,,

--

c1i

,) !

tLi_r,

ancl,

nsirg

(2.1e) wc caÌr pror.c

by intlucliou

tha,t

(2,15) I,4 il,u(

(1,-r

' <

-

7+V2 -l -= rtntl

cl,

4 2tti ,, fol all n >

L,

r\ow, b¡.

(2.18c)

and

(2.19)

llr,,,*, - rnll " "'f'

.n,

<Qd,)'_'

.r --

(2¿¡r,,,,,

¡!,, 4.lt I _(2¿ty

!t,) i. a

Carrchv sr:

J,cL x:,r,

--ìì;ì,"ä,.

ñt'qllcncc' so

it is

corrvelgort.

rorn tJrc abovc

iuctlrratiiy"ü

Ona

1/t,

Ncrvt.ou TJ'pc ì{eilrocl

(2¿¡'"

7

-

(2cl)2" 1

777

,

for all

tt,

tn ÞIt

flrer.efote

follorvs ilrai, for,

all lr, I,

(2.20)

llr,-r* //< I

.1,

<

.1 (!r1)r" , l

(:'l i)

(1,, tli,;t

< I ,1., < l

(2tr\.¿,

t,' -t; ' I/

foL' a1t n

>7..

So,

it

calì bo shou,n lrv ilclrLctiotr tlli¡l,

fot all

n, Þ

I

(2.

tLla) r,

e rS( rrru, t') :

(2,,t3b) thclc

crisl.s

,l;r

e

,(-\) I) arxl ij,1;'i <

l,l;.1,

lii(l --rl,-t)) (2.18c)

liÄ,,11(r',,) I

{

.4, ,

(2.1,!tl)

), ,4,,(11'(a;)

- lt"(ll)) i(

/:,,' ,.,

-'!l

1,

fol :ll[ nt ll eB(ro,

r') ;

(2.

tEt,)

ii

/

':1,,11"(:r,,) '

a

rtu.

Lr'<.r' r¿

.:

1, (2.18)

holl.

\\¡c suplri)se ttrril,

(2.t8)

holcl

lol

i,

: I,

...,,u,

SitL<rc,

fot

r¡11

j, n >

1

iiDt t .,,,4-7.''i'l ',:l¡ll:'( t 'i,'(Zt¿)-ittt¡-

,8, ' ' 4t; .ri

,r1,

(l'

I

l))

ed)' 't

12¿ ¡''

''

rl l;: (2rl)t'

zr,rcl f/ cr;

,-

noli<

//

,i _

noll 1_ ll n,

_n*

/l

<

.n _L

_d2

[Jr¿Í, is a,'n.e S(ro,

r,). ¡ \ '/ '

fu(.1 -_ J¿-i)

*

r''

, l.ru,thct,

¡is

,,e r'riltcn ,

,, w

prol¡e

1,Itat

s,r is

¿t,

solution of (1.1).

Silc

c (L2)

catt

il,.is

sufficie't to sh

A;',(n,*t

-

nu)

:

-F(u,,),

Jlrrsi;, b1,

(2.1g)

ow

tJi t the

sequence (//

A;til)

is bou¡clecl.

(2.27)

!!

t _ tl,,F,(#*)

ii

<

ll

I _

A,,L.,(c),,)

ll

+

li

A,(,.,(n,)

__ Lt,(rx))

< q,

+_ l;,Jl a,,

_

ø* ¡¡

i/, for

al_ rt,

¿

J, ancl,

br. (Z.llj) anrl (2.1õ)

")

.tn+tl-4,t

-- ,:(, -l

el,,_,)(r:,,

l_

rt)],_r¡<

1.

tl"

-- 1'' '4"

- -

,,.,r-

< ¿/,, !t:L lt _ r!',4, _ hl;

l

l\-lìelì cr_-,

tl'h

ich

iu Lplics

¡/Ð o.), ,, æ

\a'4¿)

L

r,,

,,* i]

<

X

2 'r¡, 1

1-rt {

[r,'au, I'r,t' ,¡¡1 tt , ttt ) .l .

