Dynamic Linear Programming - CORE
Transcript of Dynamic Linear Programming - CORE
Dynamic Linear Programming
Propoi, A.I.
IIASA Working Paper
WP-79-038
May 1979
brought to you by COREView metadata, citation and similar papers at core.ac.uk
provided by International Institute for Applied Systems Analysis (IIASA)
Propoi, A.I. (1979) Dynamic Linear Programming. IIASA Working Paper. WP-79-038 Copyright © 1979 by the author(s).
http://pure.iiasa.ac.at/1145/
Working Papers on work of the International Institute for Applied Systems Analysis receive only limited review. Views or
opinions expressed herein do not necessarily represent those of the Institute, its National Member Organizations, or other
organizations supporting the work. All rights reserved. Permission to make digital or hard copies of all or part of this work
for personal or classroom use is granted without fee provided that copies are not made or distributed for profit or commercial
advantage. All copies must bear this notice and the full citation on the first page. For other purposes, to republish, to post on
servers or to redistribute to lists, permission must be sought by contacting [email protected]
Working Paper DYNAMIC LINEAR PROGRAMMING
Anatoli Propoi
May 1979 WP-79-38
International Institute for Applied Systems Analysis A-2361 Laxenburg, Austria
The paper p r e s e n t s a survey of dynamic l i n e a r programming (DLP) models and methods, i nc lud ing d i scuss ion of d i f f e r e n t f o rma l i za t i ons of dynamic l i n e a r programs, models which can be w r i t t e n iri such a form, d u a l i t y r e l a t i o n s and o p t i m a l i t y con- d i t i o n s f o r DLP, numerical methods - -bo th f i n i t e - s t e p s and i t e r a t i v e .
DYNAMIC L I N E A R PROGWL!ING*
A . Propoi**
INTRODUCTION
Dynamic l i n e a r programming (DLP) can be cons idered a s a n e x t s t a g e of l i n e a r programming (LP) development [ I -31 . Nowadays it becomes d i f f i c u l t , even sometimes imposs ib l e , t o make de- c i s i o n s i n l a r g e sys tems and n o t t o t a k e i n t o accoun t t h e con- sequences o f t h e d e c i s i o n over some p e r i o d o f t i m e . Thus, a lmost a l l problems o f o p t i m i z a t i o n become dynamic, m u l t i s t a g e ones. Examples a r e long-range energy , wa te r and o t h e r r e s o u r c e s supp ly models, problems of n a t i o n a l s e t t l e m e n t p l ann ing , long- range a g r i c u l t u r e inves tment p r o j e c t s , manpower and e d u c a t i o n a l p lann ing models, r e s o u r c e s a l l o c a t i o n f o r h e a l t h c a r e , etc.
N e w problems r e q u i r e new approaches. Wi th in DLP it i s d i f f i c u l t t o e x p l o i t on ly LP i d e a s and methods. While f o r t h e s t a t i c LP t h e b a s i c q u e s t i o n c o n s i s t s of de te rm in ing t h e op t ima l p rograv , t h e implementat ion of t h i s program ( r e l a t e d t o t h e q u e s t i o n o f t h e feedback c o n t r o l o f such a program, i t s s t a b i l i t y and s e n s i - t i v i t y , etc . ) i s no less impor tan t f o r t h e dynamic problems. Hence, t h e DLP theo ry and methods shou ld be bo th based on t h e methods o f l i n e a r programming [ 1 , 2 ] and on t h e methods o f c o n t r o l t h e o r y , P o n t r y a g i n ' s maximum p r i n c i p l e [ 4 ] and i t s d i s c r e t e v e r s i o n [ 5 ] i n p a r t i c u l a r .
DLP CANONICAL FORM
I n f o rmu la t i ng DLP problems it i s u s e f u l t o s i n g l e o u t [ 3 ] :
- s t a t e e q u a t i o n s o f t h e system w i t h t h e s t a t e and c o n t r o l v a r i a b l e s ;
- c o n s t r a i n t s imposed on t h e s e v a r i a b l e s ; - p lann ing hor i zon T and t h e l e n g t h o f each t i m e p e r i o d ;
*On l eave from t h e I n s t i t u t e f o r Systems S t u d i e s , MOSCOW, USSR. **To appear i n "Numerical Opt imizat ion of Dynamic Systems"
(Eds. L.C.W. Dixon and G.P. Szeg6) North-Hol land.
- o b j e c t i v e f u n c t i o n (per formance i n d e x ) , which q u a n t i f i e s t h e q u a l i t y o f c o n t r o l .
S t a t e e q u a t i o n s . S t a t e e q u a t i o n s have t h e f o l l o w i n g form:
x ( t + l ) = A ( t ) x ( t ) + B ( t ) u ( t ) + s ( t ) (1 )
where t h e v e c t o r x ( t ) = x t . . . x n t 1 d e f i n e s t h e s t a t e o f
t h e system a t s t a g e t i n t h e s t a t e space X , which is supposed
t o b e t h e n-dimensional e x c l i d e a n space E"; v e c t o r r
u (t) = {u l (t) , . . . , ur ( t ) 1 E E s p e c i f i e s t h e c o n t r o l l i n g a c t i o n
a t s t a g e t ; s ( t) = s t . . s n t 1 i s a v e c t o r d e f i n i n g t h e
e x t e r n a l e f f e c t on t h e system ( u n c o n t r o l l e d , b u t known a p r i o r i
i n t h e d e t e r m i n i s t i c model ) . 0
I t is a l s o assumed t h a t t h e i n i t i a l s t a t e of t h e sys tem i s
given :
PIann inq h o r i z o n T i s supposed t o b e f i x e d . Thus i n ( 1 ) :
t = 0 , 1 , . . . ,T-1. The l e n g t h o f each t i m e p e r i o d may b e t h e
same f o r t h e whole p l a n n i n g ho r i zon ( s a y , month, y e a r , 5 y e a r s ) ,
o r d i f f e r s f o r d i f f e r e n t p e r i o d s . For example, i n long-range
energy supp ly models t h e p lann ing ho r i zon might be 100 y e a r s ,
which forms 10-per iod h o r i z o n w i t h f i v e p e r i o d s e q u a l 6 y e a r s ,
t h r e e p e r i o d s equa l 10 y e a r s and two p e r i d s equa l 20 y e a r s [ 6 ] .
C o n s t r a i n t s . I n r a t h e r g e n e r a l form c o n s t r a i n t s imposed on t h e
s t a t e and c o n t r o l v a r i a b l e s may b e w r i t t e n as
m where f (t) = i f l ( t ) ,. . . , f m ( t ) 1 E E ( t = 0 , 1 , - - = ,T - I )d
O b j e c t i v e f u n c t i o n . (Pe r fo rmance i n d e x ) , which i s t o b e maxi-
mized f o r c e r t a i n t y , i s
I n v e c t o r p r o d u c t s t h e r i g h t vector is a colum.? and t h e l e f t
v e c t o r i s a row.
