4 41 01, ARMAP, Q) CARMAP, Q) DARMAp, q) CARMAP, Q) CARMAP, Q) DARMAp, q) DARMAp, q), CARMAP, Q) CARMAP, Q) Chan and Tong 1987), He and Wang 1989), Br



Similar documents
0.,,., m Euclid m m. 2.., M., M R 2 ψ. ψ,, R 2 M.,, (x 1 (),, x m ()) R m. 2 M, R f. M (x 1,, x m ), f (x 1,, x m ) f(x 1,, x m ). f ( ). x i : M R.,,

24.15章.微分方程式

日本内科学会雑誌第98巻第3号

基礎数学I


untitled

2 N(ε 1 ) N(ε 2 ) ε 1 ε 2 α ε ε 2 1 n N(ɛ) N ɛ ɛ- (1.1.3) n > N(ɛ) a n α < ɛ n N(ɛ) a n

第85 回日本感染症学会総会学術集会後抄録(I)


Stata 11 Stata ts (ARMA) ARCH/GARCH whitepaper mwp 3 mwp-083 arch ARCH 11 mwp-051 arch postestimation 27 mwp-056 arima ARMA 35 mwp-003 arima postestim

本文/020:デジタルデータ P78‐97

A A. ω ν = ω/π E = hω. E

受賞講演要旨2012cs3

チュートリアル:ノンパラメトリックベイズ

日本内科学会雑誌第101巻第12号

dプログラム_1

…K…E…X„^…x…C…W…A…fi…l…b…g…‘†[…N‡Ì“‚¢−w‘K‡Ì‹ê™v’«‡É‡Â‡¢‡Ä

空力騒音シミュレータの開発

h = h/2π 3 V (x) E ψ = sin kxk = 2π/λ λ = h/p p = h/λ = kh/2π = k h 5 2 ψ = e ax2 ガウス 型 関 数 関 数 値


tnbp59-17_Web:プO1/ky079888509610003201

放射線専門医認定試験(2009・20回)/HOHS‐01(基礎一次)

time2print4.dvi

第85 回日本感染症学会総会学術集会後抄録(III)

基礎から学ぶトラヒック理論 サンプルページ この本の定価 判型などは, 以下の URL からご覧いただけます. このサンプルページの内容は, 初版 1 刷発行時のものです.

第86回日本感染症学会総会学術集会後抄録(II)

Ł\”ƒ-2005

ブック

NumRu::GPhys::EP Flux 2 2 NumRu::GPhys::EP Flux EP



z.prn(Gray)

橡表紙参照.PDF

2007-Kanai-paper.dvi


Copyrght 7 Mzuho-DL Fnancal Technology Co., Ltd. All rghts reserved.

Part. 4. () 4.. () Part ,

,..,,.,,.,.,..,,.,,..,,,. 2

I [ ] N(µ, σ 2 ) σ 2 (X 1,..., X n ) X := 1 n (X X n ): µ X N(µ, σ 2 /n) Z = X µ σ/ n N(, 1) < α < 1/2 Φ(z) =.5 α z α

untitled


( ) Loewner SLE 13 February

x (0, 6, N x 2 (4 + 2(4 + 3 < 6 2 3, a 2 + a 2+ > 0. x (0, 6 si x > 0. (2 cos [0, 6] (0, 6 (cos si < 0. ( (2 (3 cos 0, cos 3 < 0. cos 0 cos cos

z z x = y = /x lim y = + x + lim y = x (x a ) a (x a+) lim z z f(z) = A, lim z z g(z) = B () lim z z {f(z) ± g(z)} = A ± B (2) lim {f(z) g(z)} = AB z

A comparative study of the team strengths calculated by mathematical and statistical methods and points and winning rate of the Tokyo Big6 Baseball Le

5989_4840JAJP.qxd

Microsoft Word - Wordで楽に数式を作る.docx

46 Y Y Y Y 3.1 R Y Figures mm Nylon Glass Y (X > X ) X Y X Figure 5-1 X min Y Y d Figure 5-3 X =X min Y X =10 Y Y Y Y Figure 5-