?u r-ra

(

4,, 1

,[

-r) '4,,, 1ìrl a,ll n

> l"

¡¡t:O2'r.

'-So, 1,hc estin¿l,tes (2.2.L) becorne

it

[o]lo u's t'[ttr,t

li ,r',.,-,

-rii,li< r,-t 2, \i {.'r, i ,,, rl:1 ,., <r,

Ircnce

íril1.1

c:s(oo,r'), -.,r,rt ,- ã.'s*l r,,,'

o''.' ,,,

,ootlr'ono ,rrì

rr'o\.o (2..LBlr--e) fot'

,i,'=

tL -l-

1, as

rve

ltavc plovetl

(2.8) and (2.'1.0 -2.12).

(2.23) l!r -.1,,11,(or,)i/<

q,,i_2/i:,,.r,,,_-2d,,,_

{lu{Bct)l_,=lo{ruir,,

13v

(2.18') a'ci (2,'5)

rvc

'¡ìvc ltA;t]l<i/;t¡r llle__(ty,,a,nr[

rising

(2,28)

llA;t -

I1t,(nt,¡

i/<//;l;ril,ll ¿

_ tL,,n,,(nr,)J/< tlì ir

,.r,,

(2t1¡2,,

tvlreltcc lt

,l

.r

-

rttt/ ^,;i.\

u ,\ , 4

'

"o l'

(7

-¡.¡''

t

*,:::,:i".'ii,";-í',Ííl,l[

rì.9',,"i'i ¡,un,,,,,

¡.il,;, i

Í1_f,,',ur,,- .r,c,.,rrrtrrJ,

lil.rr -

*'o == rrt. -_ tl ,rl!r(n,r) --_a,N ==

=. l,I -_.A,,'ltr,(n,) l(an

_

n4) _1. tI,l P(e*) ._

I(n,,)

__ [tj,(n,,)(6,t,

_

n,,)

_],

(4)

en+t{ (+ +

e,,)r,,,

ancl, b¡'

(2.22)

(2-24) dn+r

(

(lt^n^

*

q.)en

:

rl,en,

for all n > 1, thus, using

(2.16)

,L-1

., * (tt, ,,) ,, * + e,t)':"' ,, :rff,,,

-r, flol. Q.20) there follows

1,he

filst part of

(2.2). tror b5,Q2B) for

n :1

thele lollorvs I

I _A,tì,,(u*) ll ì

SAh .<

t

isl,s

.lAr7'("iJl_-l

e

L(X) antr

ll ArV,1s*¡-r

ll i l¡1r _B(t\',

(2.4) and

(2.7)

ll.Ér'(¿x¡-ril

<1/(1 -ldr)ll1,ll < I+d

ll/^¡.

r_3d2

"

Finally,

because A,,

- l'1¡*)-t : - lI - A,fÌ,(a*)lP,(n*)-r, using

(2.28)

rve ge!

ll

A,,-fi'1ø'*)-t ll 4

Or(2d,)2''.

ifhe a

posteriori

örror

bound (2.S) follorvs

florn

au

4

dn_t €n_L¡ €,,_1

4 2.qn¡ ancl

(2.16).

.

- ithe-uniq,*eness

of ilre solution u*, if r < (r -rr)lt¡,

follo.rvs

from tlrc fact

thal, 1,he operator

p(r) : r -A9!(ø¡ i!_coìtrac"ii""-óu" Sir;;

;,i:

rnclecd,-

+ ll/o(¡"( p no)- is

c'rifferenriabre, F'(n)) ll

<

/Lr

I

ìl

¡,'(r)

q

< r,

ll

: if

¡it ø e S1ä0,

-!rn'irjl

'r¡.