~ e - f i n i t t c n s . The v e c t o r sequence u = ( u ( O ) , . . . , u ( T - 1 ) 1 i s a
n o n t r o l ( ~ r o g : ? a r n ) o f t h e s y s t e m . The v e c t o r s e q u e n c e
x = ( x ( 0 ) , x (1 ) , . . . , x ( T ) ), which c o r r e s p o n d s t o c o n t r o l u f rom
( I ) , ( 2 ) , i s t h e s y s t e m ' s t r c , i e c t o r y . The p a i r { u , x ) , which
s a t i s f i e s a l l +-he c o n s t r z i n t s o f t h e p rob lem ( e . g . ( 1 ) - ( 4 ) ) i s
a f e a s i b l e p r s c e s s . A f e a s i b l e p r o c e s s { u * , x * > wh ich maximizes
t h e o b j e c t i v e f u n c t i o n ( 5 ) i s o p t i m a l .
The DLP p rob lem i n i t s c a n o n i c a l fo rm is f o r m u l a t e d a s f o l l o w s .
ProbZen; I ? . F i n d a c o n t r o l u* and a t r a j e c t o r y x* , s a t i s f y i n g
t h e s t a t e e q u a t i o n s w i t h t h e i n i t i a l s t a t e a n d t h e con-
s t r a i n t s ( 3 ) and ( 4 ) , which maximize t h e o b j e c t i v e f u n c t i o n ( 5 ) . . I n Problem 1P v e c t o r s xO, s ( t ) , f ( t ) , a ( t ) , b ( t ) and matrices
~ ( t ) , ~ ( t ) , ~ ( t ) , D ( t ) are s u p p o s e d t o be known.
The c h o i c e o f t h e c a n o n i c a l form is t o some e x t e n t a r b i t r a r y
and t h e r e are v a r i o u s p o s s i b l e v e r s i o n s , a n d m o d i f i c a t i o n s of
Prob lem 1P. I n p a r t i c u l a r , b o t h s t a t e and c o n t r o l v a r i a b l e s
may be n o n - n e g a t i v e ; s t a t e e q u a t i o n s may i n c l u d e t i m e l a g s ; con-
s t r a i n t s on s t a t e and c o n t r o l v a r i a b l e s may b e s e p a r a t e , g i v e n
a s e q u a l i t i e s or i n e q u a l i t i e s ; t h e o b j e c t i v e f u n c t i o n may b e
d e f i n e d o n l y by t h e t e r m i n a l s ta te x ( T ) o f t h e s y s t e m , etc.
[ 5 , 7 ] . I t s h o u l d be n o t e d , nowever , t h a t s u c h m o d i f i c a t i o n s c a n
be e i t h e r r e d u c e d t o Prob lem 1P or t h e r e s u l t s s t a t e d below f o r
Problem 1P c a n be e a s i l y e x t e n d e d fo r t h e s e m o d i f i c a t i o n s .
Note, t h a t i f T = 1 , t h e n Prob lem 1P becomes a c o n v e n t i o n a l LP
prob lem. Prob lem 1P i tse l f c a n a lso be c o n s i d e r e d as a c e r t a i n
s t r u c t u r e d LP p rob lem w i t h c o n s t r a i n t s ( 1 ) - ( 4 ) . I n t h i s case
t h e o p t i m a l c o n t r o l P rob lem 1P t u r n s o u t t o b e a n LP p r o b l e m
w i t h t h e s t a i r c a s e c o n s t r a i n t matr ix .
Sone t imes dynamic LP p rob lems are f o r m u l a t e d u s i n g o n l y LP l a n -
guage , a s f o r examp le , P rob lem 2 , [8 ] - [ I 01 :
. - ?~o.?Lern 2 . F i n d v e c t o r s {x* ( 1 ) , . . . , x * ( T ) 1, wh ich m a x i n i z e
s u b j e c t t o
One can a l s o e x p r e s s t h e s t a t e v a r i a b l e s x ( t ) i n Problem 1P a s
an e x p l i c i t f u n c t i o n o f c o n t r o l . A s a r e s u l t , t h e fo l l ow ing LP
problem w i th a b l o c k - t r i a n g u l a r m a t r i x i s ob ta ined .
Problem 3 . Find v e c t o r s {u* ( 0 ) , . . . ,u* (T-1 ) 1 , which maximize
where t h e v e c t o r s h (t) , w ( t ) and t h e m a t r i c e s w ( t , - r ) depend on
t h e known parameters of Problem 1P.
Problems 2 and 3 a r e t y p i c a l examples of s t r u c t u r e d LP &,?d admit
va r i ous mod i f i ca t i ons i n t h e same way a s Problem 1P .(e.g. a block
d iagona l s t r u c t u r e w i t h coupl ing c o n s t r a i n t s and/or v a r i a b l e s ,
e t e . ) . They have been s t u d i e d i n t e n s i v e l y (see, e .g . [1,2,8-101.
But u n l i k e c o n t r o l Problem 1P, such f o r m a l i z a t i o n s of dynamical
problems do n o t use e x p l i c i t l y t h e s t a t e equa t ions o f a system.
Therefore t h i s approach makes it d i f f i c u l t t o a t t r a c t t h e i d e a s
and methods o f t h e c o n t r o l t heo ry . The DLP problems i n t h e form
of Problem 1P w e r e i n t roduced i n [11 ,5 ] .
DLP MODELS
There i s a l a r g e and r a p i d l y expandigg f i e l d o f a p p l i c a t i o n s of
dynamic l i n e a r p r o g r a m i n g . H e r e on ly some of them a r e mentioned.
Z??:alr;ic ntlZt?:-SPC t o r economtc mode 2 s . These a r e b a s i c a l l y exten-
s i o n s of dynamic i npu t -ou tpu t models, when s u b s t i t u t i o n i s
al lowed ( s e e f o r exanp le [ 1 ,2 ,12 -141 ) . A longs ide w i t h d i s c r e t e
t ime f o rmu la t i on con t i nuous t i m e economy models have a l s o been
deve loped ( e . g . i n [ I51 1 .
Enera; n iode is . For long- range p l a n n i n g t h e prob lem i s t h e
f o l l ow ing : f o r e x i s t i n g i n i t i a l s t r u c t u r e o f t h e secondary
energy p r o d u c t i o n c a p a c i t i e s and under g i v e n p r imary energy and
nonenergy r e s o u r c e s supp l y c o n s t r a i n t s f i n d a t r a n s i t i o n t o such
a mix o f t h e energy p r o d u c t i o n o p t i o n s ( f o s s i l , n u c l e a r , e t c . ) ,
which s a t i s f i e d t h e p r o j e c t e d energy demand and min imizes t h e
t o t a l c o s t o f such t r a n s i t i o n . These ene rgy supp l y models have
been s t u d y i n g i n t e n s i v e l y i n t h e r e c e n t y e a r s [6,16-181. The
n e x t s t a g e i n t h i s d i r e c t i o n i s t h e i n v e s t i g a t i o n o f energy-
economy i n t e r a c t i o n . T h i s problem can be a n a l y s e d e i t h e r a s a n
i n t e g r a t e d DLP model o r i n some man-machine i t e r a t i v e mode [ 181 . Pr imary r e s o u r c e s mode 2 s . These models r e l a t e t o t h e p l a n n i n g
o f p r imary r e s o u r c e s e x t r a c t i o n and/or e x p l o r a t i o n a c t i v i t i e s .