特-3.indd

untitled

第52回日本生殖医学会総会・学術講演会

2 H23 BioS (i) data d1; input group patno t sex censor; cards;

1 Fourier Fourier Fourier Fourier Fourier Fourier Fourier Fourier Fourier analog digital Fourier Fourier Fourier Fourier Fourier Fourier Green Fourier

b3e2003.dvi

Isogai, T., Building a dynamic correlation network for fat-tailed financial asset returns, Applied Network Science (7):-24, 206,

日本糖尿病学会誌第58巻第2号

330

80 X 1, X 2,, X n ( λ ) λ P(X = x) = f (x; λ) = λx e λ, x = 0, 1, 2, x! l(λ) = n f (x i ; λ) = i=1 i=1 n λ x i e λ i=1 x i! = λ n i=1 x i e nλ n i=1 x

(CFW ) CFW 1

i 18 2H 2 + O 2 2H 2 + ( ) 3K

一般演題(ポスター)

example2_time.eps

E B m e ( ) γma = F = e E + v B a m = 0.5MeV γ = E e m =957 E e GeV v β = v SPring-8 γ β γ E e [GeV] [ ] NewSUBARU SPring

I.2 z x, y i z = x + iy. x, y z (real part), (imaginary part), x = Re(z), y = Im(z). () i. (2) 2 z = x + iy, z 2 = x 2 + iy 2,, z ± z 2 = (x ± x 2 ) +

(w) F (3) (4) (5)??? p8 p1w Aさんの 背 中 が 壁 を 押 す 力 垂 直 抗 力 重 力 静 止 摩 擦 力 p8 p

地域総合研究第40巻第1号

L A TEX ver L A TEX LATEX 1.1 L A TEX L A TEX tex 1.1 1) latex mkdir latex 2) latex sample1 sample2 mkdir latex/sample1 mkdir latex/sampl

1 180m g 10m/s v 0 (t=0) z max t max t z = z max 1 2 g(t t max) 2 (6) r = (x, y, z) e x, e y, e z r = xe x + ye y + ze z. (7) v =



03.Œk’ì

untitled

第5章 偏微分方程式の境界値問題

確率論と統計学の資料

ウェーブレット分数を用いた金融時系列の長期記憶性の分析

( ) a, b c a 2 + b 2 = c : 2 2 = p q, p, q 2q 2 = p 2. p 2 p q 2 p, q (QED)

日本内科学会雑誌第102巻第10号

人文学部研究年報12号.indb

「数列の和としての積分 入門」

L. S. Abstract. Date: last revised on 9 Feb translated to Japanese by Kazumoto Iguchi. Original papers: Received May 13, L. Onsager and S.

4

p *2 DSGEDynamic Stochastic General Equilibrium New Keynesian *2 2

149 (Newell [5]) Newell [5], [1], [1], [11] Li,Ryu, and Song [2], [11] Li,Ryu, and Song [2], [1] 1) 2) ( ) ( ) 3) T : 2 a : 3 a 1 :

1 4 1 ( ) ( ) ( ) ( ) () 1 4 2


204 / CHEMISTRY & CHEMICAL INDUSTRY Vol.69-1 January

Chap11.dvi

yakuri06023‡Ì…R…s†[

( ) sin 1 x, cos 1 x, tan 1 x sin x, cos x, tan x, arcsin x, arccos x, arctan x. π 2 sin 1 x π 2, 0 cos 1 x π, π 2 < tan 1 x < π 2 1 (1) (

untitled

A B C D E F G H J K L M 1A : 45 1A : 00 1A : 15 1A : 30 1A : 45 1A : 00 1B1030 1B1045 1C1030

研究シリーズ第40号

(1.2) T D = 0 T = D = 30 kn 1.2 (1.4) 2F W = 0 F = W/2 = 300 kn/2 = 150 kn 1.3 (1.9) R = W 1 + W 2 = = 1100 N. (1.9) W 2 b W 1 a = 0

Note5.dvi

A Feasibility Study of Direct-Mapping-Type Parallel Processing Method to Solve Linear Equations in Load Flow Calculations Hiroaki Inayoshi, Non-member