ãìl-¡-

,+rn;i*"ry,

í

772 1. I.az.irr

by (2.18) rve

get

3. NUì,IIìRIC^1, lìXr\tr[I,LE

Jret

us

considel

the

l_lammerstein

integral

ectrual,ion

r'et ñ : (ø');rro'

D=

:

á(s,)ilo.

so thc

s¡'ste'r (3.3) can be *,rittclr (3.4) I@): e - lf )e _ - _ 0,

rvhere

?: Àr+, +

lJ.v1r.

tr-or ø- e ll.\,

lr ll,e

set ,7

:

orr l?À' r

j t,,(ã) :

./

_

f7,1åjr"à,,

'ìâx :2 ,ra* È",

eniç.r.

,so 'sircll

-

â.,/

<

(2,v

(3.4) rve shall use

lro :1

(rhe

,.,""*,,Tli,ii"1i îî,i;,?

(3'5)

,r-(,{rr)

_-

ùth)

- Arpl¡ut¡,

(3'6)

t7¿,¡¡

-

A,_t(21

_

I1t,(ù(i))Ai_1)

u,

.

_

11

2, ...,

lt.

5,?j1_:t_

eonrputing ilre rnatr,jx,4,

ivvely. So rve musù

ps.

sat'c

lhe

Fréchtt

of application of (8.5)

ana

tes for

qo, c/.

i

csrirnatäs (å.'Bi1:,. "il

rl,T"l

6 7

On a

^\c\\,ton

t of arìilunetic

operations

\{/i'iting

s

--

,s,

jn

(8.2)

for the

rnethod G,2,8) applieri

to

solrre

I'e

obl,ain

the

nonlineâ,r s¡rsN6¡n

ø¿

- ErI(si,,e¡, ñ¡)ru¡:

ó(s,),

l:

o¡1r..

'l'l'pc lIcilìod atìl0un

(3.2)

(3.3)

,/Y

(3.1) ø(s)

-

s?(u(t))z U,

::os,

s e

[0,

1],

in C

[0,

1], anrllet H

(s,

f,

ø)

:

stttz, ö(s)

:

9120 s.

^ Using

t_he repea-ted trapezoidal

rule rvlth

fr,¡

:i-lM,

s,

: iffl, [:

I 0',1',.-.., l, w .

llyt

i - !,...,T11

and,_wo

:

r,oN

:

n,*¡2,

*iaþpr*

ximate the

exact solution ø(s)

:

sl2

of

(S.1)

byihe sölution'oî iheËci;_

tion

(3.2)

ø(s)

- r&t(r,

s¡,

ã(s¡))w.:

ó(s), s e [0, 1].

v[e are not

concernecl

l'ith the

existence

of the

solutions

of (3.1)

and

(3.2) or

wil,h

the approximatiou error but we

stratt,

rai¡ãr

estirnate the

(5)

114 I, I.ttzät' tJ NIì\'UIì ì},.,\N'\I,YSII NUI.IIìIQUIì ]'.f ])I] TIII¡ORII, ])]I I,!\I'PROXIIìI,{TION

^V a plior i

cstinute.s

a p os tclior i

cs tiìnâ Lcs

'lorle 2J, No 9, lgg¿, pp. f Zã.-f7fì

(lo do

4 16 0.1

1.32 .10 1

1 .25 .10-1 7.24.70-1

2 .96 .10-1 2 .81 .10-1 2.78 .1C-1

2 . 1r) '10-s

1 .32 .10-3 1.29.10-3

1.03 .10-ã 3.31 .10-a 3.12 .10-0

6 .62 .10-3

4 .03 .10-'r 2.51 .70-É

ON (u,

p, m.)

-- CON\/trX trUNCTIONS

lìItIrItP. Ilr\CllS \/,\SILI' J\,III I-ESAN

(Cìu j-Na¡roca)

'1' I'l'trr, S, Ott tttc Ilcratiue llcll¡octs utillt SinttútantotLs ,\¡t¡tro:uitnaliot't oI tlte Inuersc rsf ltc

O1tcrrLlor. Iz.r'...\cad. Na[ì< ]jsLoDsltoi S.S.lì., 1,{:,4,,IOi_ ,111 (1g67).