The su rvey o f d i f f e r e n t DLP energy-resources-economy models i s
g i ven i n [ I 81 . Wa%er m o d e l s . The DLP models f o r wa te r sys tems are t h e wa te r
supp ly models (which a r e concep tua l y c l o s e t o energy supp l y
models) , t h e prob lems o f a l t e r n a t i v e s e v a l u a t i o n i n a r i v e r
b a s i n , etc . [ I 9-21] . Models .for e d u c a t i o ~ a Z and manpower ? l a n n i n ? . The s i m p l e edu-
c a t i o n a l model a i m s t o f i n d s u c h e n r o l l m e n t s t o d i f f e r e n t edu-
c a t i o n a l i n s t i t u t i o n s which a s w e l l a s s a t i s f y a l l t h e con-
s t r a i n t s of t h e sys tems ( e . g . t e a c h e r s , b u i l d i n g s and o t h e r
e d u c a t i o n a l c a p a c i t i e s a v a i l a b i l i t y ) and be a s close as p o s s i b l e
t o t h e p r o j e c t e d manpower demand. I n t h e manpower p l ann ing
n o d e l s c o n t r o l s a r e r e c r u i t m e n t s and/or p romot ions . t
There a r e d i f f e r e n t m o d i f i c a t i o n s o f t h e s e models , b o t h f o r na-
t i o n a l and i n s t i t u t i o n a l l e v e l s [ 2 2 ] .
., . ...:- - 3 ~ s e tt Z ~ ~ ~ e n t p l a n n i n g . These DLP models re la te t o f i nd -
i n g an o p t i m a l , i n some ' sense, m i g r a t i o n f l ows i n a r e g i o n O r a
cou:1tr-y [ 231 .
3 e ~ Z t h cQre sgstems. The problem o f h e a l t h c a r e sys tem p lan -
n i n g may be reduced t o an a l l o c a t i o n f unds , h e a l t h manpower and
o t h e r r e s o u r c e s o v e r t ime among d i f f e r e n t d i s e a s e s t r e a t m e n t
a c t i v i t i e s i n such a way a s t o y i e l d t h e b e s t r e s u l t s i n terms
o f reduced m o r t a l i t y , m o r b i l i t y and o t h e r l o s e s [ 24 ] .
Agricu Z ture mode is. S e ~ a r a t e DLP models concern p l a n n i n g prob-
l e m s i n l i v e s t o c k b r e e d i n g , c r o p p r o d u c t i o n , r e s o u r c e s u t i l i z a -
' t i o n , e tc . The i n t e g r a t e d models r e l a t e t o p l a n n i n g o f d i v e r s i -
f i e d a g r o - i n d u s t r i a l complexes and t h e problems o f fa rm growth
[25-281 . Production models. These DLP models a r e b a s i c a l l y a s s o c i a t e d
w i t h t h e p r o d u c t i o n s c h e d u l i n g and i n v e n t o r y problems i n
i n d u s t r y [ 8 , 9 , 2 9 , 3 0 ] .
Dynamic t r a n s p o r t a t i o n problems. Transh ipment and o t h e r n e t -
work prob lems a r e t h e i m p o r t a n t c l a s s o f L i n e a r programming.
Dynamical a s p e c t s o f t r a n s p o r t a t i o n prob lems a r e c o n s i d e r e d i n
[3 ,31 ,32 ] .
Miscellaneous models. I t is wor thwh i le t o ment ion' h e r e some
" i n d i v i d u a l " DLP a p p l i c a t i o n s : conges ted u rban road t r a f f i c
c o n t r o l [ 3 3 ] , p l a n n i n g t h e a c q u i s i t i o n and d i s p o s a l o f equipment
i n a t r a n s p o r t f l e e t [ 3 4 ] , m u l t i s t a g e s t r u c t u r a l d e s i g n prob lems
L3.51.
THEORY OF DCP
The t h e o r y o f DLP i s connec ted w i t h d u a l i t y r e l a t i o n s and
o p t i m a l i t y c o n d i t i o n s 11 1 ,36 ,7 ,5 ] . Problem ID. To f i n d a d u a l c o n t r o l X = { A (T-1) , . . . , X ( O ) and
a d u a l t r a j e c t o r y p = { p ( ~ ) , . . . ,p (O) s a t i s f y i n g t h e c o - s t a t e
( d u a l ) e q u a t i o n C
w i t h t h e boundary c o n d i t i o n
s u b j e c t t o t h e c o n s t r a i n t s
and minimize t h e o b j e c t i v e f u n c t i o n
Here p ( t ) IZ En and A ( t ) IZ Ern are Lagrange m u l t i p l i e r s f o r con-
s t r a i n t s ( 1 ) - ( 4 ) .
The d u a l Problem I D i s a c o n t r o l - t y p e problem as i s t h e p r i m a l
one 1P. Here t h e v a r i a b l e A ( t ) i s a d u a l c o n t r o l and p ( t ) i s
a c o - s t a t e o r a d u a l s t a t e a t s t a g e t. W e have r e v e r s e d t i m e
i n t h e d u a l Problem I D : t = T-1, ..., 1,O.
Theorem I . ( T h e DLP NGZobaZ" D u a l i t y T h e o r e m ) . The s o l v a b i l i t y
o f e i t h e r o f t h e 1P o r I D prob lems i m p l i e s t h e s o l v a b i l i t y o f * t h e o t h e r , w i t h Jp (u * ) = JD ( A ) . .
L e t u s i n t r o d u c e Hami l ton f u n c t i o n s
f o r t h e p r i m a l and d u a l p rob lems r e s p e c t i v e l y .
Theorem 2 . (The DLP " L o c a l u D u a l i t y T h e o r e m ) . The s o l u t i o n s
of t h e p r i m a l {u* ,x* ) and d u a l p rob lems are o p t i m a l if
and o n l y i f t h e v a l u e s of Hami l ton ians c o i n c i d e :
Thus, t h e s o l u t i o n o f t h e p a i r of dynamic d u a l p rob lems can be
reduced t o a n a l y s i s o f a p a i r o f s ta t i c l i n e a r programs
max H p ( p ( t + l ) , u ( t ) (10)
G ( t ) x ( t ) + D ( t ) u ( t ) = f ( t ) ; u ( t ) > 0 -
min H D ( x ( t ) , A ( t ) )
l i nked by t h e s t a t e ( I ) , ( 2 ) and c o - s t a t e ( 6 ) , ( 7 ) e q u a t i o n s .
I n p a r t i c u l a r , i t can be shown, t h a t maximum p r i n c i p Z e fo r dua l
Problem I P and minimum p r i n c i p l e f o r dual Problem I D a r e ho ld
[ 5 , 7 ] . An economic i n t e r p r e t a t i o n o f t h e DLP d u a l i t y r e l a t i o n s
i s given i n [361 .
DLP METHODS
We s h a l l d i s t i n g u i s h f i n i t e and i t e r a t i v e methods f o r s o l v i n g
DLP problem.