1 filename=mathformula tex 1 ax 2 + bx + c = 0, x = b ± b 2 4ac, (1.1) 2a x 1 + x 2 = b a, x 1x 2 = c a, (1.2) ax 2 + 2b x + c = 0, x = b ± b 2


i I

A B 5 C mm, 89 mm 7/89 = 3.4. π 3 6 π 6 6 = 6 π > 6, π > 3 : π > 3

26 Development of Learning Support System for Fixation of Basketball Shoot Form

Transcription:

41,, 01 3 41 444 ARMA Can We Have Correspondence between Discrete-Time ARM A Process and Continuous-Time ARM A Process? Mituaki Huzii ARMA Huzii 006) Huzii 006) ARMA3, ) We assume that a time series is a continuous-time weakly stationary process. We observe the process at equal sampling intervals and construct a discrete-time weakly stationary process. Does the discrete-time weakly stationary process satisfy a DARMAdiscrete-time autoregressive moving average) model, if the continuous-time weakly stationary process satisfies a CARMAcontinuous-time autoregressive moving average) model? And, conversely, we assume we have a discrete-time weakly stationary process which satisfies a DARMA model. Then can we have a continuous-time weakly stationary process, which satisfies a CARMA model and whose sampled process satisfies the DARMA model? In this paper, we review the papers, which have been published by now and treat of these two problems mainly the latter problem), and show more rigorous formulations of the problems and more general answers to these two problems, by using the resuls shown in Huzii 006) and adding new results. : DARMA, CARMA, 1. ARMAp, q) DARMAp, q) 11-8551 1-13-7 E-mail: huzii@indsys.chuou.ac.jp).

4 41 01, ARMAP, Q) CARMAP, Q) DARMAp, q) CARMAP, Q) CARMAP, Q) DARMAp, q) DARMAp, q), CARMAP, Q) CARMAP, Q) Chan and Tong 1987), He and Wang 1989), Brockwell 1995), Brockwell and Brockwell 1999) n {X n } E X n <, EX n = m m = 0, EX n+j Xn = γ j j X n = 1/ expπnµi)dzµ), γ j = 1/ 1/ 1/ expπjµi)f d µ)dµ 1.1) Brockwell and Davis 1991)) EdZµ)d Zµ ) = 0µ µ ), E dzµ) = f d µ)dµ f d µ) f d µ) f d µ) {X n } µ DARMAp, q) p φ k X n k = k=0 q θ l ɛ n l 1.) l=0 {X n } {ɛ n } 0 1 n {φ k } {θ l } t φ p 0, θ q 0 1.1) 1/ 1/ expπnµi) p φ k exp πkµi) ) dzµ) = k=0 1/ expπnµi) q θ l exp πlµi) ) dz ɛ µ) 1/ l=0 Z ɛ µ) {ɛ n } 1.1) Zµ) E dz ɛ µ) = 1dµ = fd ɛµ)dµ {X n}

ARMA 43 f d µ) q f d µ) = l=0 θ l exp πlµi) p k=0 φ k exp πkµi) 1.3) Φz) = p k=0 φ kz k = 0 Θz) = q l=0 θ lz l = 0 DAR DMA Φz) = 0 {1/β τ ; 1 τ p} z > 1 X n ɛ n {ɛ n j ; j 0} {X n j ; j 0} {X n } γ j {β τ } p j γ j = g τ β τ τ=1 1.4) {g τ } t {X t } E X t <, EX t = m m = 0, EX t+h Xt = R h h X t = expπtλi)dzλ), R h = expπhλi)f c λ)dλ 1.5) Brockwell and Davis 1991)) EdZλ)d Zλ ) = 0λ λ ), E dzλ) = f c λ)dλ {X t } CARMAP, Q) k=0 a k d k X t dt k = Q l=0 d l ) dbt b l dt l dt 1.6) Brockwell 1995)) a 0 = 1 {a k }, {b l } {B t } EB t B t1 )B t4 B t3 ) = 0t 1 < t t 3 < t 4 ), EB t B u ) = t u a P 0 b Q 0 1.6) Brockwell 1995) t 0 X t = b U t, du t = AU t dt + edb t.

