
第3 6 卷 第 5 期 上海师范大学学报(自然科学版) Vol.36,No.5
2007年10月 JournalofShanghai N orm al U niversity( N atural Sciences) 2 007,Oct.
An affinescaling inexactNewton m ethod with
in te r io r p o in t b a c k tr a c k in g te c h n iq u e fo r
nonlinear problem s subject to bounds on variables
GU Yi-ming, ZH U D e-tong
( C ollege of M athem atics and Sciences, Shanghai N orm al U niversity, Shanghai200234, China)
A bstract:
W e prop ose a new affine scaling inexactN ew ton m ethod w ith in te rior p oin t lin e sea rc h tec h n iq u e for so l-
ving the box constrained nonlinear optim iza tion . C h o o sin g the scaling m atrices which have sm oothness properties, we
consider reform ulated least squares with bounded constra in ts su b stitu tin g th e origin a l p rob lem . B y the inexact N ew ton
m eth od w e get a decreasing dire c tion in e ac h ite rate . W e a lso u se lin e sea rc h a lo n g th is d e c rea sin g d ire ctio n to m ak e
th e m e rit fu n c tion adequately degressive and m aintain s tr ic t feasib ility. The alg orith m is sh o w n to b e glob a l an d loc a l
convergent under som e reasonable assum ptions. The results of som e num erical experim e n ts are re p o rte d to sh o w th e
effectiveness of the proposed algorithm .
Keywords:
lin e se a rch ; inexactNewton method; interiorpoint
C L C num ber:
O221.2
D ocum ent code:
A
ArticleID:
1000-5137(2007)05-0022-08
R eceived date: 2007-05-02
F o u n d a tio n ite m : T h e a u th or g ratefu lly a ck n ow led g es th e p artial supports of the Ph. D . Foundation G rant (0527003) of
C hinese E ducation M inistry and the Science Foundation G rant (06DZ037,05DZ11) ofShanghaiEducation Com m ittee.
B iography: G U Y i-m ing(1974 - ) , m ale, C ollege of M athem atics and Sciences, Sh anghai N orm al U niversity; ZH U D e-
ton g( 1 954 - ) ,m ale, doctor, professor, C ollegeofMath em atics an d S ciences, S han gh ai N orm al U niversity.
1 In tro d u c tio n
W e c o n sid e r th e so lu tio n o f th e n o n lin e a r o p tim iz a tio n p ro b le m w ith b o x c o nstraints:
min f(x) s.t.l≤ x≤ u , (1.1)
where f: R
n
→ R is a s m o o th n o n lin e a r fu n c tio n a n d th e v e c to rs l, u ∈ { R ∪ { ±∞ } }
n
,l< u.Letthefeasi-
ble setF = {x ∈ R
n
| l ≤ x ≤ u } a n d th e s tric t in te rio r fe a s ib le s e t F
0
={x∈R
n
|l< x < u}.
Letthe vectorx
*
∈ F is a stationary pointfor the problem (1.1). Then for every i∈{1,2,…,n}
f
i
(x
*
)
=0 if l
i
<x
*
i
<u
i
≥0 if x
*
i
=l
i
≤0 if x
*
i
=u
i
{
,(1.2)
where f
i
( x) is the ith com ponent of the gradient vector of f( x) , l
i
,u
i
an d x
*
i
are the ith com ponents of l, u
and x
*
, respectively.
D uring the last few years, there have been quite a few papers dealing w ith problem (1.1). These refer-