F i n i t e - s t e p s methods. These methods are t h e ex tens ion o f f i n i t e
l a rge -sca le LP methods f o r t h e dynamic problems. Two b a s i c
approaches may be s i n g l e d o u t here: compact i n v e r s e and decom-
p o s i t i o n methods [9,10,37,38: . Compact i n v e r s e methods. These methods are v a r i a n t s of t h e
s implex method us ing b a s i s f a c t o r i z a t i o n w i t h v a r i o u s s t r a t e g i e s
a s t o t h e v e c t o r p a i r t o en t ,e r and t o l eave t h e b a s i s . The
approach i s be ing developed both f o r s t r u c t u r e d and g e n e r a l LP
( i n t h e l a t t e r c a s e t h e s p a r s e n e s s o f c o n s t r a i n t m a t r i x i s
e x p l o i t e d ) ( s e e e .g . [9 ,10,37,38,39,40] ) . For s t a i r c a s e s t r u c t u r e
(Problem 2 ) t h i s approach was used i n [37,41-441 . For DLP
Problem 1P t h e approach was prcposed i n [45,46] and desc r ibed
i n (47-491, The method a r i s e d i s a n a t u r a l and s t r a i g h t f o r w a r d
ex tens ion o f t h e s implex method. The main concept of t h e s implex
method--the b a s i s - - i s r e p l a c e d by t h e s e t o f l o c a l b a s e s , i n t r o -
duced f o r t h e whole p lann ing pe r iod . The i d e a of t h e method i s
t h e fo l lowing. Cons ider t h e c o n s t r a i n t s ( 3 ) f o r t = O . Using
( 2 ) , they can be w r i t t e n i n t h e form
and accord ing t o t h e convent iona l s implex procedure be p a r t i -
t i oned a s
\.r:rlcre D ( 0 ) i s a s q c a r e n o n - s i n g u l a r ( m m ) m a t r i x . I f T = 1 0
( s t a t i c : p r o b l e m ) , t h e n onr- c a n s e t u1 ( 0 ) = 0 a n d f i n d f r o m ( 1 3 )
a L a s i c f s a s i b l e s o l u t i o n u 0 ( O ) - > 0 . I n t h e dynz?.nic c a s e
(T > 1 ) n o t a l l componen ts o f u 1 ( 0 ) a r e z e r o s f o r b a s i c
s a l u t i o x s . T h e s e n o n z e r o c c m p o n e n t s s h o y ~ l d be t r a n s f o r m e d t o
t h e next s t r p s t > 0 . 3 a r t i t i o n i ~ g t h e s t a t e e q u a t i o n s ( 1 ) f o r
t = 0 i n t h e same way a s i n ( 1 3 ) a n d s u b s t i t u t i n g u o ( 0 ) f r o m
( 1 3 ) t o t h i s p a r t i t i o n e d s t a t e e q u a t i o n s , o n e c a n o b t a i n t h e
c o n s t r a i n t s f o r t h e nexE s t e p i n t h e form, s i m i l a r t o ( 1 2 ) : A
j ( l ) ; ( l ) = 1 ) e ( 1 = , 0 ; ~ ( 1 ) ; . Tne m s t r i x 6(1) depenscis or, G ( 1 1 , C , 3 ( 3 ) a22 car! be a g a i n p a r t i t i o z e d i n t h e
h A O h A
s a n e 1 ~ 2 : ~ : D ( 1 ) = [ D O ( l ) ; D l ( i ) ] , d e t D O ( l ) # 0 and s o o n , u n t i l tr.c l a s t s t e p u l (T-I) = 0 .
h
The s e t o f n l i n e a r l y i n 6 e p e n d e r . t c o l u r r n s of t h e m a t r i x D ( t ) i s
Loccl 3 c s l s a t s t e p t . The s e t of l o c a l b a s e s { D o ( t ) ) f o r t h e
w h o l e p la r , r . i ng p e r i o d t = 0 , 1 , . . . ,T-1 p l a j r s t h e s a m e r o l e a s '
5 a s i s i n s t a n . i z r d s i m p l e x - m e t h o d . The i n t r o 2 u c t i o n o f l o c a l b a s e s
a l l o w s t o i m p l e m e n t a l l s i m p l e x procedures ( s e l e c t i o n o f v e c t o r s
t o b e removed f r a x a n d t o b e i n t r o d u c e d i x t o t h e b a s i s , p r i c i n g ,
u p d a t i n g ) v e r y e f f e c t i v e l y [ 4 9 ] . I t s h o u l d be a l s o n o t e d , t h a t
t h e d y n a m i c s imp lex -me thod g i v e s n o t o n l y c o m p a c t s u b s t i t u t e
f o r b a s i s i n v e r s e , b u t u s e s o n l y a p z r t of t h e l o c a l b a s e s
r e p r e s e n t a t i o n a t e a c h i t e r a t i o n .
The e x t e n s i o n o f t h e dynamic s i m ~ l e x - m e t h o d t o DLP p rob le i ns w i t h
t i m e l a g s o r n c n - n e g a t i v e s t a t e v a r i a b l e s as w e l l a s d u a l ver-
s i o n s o f t h e method are c o n s i d e r e d i n [ 4 6 , 4 9 1 .
2ecor?=,cs i t i o r . ns rkods. T h e s e m e t h o d s f o l l o w s fror;l r e p e a t e d ap- - p l i c a t i o n o f d i f f e r e n t d e c o m p o s i t i o n m e t h o d s ( f i r s t o f a l l , ~ a n t z i c j -
Wol fe d c c o m ; ~ o s i t i o n ~ r i ~ c i 2 l e j t o t h e DLP ~ r o b l e m , and t h e s t a i r c a s e
s t r u c t u r e ( ? r o b l e m 2 ) was d e s c r i b e d i n [ ? ,50 -541 .
F o r D L ? P r o b l e m 1P t h i s a 2 p r o a c h was u s e d i n [ S S ] . C o n s i d e r t h e
s e q u e n c e o f LP p r o b l e m s ( t = T- 1 , . . . , I , 0 ) :
max p ( t + l ) x ( t + J )
where R (xo) i s t h e s e t of a l l s t a t e s x ( t ) f e a s i b l e from i n i t i a l t s t a t e x0 f o r t s t z p s [ 5 1 ; p ( t + l ) i s d u a l s t a t e v e c t o r .
So lu t i ons o f t h e s e LP problems w i th op t ima l va lues of p * ( t + l )
g i v e an op t ima l s o l u t i o n o f t h e o r i g i n a l Problem 1P. Any v e c t o r
x ( t ) of R (xo) can be r e p r e s e n t e d a s a l i n e a r convex combina- t t i o n o f i t s extreme p o i n t s (assuming boundness o f R ( x o ) ) . t Thus, problem (1 4 ) can be r e w r i t t e n as
max p ( t + l ) [ L A ( t ) x i ( t ) p i ( t ) + B ( t ) u ( t ) + s ( t ) l i -
Problem (15) i s a master problem a t s t e p t. For i t s s o l u t i o n
convent ional column g e n e r a t o r procedure can be adopted. Other
methods f o r s o l u t i o n o f Problem 1P based on d i f f e r e n t decomposi-
t i o n and p a r t i t i o n p r i n c i p l e s w e r e cons idered i n [56-601.
The DLP i t e r a t i v e m e t h o d s . The a p p l i c a t i o n o f t h e LP f i n i t e
methods t o t h e dynamic problems may cause c e r t a i n d i f f i c u l t i e s ,
e s p e c i a l l y f o r l a r g e p lann ing peri'ods T. The reason f o r it i s
t h a t s o l u t i o n p a t h s i n t h e s e methods c o n s i s t o f a number o f
v e r t i c e s of a f e a s i b l e po l yhedra l set ( i n some s p a c e ) , and t h i s
number w i l l i n c r e a s e e x p o n e n t i a l l y w i t h T.