44 41 01 b = b 0, b 1,, b P 1 ), b Q+ = b Q+3 = = b P 1 = 0) e = 1 0, 0,, 0, ), P ) a P ) U = U t, U 1) P 1) t,, U t, 0 1 0 0 0 0 1 0 A =.... P P ) 0 0 0 1 1 a P a 1 a P a a P a P 1 a P {U t } {B t } U 1) t dt = du t, U ) t dt = ddu t ), Aω) = a k ω k, Bω) = k=0 Q b l ω l l=0 Aω) = 0 Bω) = 0 CAR CMA X t B t {B t u ; u 0} {X t u ; u 0} R h f c λ) R h = τ=1 ) G τ exp α τ h. 1.7) {α τ } CAR {G τ } {X t } α j j α j = ᾱ j, G j = Ḡj Q f c λ) = l=0 b lπλi) l P k=0 a kπλi) k 1.8) Priestley 1981, Vol. 1), Yaglom 1987) DARMAp, q) CARMAP, Q) CARMAP, Q) p, q

ARMA 45 DARMAp, q) Priestley 1981, Vol. 1) DARMAp, q) P, Q CARMAP, Q) DARMAp, q) CARMAP, Q) Huzii 006) Chan and Tong 1987) DARMA1, 0) 1.) p = 1, q = 0 f d µ) = f c µ + k) 1.9) k= f c λ) 0 < φ 1 < 1 Priestley 1981, Vol. 1) CARMA1, 0) DARMA1, 0) 1 < φ 1 < 0 1.9) f c λ) CARMA1, 0) CARMA1, 0) CARMA, 1) DARMAp, q) p > 1 He and Wang 1989) DARMAp, q) CARMAP, Q), P = p + p p p k=0 φ kz k = 0 Brockwell 1995) Brockwell and Brockwell 1999) He and Wang 1989) Brockwell and Brockwell 1999) f d µ) q l=0 θ l exp πlµi) µ 0 CARMAP, Q) Brockwell 1995) DARMA, 0) DAR DARMA, 0) CARMA, 1) CARMA4, ) He and Wang 1989) DARMA, 0) CARMA, 1) CARMA4, ) DARMA CARMA

46 41 01 Phillips 1959) Robinson 1978) CARMAP, Q) DARMAp, q) Jones 1981) CARMAP, Q). 1 n n = 0, ±1, ±, ) = 1 n 1 {X n } X n 1 1.1 ) γ j DARMA, CARMA,.1 {X n } f d µ) {φ k ; 0 k p} {θ l ; 0 l q} φ p 0, θ q 0 1.3) {X n } DARMAp, q) p > q f d µ) 1/ 1/ log f d µ)dµ > X n {X j ; j n 1} Φz) = 0 ξ Θz) = 0 ξ ξ > 1, ξ > 1 Priestley 1981, Vol.

ARMA 47. {X t } f c λ) {a k ; 0 k P } {b l ; 0 l Q} a P 0, b Q 0 1.8) {X t } CARMAP, Q) P > Q f c λ) log f c λ) 1 + λ dλ > ) X t {X u ; u < t} Aω) = 0, Bω) = 0 {α k }, {α k } α k, α k Priestley 1981, Vol. CARMAP, Q) 1.6 X t P B t DARMAp, q) CARMAP, Q) CARMAP, Q) DARMAp, q) {Y n } {X t }.3 K K n 1, n,, n K Y n1, Y n,, Y nk ) X n1, X n,, X nk ) {Y n } {X t } embeddable) 1 {X t } P QQ < P ) CARMAP, Q) {X n } p qq < p) DARMAp, q) {Y n } p qq < p) DARMAp, q) P QQ < P ) CARMAP, Q) {X t }

48 41 01 3. GDARMAp, q) GCARMAP, Q) 3.1 Priestley 1981, Vol. 1 CARMAP, Q) DARMAp, q) 3.1 CARMAP, Q) {X t } {X n } DARMAP, P 1) 3.1 Aω) = 0 {α k } 1.7) {X n } f d µ) γ j j j 0) 1 z expα 1 ))1 z expα )) 1 z expα P )) = φ k z k {φ 0 = 1), φ k ; 1 k P } γ j = P l=1 G l expα l j) φ k γ j k = 0 j P ) k=0 j P 1/ 1/ k=0 P expπjµi) φ k exp πkµi)) fd µ)dµ = 0 k=0 j P 1/ 1/ expπjµi) φ k exp πkµi) fd µ)dµ = 0 {θ l ; 0 l P 1} k=0 φ k exp πkµi) P 1 fd µ) = θ k exp πkµi), a.e., k=0 {X n } DARMAP, P 1) k=0

