Parameter Estimation with Constraints Based on Variational Method

J Ma n c Ap 1 2 1: 0—0 . r e Si p. 00 9 1518 i ( Do I 1 0 7 s 1 0 . 0 9 0—: 0 1 0/ 1 8 4 0l— 0 2 3

上 ara e e s i at w - ' J ons- ai s 1 h, m t r… t E m i on i n t t‘ r nt

Ba e n V a i to lM e ho s d o r a i na t d We . i i n m ng Sh P s r d aeB ia eNa a u ma ie a e y Qig a 6 0 1 Chn ot a u t r d, v l b rn d m, n d o2 6 7, ia g g S Ac Absr c: Th c u a y f paa ee e t a in s rtc lw h n iial o e ig s i ta t e a c r c o r m t r si to i ciia e dgt ly m d ln a hp.A a a ee m p rm tr e tm ain m eh d wih c n tan sw a e eo d si to t o t o sri t s d v lpe .ba e h a ito 1m eh .Pe f r n e f ncins s d on t e v ra ina tod ro ma c u to

a d c n t ite u t n h a it n l t o r o sr ce y a ay ig i p t n up t q ain f n o sr n q ai si te v r i a h d ae c n tu td b n l zn n u d o t u u t so a o n ao me a e o t yse .Th r blm fp r me e si ain w a r n f r e n o ap o lm fla ts uae si ain. hes tm e p o e o a a tre tm to sta so m d it r be o e s q r se tm to Th a a tre tm ain e uain sa ayz d i r e o g ta o tm ie e t a in ofp r me esba e e p r me e si to q t o wa n l e n o d rt e n pi z d si to a a tr s d m o t La rng m ulplc to o rt r Sm u ai r s ls h n he ga e t i ain pe ao . i lton e ut s owe t t li m e

h i b te t a te i d ha tl s tod s et r h n h ta ii a e s q a e si to r dton lla ts u r s e tma in,pr d i g a hgh rprc so o ucn i e e iin whe de tl i g paa ee s t h s ey n i n iy n r m t r.I a v r i p ra r ci a auei r a ppia in uc ss tm ntfc to n r m ee si ai n. m o tnt a tc lv l n ae sofa lc to s h a yse i p de i ain a dpaa t re t i m to K e w o ds 1a ts uae si ai n; a a e e si ai; raina eh;c n tan y r:e s q r se t m to p rm tre tm t on va it o l m tod o sr i t

Aril I l 7一4 32 l) 1O O一4 t e D: 6 l9 3 (0 oo一 15O c

1I t O uc i n n r d t0 T a i o a la t s u r s me h d h s c mp e e sv r l d t n l e s q a e i t o a o rh n ie a p i ai n y tm e t c t n a d p r m ee si t n p l to si s se i n i a i n a a tre t c n d i f o ma i . o

I q()“a d r ep r c s s m p t n up t nE ., n aet ef ty t i u do tu 1 y h e e n a r s e t ey a d e p c i l, n a, v e p c e si t . x e t d t e tma e o a et epa a ee swh c r r r m tr i h a e h

Bu i d s d a t g i t a e tma e p r m e e s a e t t s ia v n a e s h t si t d a a tr h v ay s mp o i i swh n i p t n u p td t r o tm i a e t t b a e u d o t u aa a e c n a n t d c n a

∑ 7 a d 7 ) n ( ( ) ae sp o e t b te rn o r u p sd o e h a d m ,L

b oss n Xu(9 6,Hat gJme n a e(9 9, y

n i .I e 1 8 ) s n—a sa d S g 16 ) i Hs 1 7 ) hn ad Wa g(0 3,teg n rl e e s i 7,C e n n 2 0 ) h e eai d l t a(9 z a s u e m eh d wa i to u e t o e c me t i q a s r to s nrd cd o v ro hs d s d a t g .b ti e u tc n n t g a a t e a si t n ia v n a e u t r s l a o u r n e n e tma i s o

i t re e ta o s e il,a d t e t e r n o v ra l n e f r n i l n ie s ra s n o b h a d m a ib e

一

s raswh s e n r eo a d d srb t g i n ta h d eil o em a sae z r n itiu i su atc e n

a d ie t a. z a d ( ) a etern o v r be f n ni 1 () n d c r d m ai ls h a a o t e s se o t u n n u ih a e o sr e cu l . h y tm up ta d ip twh c r b e v d a t al y An h I dt e】

wi oay tt is L dFse, 9 Diga dY n, t n smpoi ba ( ua i r 18; n a g h c n h 9 n 1 9 a Dig ad Y n,19 b . arn e eu t n hc 9; n n a g 9 ) L gag q ai,w ih 9 9 o me t e s se i p t n u p t i e t b ih d i i a e e st y t m u d o t u, s s a ls e t sp p L h n a n h I i s d a h e e a i e e ti t n c n i o . r me e s t su e st e g n r l d r src i o d t n Pa a t r z o i

f() () () z=+

1 k“ )叩 ) ) (+(=后 E . ) s u q()te e q( ip tnE .,h nwe t 2 i 1 g

() 2

a e e t t d b e s q a e t o a e n v r to a r si ma e y l a t s u r s me h d b s d o a i i n l a me h d S mu a i

n r s l s o h t h sme h d c n r d c t o . i lto e u t h wst a i t t o a e u e teay h s mp o i i s a d h sb t rr c g i o r cso a t tcb a, n a e t e o n t n p e ii n t n e i h t e t d t n lla ts u r se t h a i o a s q a e si t n Th s me h d h sv r r i e ma i . i o to a ey i p ra tp a t a p l a i n v l e i y t m d n i c t n m o t n r c i l p i t au n s se i e t a i c a c o i f o a d p r m ee si ai n Fo x mp e wh n t e p a t r n a a tre t m t . re a l, o e h a me e s r o h p m o e r si a e。 h sm e h d i b t r t a n fs i d l a e e t m td t i t o s e t h n a y e o h r eh d b c u e t ep r me e sg t v t i m eh d h v t e t o e a s a a t r o s m h b h to a e Suppos e

z )∑ ik+ ik 后一 (= (—)∑b( ) ( z x—+ ) i =l i =1

∑b(￣k U

n s mp o i i s An h s s e n p o e y e a p e . oa y t t b a . d t i e r v d b x m l s c ha b

2 Tr dii na e s qua e s i a i n a to ll a ts r se tm to I r e o i u nae t e p n i l fta i o a e s q a e n o d rt l mi t h r cp e o d t n l a ts u r l i r i l e tma i n S se eo i o sd r d si t, y tm b l w sc n i e e . o S o

v )后-七一 (一) (= ( Z口(一)∑七 ) 】 I 1=

t/

m

(= (—)∑b(—+(尼∑ z+ ik f V ) ) k x ) y )∑ ( (= y ∑6( k j一)“ S p oe u p s Re ev d da e 2 0—— 6 c i e t: 0 9 01 0 . F u d t

n i m:S p o d b eNa y E u p n e at e t o n ai n o n a o t i e u p ge y t v q i me t p r n u d t h D m F o l =i

( 4 )

0= a, a,,, (l 2,…,岛 b 2

u d r rn . 0 9 1 9. n e G a t 2 0 (8 ) No Co r s n i g a t o r e po d n u h rEma l s i n n i: h we mi g _

20@ 16 o 05 2 r cn

[, 1 (+ ),n J] z2 ) 2… z+ ) (+, z, ( v

◎ H ab nEn i e rn n v ri ndS rn e - ra ri ed l eg 2 0 r i g n e gU i est a p i g r i y￣el gBe l H i eb r 01 n

第1页