The i t e r a t i v e methods seem t o by-pass t h e s e d i f f i c u l t i e s . They
a l s o c h a r a c t e r i z e d by low demand t o computer ' s co re memory,
s i m p l i c i t y of t h e computer implementat ion, low s e n s i t i v i t y t o
d i s tu rbances . Disadvantages of t h e s e methods may b e low con-
vergence r a t e , y e i l d i n g o n l y approximate s o l u t i o n . .\
We s h a l l s i n g l e o u t t h e fo l l ow ing i t e r a t i v e methods.
; 1 L i : r ~ c r ~ ! ! : ~ n t m?t i ;oJ ; ; . T h i s g r o u p o f m e t h o d s i s bclscd -- __ I- o n f i n d i n g ex t remum o f f u n c t i o n s :
' (:, p = max L ( u f x ; A , p ) ; ; ( u , ~ ) = min L ( u f x ; ; \ ,PI ( 1 6 ! X f U L 0 p i x > 0 -
w h e r e L ( u , x , A , p ) i s t h e L a g r a n g e f u n c t i o n o f P r o b l e m 1P ( I D ) .
M i n i m i z a t i o n o f Y i s e q u i v a l e n t t o t h e s o l u t i o n o f t h e d u a l
P r o b l e m ID, w h i l e t h e m a x i m i z a t i . o n o f c$ i s e q u i v a l e n t t o t h e
s o l u t i o n o f p r i m a l P r o b l e m 1P [ 7 ] . A s t h e s e f u n c t i o n s a r e non-
d i f f e r e n t i a b l e b y n a t u r e , t h e g e n e r a l i z e d g r a d i e n t t e c h n i q u e i611
is n e e d e d . The a p p l i c a t i o n o f t h i s a p p r o a c h t o t h e s o l u t i o n o f
d u a l P r o b l e m ID l e a d s t o t h e f o l l o w i n g a l g o r i t h m . W e c o n s i d e r
P r o b l e m 1P w i t h a d d i t i o n a l c o n s t r a i n t s o n c o n t r o l v a r i a b l e s :
u ( t ) EUt ; U t = ! u ( t ) I ~ ( t ) u ( t ) 2 r ( t ) , u ( t ) 2 0 ) .
v ( 1 ) C h o o s e a r b i t r a r y d u a l c o n t r o l { A ( t ) j ( t = 0 , 1 , . . . ,T-1 ) . ( 2 ) Cernpute d u a l s t a t e t r a j e c t o r y ~ ~ ' ( t ) ; f r o m t h e d u a l
s t a t e e q u a t i o n s ( 6 ) w i t h b o u n d a r y c o n d i t i o n ( 7 ) . v
( 3 ) F o r p\J ( t + l ) s o l v e T LP p r o b l e m s : rnax H p ( p ( t + l ) .u ( t ) ) . u ( t ) E U t . L e t uV ( t ) b e a s o l u t i o n o f t h e s e p r o b l e m s .
v ( 4 ) Compute p r i m a l s t a t e t r a j e c t o r y x ( t l f r o m ( 1 ) and
( 2 ) f o r u V ( t ) .
v v (5) Compute g e n e r a l i z e d g r a d i e n t o f Y : S Y ( A ) = h ( t ) , w h e r e h v ( t ) = f ( t ) - ~ ( t ) x ' ( t ) - ~ ( t ) u \ ; ( t ) . ( 1 7 )
v+ 1 ( 6 ) Compute new d u a l c o n t r o l A : . v+l v v v
= A - a ( ' 3 ) I ( ( V = O , l , 2 . . . . ) ( 1 8 )
- - - * r . . 3 : 1 i . ? ~ : . L e t n V - t O ; v - + ~ ; - . 1 a v - -. Then ?' ( i V ) -+ ! ' (A ) ,
:>=o * * *
) ~ ' + . I , 1 = {A ( t ) is a n o p t i m a l c o n t r o l o f d u a l P r o b l e m I D .
The a l g o r i t h m ( 1 ) - ( 6 ) g i v e s a d u a l s o l u t i o n o f t h e p r o b l e m . To
o b t a i n t h e p r i m a l s o l u t i o n o n e c a n a p p l y t h e same a p p r o a c h t o
f u n c t i o n : ( u ) i n ( 1 6 ) .
,,--,:*.c.. - - - ... , .. -, L Q C L : : n 3 2 i i :C ~ u Y ~ c ? . ~ s ; : . - : e z , : z : [ 6 2 631 . T h i s a p p r o a c h com-
b i n e s t h e a d v a n t a g e s o f b o t h f i n i t e a n d i t e r a t i v e m e t h o d s : u n d e r
ain imal requirements f o r s t o r a g e i t ensures monotonic conver-
cjsnce i n a f i n i t e number of s t e p s . I n a sense , t h i s method c-n be
considered a s a "smooth" a l t e r n a t i v e t o t h e nonsmooth approach,
dsscr ibeci above. Le t t h e L a g r a ~ q e f ~ i c t i o n 5 ( u , ) . ) o f Problsm 1 2
( I D ) be ailqnented by q ! ~ a d r a t l c te~rix:
where 0 > 0 , v e c t o r h ( t ) i s d e f i n e d from ( 1 7 ) .
Now i n s t e z d of ( 1 6 ) t h e fo l l ow ing p a i r of problems can . be
cor?s iders2
\U ( A ) = rriax i?(u ,A) ; t , ( u ) = in M(u,X! f1 . -
U X
Opposi te t o ?' ( A ) , t h e f u ~ c t i o n P (?, j i s a s a ~ o t h concave m f u n c t i o n ; d e r i v a t i v e o f '? ( A ) i s a con t inuous f7 t?c t i on of X m d m a'? ( E , ) / a A ( t ) = h ( t ) where h ( t ) i s computed from (171, when M
u* = ui(A) i s a s o l u t i o n o f t h e l e f t problem i n ( 2 0 ) (621.
Problems n i n Y ,(1) and max $ ( u ) a r e t h e p a i r of modi f ied d u a l )I M
~ r o b l e ~ . s , a s s o c i a t e ? w i t h o r i g i n a l Problem 1P. I n [621 i s
shown, t h a t t h e m o d i f i c a t i o n does n o t chznge t h o g e n e r a l d u a l
r e l a t i o n s . Thus i;l o r d e r t o o b t a i n a s o l u t i o n o f Troblem
one can sol ire e i t h e r t h e d u a l p r o b l e ~ min ? 0.) with t h e no2-
smooth fu-rlction Y o r t h e modi f ied Zual p rob len n i n ? ( A ) w i t h >! t h e smooth fuirlction !'
M'
The l a t t e r g i v e s t h e f o l l o w i n s a lgor i thm:
( 1 ) Choose a r b i t r a r y d u a l c o n t r o l A V .
( 2 ) Solve t h e l e f t p r o b l e n i n ( 2 0 ) .
Th is ?robLen has l i n e a r s t a t e e q u a t i o n s ( 1 ) , g u a d r a t i c ob jec-
t i v e f u n c t i o n ( 1 9 ) and s imp le c o n s t r a i n t s on c o n t r o l u ( t )
( e . 9 . I) - < u(t) - < c ( t ) 1 . There fo re f o r i t s s o l u t i o n o p t i n a l
c o n t r o l t2chnique (iv-hich reduces i n t h i s c a s e t o s o l u t i c n of 2
n a t r i x A i c c a t i e q ~ a t i o n ) can be e f f e c t i v e l y 2 ~ 2 l l e d [ 6 3 1 .
v ( 3 ) L e t u b e a s o l u t i o n o f t h e abo l re p r o b l m . C c m ~ a t e
h" ( t ) f r o m ( 1 7 ) f o r t h i s uV and new d u a l c o n t r o l X v+ 1
f r o m
p + 1 ( t ) = X v ( t ) - 4 h v ( t ) ( \ J = 0 , 1 , 2 , . - . )
T k e o r ~ s , ~ 4 . F o r a n y e > 0 ,
* * and b e g i n n i n g f r om some f i n i t e vO : X v €11 , u;' EU , v > \ J ~ ,
& -
I
where A T I U-, a r e s o l u t i o n sets o f d u a l a n d p r i m a l p r o b l e m s
I D and 1P r e s p e c t i v e l y .