ARMA 49 3. 3.1.3 {X t } {X n } K K {t 1, t,, t K } X t1, X t,, X tk ) K CARMAP, Q) DARMAp, q) GCARMAP, Q) GDARMAp, q) {Y n } GDARMAp, q) P Q {Y n } GCARMAP, Q) {X t } 0 h R h λ f c λ) 3. {Y n } GDARMAp, q) {X t } GCARMAP, Q) {Y n } {X t } a) b) a) γ j = R j j b) µ, 1/ < µ 1/, f d µ) = f c µ + k). 3.1) k= Priestley 1981, Vol. 1 {Y n } GDARMAp, q) GCARMAP, Q) Huzii 006) f c λ) f c λ) f c λ) = = Bπλi) Aπλi) 1 C l πλi α l. 3.) l=1 {α l } CAR {C l } {α l ; 1 l P } f c λ) {C l } {α l }

430 41 01 GDARMAp, q) GCARMAP, Q) GDARMAp, q) {φ k } Φ1/z) = p k=0 φ k1/z) k = 0 {β k }, β k = 1/ξ k, GCARMAP, Q) {a k } Aω) = P k=0 a kω k = 0 {α k } [β l, β l ] = expαl ), expα l ) ) = β l βl ) m 1 =, 1 β l βl m=0 0 1 expα l u + ᾱ l u)du = α l + ᾱ l [β l, β l ] expα l ), expα l )) {X t } 0 f d µ) 3.1) f c λ) f d µ) {β l ; 1 l p} f d µ) = p Ψ l 1 β l exp πµi) l=1 = p exp πjµi) Ψ l β j l j=0 + 1 j= l=1 exp πjµi) p Ψ l [β l, β l ] l =1 p l =1 Ψ l β j l p Ψ l [β l, β l ]. 3.3) 3.) f c λ) 3.1) l=1 f d µ) = = = = + k= k= f c µ + k) 1 C l πµ + k)i α l l=1 C l exp ) ) πµ + k)i α l u du k= j=0 l=1 0 exp πjµi) C l expα l j) 1 j= l=1 l =1 l =1 C l expαl ), expα l ) ) exp πjµi) C l expᾱ l j ) C l expαl ), expα l ) ) 3.4) l=1

ARMA 431 Huzii 006) 3..1 Ψ l 0 1 l p) 3.. {β l } β l 0 3..3 Φz) = 0 Θz) = 0 3.3) 3.4) 3.3 Huzii 006)) 3.4) a), b), c) a) P p b) P = p expα l ) = β l 1 l p) c) P > p, 1 l p β l = expα l ) p + 1 l P l 1 l p l I l α l = α l + πi l i 3.4 Huzii 006)) {Y n } {X t } GDARMAp, q) GCARMA P, Q) 3..1 3..3 {C l, 1 l P } {Y n } {X t } a) P = p 1 l p l p Ψ Ψ p l l =1 l [β l, β l ] = C C l l =1 l expαl ), expα l ) ) b) P > p 1 l p l p Ψ Ψ l l =1 l [β l, β l ] = j Ll) C P C j l =1 l expαj ), expα l ) ) Ll) = {j : expα j ) = β l } DARMAp, q) β l1 l 1 β l1 = β l1, Ψ l1 = Ψ l1 CARMAP, Q) α l l α l = ᾱ l, C l = C l 3.4 a), b) [β l, β l ] expα l ), expα l ) ) 3.3 3.3 3.4 GDARMAp, q) GCARMAP, Q) p = 1 p = 3.3 3.4 p 3