õ. Zclrtrtìcr', .lj. .L ¡! llcn¡.¿r'li ttbotLL \4olt¡tt's lleiluttt'. (ìonrl¡. l.)r¡.c ¡\ppl. ttaf tt., 32, 3G1_366

(1 e74).

1, t\:l.Itotllr(;1.t{)ì\

lìccr.ivc(l lã XII 1903

Ilr

13

l *'c trcfillcrl a

cr:ìs:i ol,gerrr*itrizc(-[

c.,r\.c\

l,lru(.ti..,rr)t \vrlich

å,;ll1:ì rl,i,,Tr1":,:]l,o

,ro,,urnl,iä'i,,n,,"r"*i,,s., _i^,ì_l,i;;n,'r.

àn,.,."*

arrrt rlrrir,si_

an aflit,tÌìai,ìr,c ¿ìns\\.el

to fltis

orras_

s alrcl 1,o

sive

i:iomo

clìa,],a.t."ir;iiì;i.

Inslilule oI Oulctilus llcpttbIicii,:)'l

PO l.lo¡: 63

l,l0t) Clu.j-Nullc:rt llontrinitt

2, lz, þ, nt) - [:ON\ ItX Iì{;\(:.t,tOÀiS

I)r';rrr''rl'r'ro'r

:.1.

,1,.'J

./:

10,

irl

1,j,,:..,,,,1t l.ì ,t);

1,,?/¿

c [0, 1]=.,{

T";, ;:'ì:î','oì,)r,lr'i ,'ï1"'1,,,(,t il' ,,í¡l..""'

''' 'rr

fr), ú

| 'ir

¿,,r1

c,ri'

ir :

(2.I) tft*

1- nt(.I -_

r)y) <

1¡:r,1.,,(t)

-j r¡(1 _

te)J,1t¡¡Str,

tu, ,1,'f'-3'å, otå"11,,,,1ì!t?r,ji:l;t

'11

l,lrc (o., p,

,,)-conr.c,r

.ru'cbio's

on ,,,,, ,¡rÏ*Il,;IosÌ'f'to'N

2'r.

f,et

ae[-0,

Lt]. lt'tretr.

Jer(l¡;t))(tt) i7 untt orlq

,i,f

l2'2)

Í{,,,!1,)(,,1

..:Ii_ll) _

"_!:l'!!:)_

(t.__ 1¡¡1¡?,

,is ittcreusittq ou, (,nLu, l).1.

Referințe

DOCUMENTE SIMILARE

In case (2a) p −1 4 = − 1, M¨uller’s algorithm requires one exponentiation (Step 1) and five multiplica- tions (two multiplications in Step 2, one multiplication in Step 3 and

Astfel, problema P 1 este u³oar  dac  graful este bipartit: se rezolv  problema LP 0 1 ³i soluµia (întreag ) optim  g sit  este o soluµie op- tim  ³i pentru P 1..

Model 1: hammock, unidirectional fiber PECVD, instructor Model 2: hammock, unidirectional fiber PECVD, student 1 Model 3: hammock, unidirectional fiber PECVD, student 2 Model

Although 1:2 ratio has relatively high drug content than 1:3 and 1:4 ratios, this formulation was not selected for further studies due to low drug entrapment (58%) which implies a

Totuşi, prima mea întâlnire cu pictorul Ciobanu, un tânăr ca- re degaja atâta mister şi o atât de puternică atracţie, m‑a urmărit mult timp după aceea.. L‑am revăzut cu

2001: conferenţiar la Catedra de Engleză a Universităţii din Iaşi; titular curs Literatură Victoriană, curs de Literatură americană pre-modernă şi curs de

Specializarea: Limbă şi literatură italiană - modernă / română / comparată Lista candidaţilor. admişi cu taxă - ordonaţi după media

Due to the fact that beyond the goal of this paper, our final goal is to identify CA rules that are able to successfully segment images, we intend to study the application of CA