Theorem 4 s t a t e s t h a t , o p p o s i t e t o t h e g e n e r a l i z e d g r a d i e n t a l g o r i t h m ( 1 ) - ( 6 ) , t h e a l g o r i t h m ( 1 ) - ( 3 ) c o n v e r g e s f o r a f i n i t e number o f s t e p s a n d t h e v a l u e s o f t h e m o d i f i e d f u n c t i o n '? ( A ) m o n o t o n o u s l y d e c r e a s e s w i t h v ( i n f a c t so d o e s t h e n o n n o d y f i e d d u a l f u n c t i o n Y ( A ) a l s o 1621 ) . M o r e o v e r , t h i s a l g o r i t h m g i v e s s i m u l t a n e o u s l y d u a l A * and ?r imal u* o p t i m a l s o l u t i o n s .
CONCLUSION
A s h o r t s u r v e y h a s b e e n g i v e n a b o v e o f DL? m o d e l s , t h e o r y and m e t h o d s . T h e r e i s n o t much l i t e r a t u r e o n c o m p u t e r imp lemen ta - t i o n o f t h e m e t h o d s . E x p e r i m e n t s , h o w e v e r , show t h a t t h e s e me thods c a n b e much more e f f i c i e n t i n c o z p a r i s o n w i t h t h e s t a n d a r d ( s t a t i c ) a p p r o a c h (see, f o r i n s t a n c e , [ 3 9 , 5 3 1 ) . T h e r e - f o r e c o m p u t e r t e s t i n g and e v a l u a t i o n o f t h e a l g o r i t h m s a r e now a n i m p o r t a n t p r o b l e m . O t h e r i m p o r t a n t d i r e c t i o n s o f r e s e a r c h a r e t h e p r o b l e m o f t h e i m p l e m e n t a t i o n o f o p t i m a l c o n t r o l a n d p o s t - o p t i m a l a n a l y s i s o f t h e mode l ( f e e d b a c k v s p rog ram o p t i m a l c o n t r o l , s e n s i t i v i t y a n d s t a b i l i t y a n a l y s i s , i n t e r r e l a t i o n be- tween s h o r t and l o n g - t e r m p r o b l e m s , e t c . )
I t s h o u l d a l s o b e n o t e d t h a t n o t a l l d y n a i c a l o p t i m i z a t i o n p r o b l e m s c a n be k e p t w i t h i n t h e f ramework o f DLP. T h e r e f o r e , t h e e x t e n s i o n o f DLP f o r s t o c h a s t i c , maxmin, n o n l i n e a r dynamic p r o g r z m i n g p r o b l e m s i s o f l a r g e p r a c t i c a l i n t e r e s t . Some w o r k h a s S e e n a l r e a d y d o n e i n t h i s d i r e c t i o n (see, e . g . [64 -6611 .
Refe rences
Dan tz i g , G.B., Lineal2 Programming and E x t e n s i o n s , P r i n c e t o n U n i v e r s i t y P r e s s , P r i n c e t o n , N . J . , 1363.
K a n t o r o v i t c h , L .V . , -The B e s t Use o f Economic Resources , Harvard U n i v e r s i t y P r e s s , Cambridge, Mass., 1965.
P r o p o i , A . I . , Problems o f Dynamic L i nea r Programming, Rb;-76-78, I r i t e r n a t i o n a l I n s t i t u t e f o r App l ied Systems A n a l y s i s , Laxenburg, A u s t r i a , 1976.
P o n t r y a g i n , L.S., e t a l . Mathemat ica l Theory o f Opt imal P rocesses , I n t e r s c i e n c e , New York, 1962.
P r o p o i , A . , Elementy T e o r i i Opt imalnykh D i s c r e t n y k h P r o t s e s s o v (E lements o f t h e Theory o f Opt imal Discrete S y s t e m s ) , Nauka, Moscow, 1973 ( i n R u s s i a n ) .
Markuse, W . , e t a l . A Dynamic Time-Dependent Model f o r t h e : A n a l y s i s o f Alternative Energy P o l i c i e s , i n Opera t i ona l Research ' 7 5 ( e d . K.B. Ha ley) , North- Ho l land P u b l i s h i n g Company, 1976, pp. 647-667.
P r o p o i , A., Dual Sys tems o f ~ y n a h c ine ear Programming, RR-77-9, I n t e r n a t i o n a l I n s t i t u t e f o r App l i ed Systems A n a l y s i s , Laxenburg, A u s t r i a , 1977.
Dan tz i g , G.B., On t h e S t a t u s o f M u l t i s t a g e L i n e a r Program- ming Prob lems, Management S c i e n c e , Vol. 6 ( 1959 ) , pp. 53-72.
Lasdon, L. S ., O p t i m i z a t i o n Theory f o r Large Sys tems , MacP-lillan, 1970.
I~ladsen;O., S o l u t i o n o f LP-Problems w i t h S t a i r c a s e S t r u c - t u r e , RZ-26-1977, I n s t i t u t e o f Mathemat i ca l S t a t i s - t i c s and O p e r a t i o n a l Research , Dennark, 1977.
I v a n i l o v , Yu. P. and A.I. P r o p o i , The L i n e a r Dynamic Pro- gramming, DokZ. Acad. Nauk SSSR, Vol. 198 , No . ' 5 (1971) pp. 1011-1014 ( i n R u s s i a n ) .
Aganbegian, A.G., K.A. B a g r i n o v s k i i and A.G. Granberg, Systema Modele i Narodnohoz ias tvennogo P l a n i r o v a n i a (Systems of Models o f N a t i o n a l P l a n n i n g ) , Moscow, "Idys Z " , .I972 ( i n Russ ian ) .
Dan tz i g , G.B., Op t ima l s o l u t i o n of a ~ y n a m i c L e o n t i e f Flodel w i t h S u b s t i t u t i o n , Econometrics, Vol . 23 (1955) , pp.295-302.
Koehler , G . , A. Whinston and G. Ylright, O p t i m i z a t i o n Over L e o n t i e f S u b s t i t u t i o n Sys tems , North-Holland Pub l i sh ing . Company, 1975.
Dukalov, A. N . , Yu.N. Ivanov and V.V. Tokarev, Cont ro l Theory and Zconomic Systems. Avtomat ika i Te Zemekhanika, No. 5 (1974) ' ( i n Russ ian) .
t lakarov A.A. and L.A. Melent jev , Metody I s s l e d o v a n i a i O p t i n i z a t z i i Energet icheskaso Khoziastva (Ana lys is and O p t i n i z a t i o n Methods o f Energy Sys tens) , Nauka, Novosib i rsk , 1973 ( i n Russ ian) .