43 41 01 p = 3 3.3 3.4 3.3.1 GDARMA1, 0) GCARMAP, Q) GDARMA1, 0) f d µ) = θ 0 1 + φ 1 exp πiµ) = θ 0 1 + φ 1 + φ 1 cosπµ), {Y n } Y n + φ 1 Y n 1 = θ 0 ɛ n {Y n } Φz) = 1 + φ 1 z Φz) = 0 1/β 1 β 1 = φ 1 Ψ 1 = θ 0 θ 0 > 0 GCARMA1, 0) β 1 > 0 φ 1 < 0 3.3 b) 3.4 a) α 1 = logβ 1 ), Ψ 1 [β 1, β 1 ] = C 1 expα 1 ), expα 1 )) [β 1, β 1 ] C 1 = ± expα 1 ), expα 1 )) Ψ α 1 ) 1 = ± 1 β1 )θ 0 GCARMA1, 0) β 1 < 0 β 1 = expα 1 ) α 1 3.3 c) α 1 = log β 1 + πi, α = log β 1 πi 3.4 b) C 1 = C = r 1 expπνi) r 1 r 1 = Ψ 1 [β 1, β 1 ] expα1 ), expα 1 ))G, 1 1 α1 = log β 1 ) ) G 1 1 = 1 + cosπθ 1 1) cos πθ 1 1 + ν) expπθ 1 1i) = α 1 + πi α 1 ) + π) ν ν r 1 GCARMA, 1) Chan and Tong 1987) β 1 < 0 GCARMA, 1) 3.3. GDARMA, 1) GCARMAP, Q) GCARMA, 1) P

1) β 1, β 0 ARMA 433 3.4 a) β = β 1, Ψ = Ψ 1, α = ᾱ 1, β 1 = expα 1 ), C = C 1 l = 1, l = GDARMA, 1) a) F 1, F Ψ 1 = Ψ 1 expπη 1 i), α 1 = α1 + α1 i, C 1 = r 1 expπν 1 i), expπδ 1 1i) = 1 expα 1) cosα1 ) + expα1) sinα1 )i 1 expα 1 ) cosα1 ) 1 ) expπθ 1 1i) = α1 + α1 i α 1 ) + α1 ) F 1 = Ψ 1 [β 1, β 1 ]F1 + F1 i), F = F 1 α 1, α 1, F 1, F 1 r 1 0, 1/ < η 1, ν 1 1/ 3.4 a) α1, α1, η 1 U num U den cotπθ 1 1), 3.5) ) ) U num = 1 expα1) sin πη 1 + δ 1 1), U den = 1 expα1 ) cosα 1 ) + exp4α 1 ) ) ) + 1 expα1) cos πη 1 + δ 1 1), 3.4 a) C 1, C C 1 = r 1 expπν 1 i) C = C 1 r 1 ν 1 Ψ 1 [β 1, β 1 ]F1 = r1 expα 1 ), expα1) ) 1 + cos πν 1 + θ ) ) 1 1) cosπθ 1 1), Ψ 1 [β 1, β 1 ]F1 = r1 expα 1 ), expα1) ) sin πν 1 + θ ) ) 1 1) cosπθ 1 1) Ψ 1 = expπ 0.15i) Ψ 1 = 1, η 1 = 0.15), β 1 = exp + 0.5i), α 1 =, α 1 = 0.5) 3.5) 1.006791 4 3.5) GDARMA, 1) Y n 0.4Y n 1 + 0.0Y n = 1.41ɛ n 0.6ɛ n 1 f c λ) f c λ) = r1 cosπν 1 ) ) πλi) + 4 cosπν 1 ) sinπν 1 ) πλi) + 4 πλi) + 4 + 0.5

434 41 01 r 1, ν 1 ) Sol 1 =.04, 0.11) Sol = 11.68, 0.3). f c λ) Sol 1 f c λ) = 3.1 πλi) + 5.16 πλi) + 4 πλi) + 4.5 CMA Sol f c λ) = 3.1 πλi) 5.16 πλi) + 4 πλi) + 4.5 CMA Brockwell 1995) He and Wang 1989) {Y n } GDARMA, 0) β 1, β 0 3.5) Brockwell 1995) 3.6) ) β 1, β β 1, β 0 β 1 = β β 1, β β 1, β 3.3 b) α 1 = log β 1, α = log β 3.4 a) GDARMA, 1) 3.4 a) F 1 = Ψ 1 F = Ψ Ψ l [β 1, β l ], 3.6) l =1 Ψ l [β, β l ] 3.7) l =1 C = expα 1), expα 1 )) expα 1 ), expα )) C 1 + C 1 = expα ), expα )) expα ), expα 1 )) C + F 1 1, expα 1 ), expα )) C 1 3.8) F 1 expα ), expα 1 )) C 3.9) C 1, C F 1 + F 0 F 1 < 0, F < 0 F 1 > 0, F > 0 C 1, C ) C 1, C 0 ± C 1, C )