Haefe le , W. and A.S. Manne, S t r a t e g i e s o f a T r a n s i t i o n from Foss iZ t o NucZear Fue l s , RR-74-07, I n t e r n a t i o n a l I n s t i t u t e f o r A?plied Systems Ana lys i s , Laxenburg, A u s t r i a , 1974.
Propo i , A. and I. Zin in , Energy-Besources-Economy DeveZop- ment ModeZs, I n s t i t u t e f o r Appl ied Systems Ana lys i s , . Laxenburg, A u s t r i a , 1978 ( fo r thcoming) .
Agarkov, S. , I . Druzhin in , 2. Konovalenko, A. Kuznetsov, T. Naumova, V. Ushakova, I. Gera rd i , N . Dorfman, R. Kogan, E. Kokorev and D. R ~ s ~ u ~ o v , The Optima2 A l l o c a t i o n o f Water Consuming P roduc t i on . Vodnye Resursy (Water Resources) , Vol. 3 (1957) , pp. 5-17.
Pa r i kh , S.C.,L inear Decomposit ion Programming o f Optimal Long-Range Operat ion o f a M u l t i p l e Mul t ipurpose
. Reservo i r System, i n The Proceed ings o f t h e Four th I n t e r n a t i o n a l Con fe rence on Operat ionaZ Research , (eds. D . Hertz and I. :.felese) Wiley, 1966.
pp. 763-777.
Biswas, A . K . , (Ed. Sys tems Approach t o Water Management, McGraw-Hill P u b l i s h i n g Company, 1976.
Propo i , A . , Models fo r Educa t i ona l and Manpower P lann ing : A Dynamic L i nea r Programming Approach, RM-78-20, ' I n t e r n a t i o n a l I n s t i t u t e f o r Appl ied Systems Iu la l ys i s , . Laxenburg, A u s t r i a , 1978.
Propo i , A. and F. ~ 7 i l l e k e n s , A DLP Approach t o Na t i ona l S e t t l e m e n t Sys tem P lann ing , RE-77-8, I n t e r n a t i o n a l I n s t i t u t e f o r Appl ied Systems Ana lys i s , Laxenburg, A u s t r i a , 1977. A lso see Envi ronment and PZanning A , Vol. 10 (1978) , pp. 561-576.
Propo i , A . , Dynamic L inea r Programming Models i n Hea l th Care Systems i n Model ing Hea l t h Care Sys tems , e d s . E. Shigan and R. Gibbs, CP-77-8, I n t e r n a t i o n a l I n s t i t u t e f o r Appl ied Systems Ana lys i s , Laxenburg, A u s t r i a , 1977.
Swart , PI. a.0. Expansion Planning f o r a Large Da i ry Farm, i n S t u d i e s i n L i n e a r Programming, Eds. H . S a l k i n and J.' Saha, North-Holland Pub l i sh ing Company, N . Y . , 1975.
P r o p o i , A . , Dynamic L i n e a r Progrcmming ModeZs f o r L i v e s - s t o c k Farms, 311-77-29, I n t e r n a t i o n a l I n s t i t u t e f o r App l ied Systems A n a l y s i s , Laxenburg, A u s t r i a , 1973.
C s a k i , C . , Dyncmic L i n e a r Programming Model f o r Ag r i cuZ - tu raZ I n v e s t n e n t and R e s o u r c e s U t i l i z a t i o n P o Z i c i e s , RM-77-36, I n t e r n a t i o n a l I n s t i t u t e f o r App l ied Systems A n a l y s i s , Laxenburg, A u s t r i a , 1977.
C a r t e r , H . , C . Csak i and A . P r o p o i , P l a n n i n g Long-Range Agr icuZtunaZ I n v e s t m e n t P r o j e c t s : A Dynamic L i n e a r Progranming Approach, W-77-38, I n t e r n a t i o n a l I n s t i t u t e f o r App l i ed Systems A n a l y s i s , Laxenburg, A u s t r i a , 1977.
G lassey , C . R . , Dynamic L i n e a r Programs f o r P r o d u c t i o n Schedu l i ng , O p e r a t i o n a l R e s e a r c h , V o l . 1 9 , 1 (1971) . pp. 45-56.
. -
P e g e l s , C . C . , S y s t e m s AnaZysi is f o r P r o d u c t i o n O p e r a t i o n s , .Gordon and Breach P u b l i s h i n g Company, London, 1976.
Be l lmore, PI. , W".. Ek lo f and G.L. Nemhauser, Transh ipment A lgo r i t hm f o r a M u l t i p e r i o d T r a n s p o r t a t i o n Problem, Naval R e s e a r c h L o g Z s t i c s Q u a r t e r l y , V o l . 19 , 4 ( 1969 ) .
Madsen, O . , A Case Study i n Problem S t r u c t u r i n g , i n Second European Congress on O p e r a t c o n a l R e s e a r c h , (ed. M . Roubens) , North-Hol land P u b l i s h i n g Company, 1976, pp. 287-293.
Tamura, H. M u l t i s t a g e L i n e a r Programming f o r Discrete Opt imal C o n t r o l w i t h D i s t r i b u t e d Lags. A v t o m a t i c a , V o l . 13 (1977 ) , pp. 369-376.
N e w , C. C. , T r a n s p o r t F l e e t P l a n n i n g f o r M u l t i p e r i o d Ope ra t i ons . O p e r a t i o n a l R e s e c r c h Q u a r t e r l y , Vol. 26, 1 , ii ( 1975 ) , pp. 151-166.
H o t J . K . , Opt imal Des ign of Mul t i -S tage S t r u c t u r e s : a Nested ~ e c o m p o s i t i o n Approach. Computers and S t r u c t u r e s , V o l . 5 ( 1 9 7 5 ) , pp. 249-255.
I v a n i l o v , Pu. P. and A . P r o p o i , D u a l i t y R e l a t i o n s i n Dynamic L i n e a r Programming, A v t o m a t i c a i T e l e m e k h a n i k a , V o l . 12 , (1973) , pp. 100-108 ( i n Russ ian ) . _..
B u l a v s k i i , V.A . , X.A. Zv iag ina and ;i.A. Yakovleva, Ch i s l ennye Metody L ine inogo Programrnirovania (Numer ical Methods o f L i n e a r Programming) , "Nauka I f ,
bloscow, 1977 ( i n Russ ian ) . Beer, K., Ldsung Grosser L i n e a r e r ~ p t i m i e r u n g s a u f g a b e n ,
V E B ~ e u t s c h e r V e r l a g d e r W i s s e n s c h a f t e n , B e r l i n , 1977 ( i n German) .
Wink le r , C . , Bas i s F a c t o r i z a t i o n f o r B lock Angular L inear Programs: U n i f i e d Theory o f P a r t i t i o n i n g and Decomposi t ion Using t h e S imp lex Method, RR-74-22, I n t e r n a t i o n a l I n s t i t u t e f o r App l i ed Systems A n a l y s i s , Laxenburg, A u s t r i a , 1974.
G i l l , P.E. and W. Murray, L i nea r l y -Cons t ra i ned Problems I n c l u d i n g L i n e a r and Q u a d r a t i c Programming, i n The S t a t e o f t h e A r t i n Numerical A n a l y s i s , ( e d . D . J a c o b s ) , Academic P r e s s , 1977, pp. 313-363.
Dan tz i g , G.B., Compact B a s i s T r i a n g u l a r i z a t i o n f o r t h e S implex Method i n Recent Advaxces i n Mathemat ica l Programming (Eds. R. Graves and P. Wolfe) , I - lcGraw- H i l l P u b l i s h i n g Company, N.Y . , 1963 , pp. 125-132.
Heesterman, A.R.G. an'd J. Sandee, S p e c i a l S implex A lgo r i t hm f o r L inked Prob lems, Management S c i e n c e , V o l . 11 , N o . 3 (1965) pp. 420-428.
K a l l i o , M. and 'E. P o r t e u s , T r i a n g u l a r F a c t o r i z a t i o n and G e n e r a l i z e d Upper Bounding Techn iques , Opera t i ons Research, V o l . 25, 1 (1977) , pp. 89-99.
W o l l m e r , R . D . , A S u b s t i t u t e I n v e r s e f o r t h e as is 6 f a S t a i r c a s e S t r u c t u r e L i n e a r Programs, Mathemat ics o f Opera t ions Research , Vol . 2, ' 3 . . ( 1977 ) , pp. 230-239.
Kr ivonozhko, V . I . and A . I . P r o p o i , A Method f o r L i n e a r Dynamic Programs w i t h t h e U s e of B a s i s M a t r i x F a c t o r i z a t i o n , i n A b s t r a c t s o f t h e I X I n t e r n a t i o n a l Symposium o f Mathemat ica l Programming, Budapes t , August 1976.
Kr ivonozhko, V . I . and A . I . P r o p o i , S implex Method f o r Dynamic L i n e a r Program S o l u t i o n , i n O p t i m i z a t i o n Techn ique , Proceed ings o f t h e 8 t h IF IP Cox fe rence , Kflrzburg 1977, (ed. J. S t o e r ) , Spr i nge r -Ve r l ag , 1978,
P r o p o i , A. and V. Krivonozhko, The Dynamic S imp lex Method, RM-77-24, I n t e r n a t i o n a l I n s t i t u t e for App l i ed Systems A n a l y s i s , Laxenburg, A u s t r i a , 1977.
Kr ivonozhko, V.E. and A.I . P r o p o i , C o n t r o l S u c c e s s i v e Improvement Method i n Dynamic L i n e a r Programming Problems I , 11, I z v e s t i a AN SSSX, Techn icheskya K i b e r n e t i k a , V o l . 3 ,4 , 1978, pp. 5-16,5-14 ( i n X u s s i a n ) .
P r o p o i , A. and V. Krivonozhko, S imp lex . Method f o r Dynamic L i nea r Programs, RR-78-14, I n t e r n a t i o n a l I n s t i t u t e for App l i ed Systems A n a l y s i s , Laxenburg, A u s t r i a .
Cobb, R.H. and J. Cord, Decompcsi t ion Approaches f o r S o l v i n g L inked Programs, in ProcceZings o f t h e P r i n c e t . 0 ~ Symposium on b iat lzenat ica l Programming ( ed . H.W. Kuhn), P r i n c e t o n u n i v e r s i t y P r e s s , N . J . 1967.
[51] Ho, J . K . and A.S. Manne, N e s t i d Decomposi t ion f o r Dynamic Models, Mathemat ica l Programming, Vol. 6 , 2 ( 1974 ) , pp. 121-140.
[52] G lassey , C . R . , Nested Decomposit ion and Mu l t i -S tage L i n e a r Programs, Management S c i e n c e , Vol. 20, pp. 282-292.
[53). Ho, J . K . Nested Decomposit ion o f a Dynamic Energy Model, Ma?zagement and S c i e n c e , Vol. 23, ' 9 (1977) , pp. 1022-1026.
[54] K a l l i o , M . , Decomposi t ion o f ~ r b o r e s c e n t L i n e a r Programs, Mathemat ica l Programming, Vol. 13 (1977) , pp. 348-356.
[55] Kr ivonozhko, V.E., Decomposit ion A lgo r i t hm f o r Dynamic L i n e a r Programming Prob lems, I z v e s t i a AN SSSR, Techn icheskya K i b e r n e t i k a , Vol. 3 (1977) , pp. 18-25 ( i n R u s s i a n ) .
[56] I v a n i l o v , Yu. P. and Yu. E. Malashenko, Method f o r L i n e a r Dynamic P r o g r a m i n g S o l u t i o n , ~ i b e r n e t i k a , Vol . 5 ( 1 9 7 3 ) , pp. 42-49 ( i n R u s s i a n ) .
[57] Be lukh in , V.P., P a r t i t i o n P r i n c i p l e and Pr imal-Dual Method f o r Dynamic L i n e a r Programming Prob lems. Zhurnal ; ' 3 c h i s l i t e l n o i Ma temat i k i i Matemat i chesko i P h y s i k i , Vol. 1 5 , 6 (1975) , pg. 1424-1435.
[58] Be lukh in , V.P., P a r a m e t r i c Method f o r Dynamic L i n e a r Programming S o l u t i o n . Avtomat i ca i Te Zemekhanika, Vol . 3 (19751, pp. 95-103.
[591 Aonuma, T., A Two-level A lgo r i thm f o r Two-stage L i n e a r Programs. Jou rna l o f t h e Ope ra t i ons Research S o c i e t y o f gapan, Vol . 21, 2 ( 1 9 7 8 ) , pp. 171-187.
[601 Fukao, T., A Computat ion Method f o r Dynamic L i n e a r Programming, Journa l o f t h e Ope ra t i ons Research S o c i e t y o f Japan, Vol . 2 , 3 ( 1960 ) .
- -. - -- - - - .
[611 Nonsmooth O p t i m i z a t i o n , P roceed ings of IIASA Workshop, March-Apri l 1977, eds . C. Lemarecha l , R. M i f f l i n , Pergamon P r e s s , 1978.
[62] P ropo i , A . I . and A.B. Yadykiri, Modi f ied Dual R e l a t i o n s i n Dynamic L i n e a r Programming, Avtomat ika i Tezemekhan ika , 5 ( 1 9 7 5 ) , pp. 106-1 14 ( i n R u s s i a n ) . -
. . -. . . - 1631 P ropo i , A . I . and A.B. Yadykin, P a r a m e t r i c I t e r a t i v e
!.lethods f o r Dynamic L i n e a r Programming. I . Non- d e g e n e r a t e C a s e , Avtomat ika i Te lemekhcn ika , Vol. 12 ( 1 9 7 5 ) , 11. Genera l C a s ~ . Lv tomat i ka i Te lemekhan ika , Vol . 2 ( 1 9 7 6 ) , pp. 127-135, 136-144.
[ 6 4 ] Ermol j e v , Yu. 14. , Metody S tokhas t i cheskogo Prograrnmirovania (Methods o f S t o c h a s t i c Programming), Nauka, M o s c o w , 1976 ( i n R u s s i a n ) .
[65] I v a n i l o v , Yu.P. and A . I . P r o p o i , Convex Dynamic Program- ming. Zhurnal Vychis lite Znoi Matematiki i Matematicheskoi Phyziki, V o l . 1 2 , 3 (1972) ( i n R u s s i a n ) .
[ 6 6 ] P r o p o i , A . I . and A.B. Yadykin, The P lann ing Under Unce r ta i n Demand, 1,II. Avtomatika i Telemekhanika, Vol. 2 , 3 (1974 ) , pp. 78-83, 1 0 2 - 1 0 8 . ( i n R u s s i a n ) .