ARMA 435 3.4 a) C 1, C GCARMA, 1) F 1, F F 1, F, β 1, β C 1, C ) GCARMA, 1) Huzii 006)) 3.4 b) GCARMAP, Q) P GCARMA3, ) Ψ 1 =, Ψ = 1, β 1 = exp ) = 0.135335 α 1 = ), β = exp.1) = 0.1456 α =.1) F 1 =.04096, F = 1.018480 C 1, C ) 4 GCARMA, 1) GCARMA, 1) Ψ 1, Ψ, β 1 β β = exp 5) = 0.006738 C 1, C ) GCARMA, 1) GCARMA3, ) GCARMA4, 3) Huzii 006)) 3) β 1, β 3.3 3.4 Huzii 006) 3.3.3 GDARMA3, ) GCARMAP, Q) GDARMAp, q) p 3 GCARMAP, Q) 3.3 3.4 GCARMAP, Q) β 1, β, β 3 β 1 > β > β 3 C 1, C, C 3 3.4 a) GCARMA3, ) GDARMA3, ) 3.4 a) 3 F l = Ψ l Ψ l [β l, β l ], 1 l 3 l =1 expα k ), expα l ) ) = α k,l, 1 k, l 3, α k, 1 k 3, α k,l = α l,k

436 41 01 3.4 a) C 3 = C 3 = C 3 = 1 F 1 α 1,1 C 1 α 1, C, 3.10) α 1,3 C 1 α 1,3 α 1,3 1 F α,1 C 1 α, C, 3.11) α,3 C α,3 α,3 1 C 3,1 ± C 3, ), 3.1) α 3,3 C 3,1 = ) α 3,1 C 1 + α 3, C C 3, = α 3,1 C 1 + α 3, C ) + 4α3,3 F 3 C 1, C, C 3 3.10) 3.11) C = 1 B 11 C 1 + B ) 1 + B 11 C 1 + B 1 ) A 11 C 1 C + D 11 1 3.13) C = 1 B 11 C 1 + B ) 1 B 11 C 1 + B 1 ) A 11 C 1 C + D 11 1 3.14) α, A 11 = α ) 1,, B 11 = α,1 α 1,1, α,3 α 1,3 α,3 α 1,3 B 1 = F 1 α 1,3, D 11 = F 1 α 1,3 B 11 + F α,3 A 11, 3.10) 3.1) C = 1 B 1 C 1 + B ) + D 1 C 1 + D ) A 1 C 1 C + D 3 1 3.15) C = 1 B 1 C 1 + B ) D 1 C 1 + D ) A 1 C 1 C + D 3 1 3.16)

ARMA 437 A 1 = α 1, α1,3 α 1,α,3, α 1,3 α 3,3 B 1 = α 1,1α 1, α 1,3 B = F 1 α 1,3 α 1,1α,3 α 1,, α 1,3 α 3,3 α 3,3 α,3 α ) 1,, α 3,3 α 1,3 D 1 = α 1,1α,3 α 1,, D = F 1α,3, α 1,3 α 3,3 α 3,3 α 1,3 α 3,3 D 3 = F 1 α 1, α,3 α 1,3 4α3,3 α ) 1,1α,3 4α 1,3 α3,3 ) + F 3 α 1, α 1,α,3. α 3,3 α 1,3 α 3,3 α 1,3 3.13) 3.14) 3.15) 3.16) C 1, C ) C 1, C 0 C 1, C ± 3.1 A 11 < 0, B 11 < 0, A 1 > 0, B 1 > 0, D 1 > 0 B 1 B D > 0, < 0, > 0, F 1 F 1 F 1 B 1 + D 1 > 0, B 1 D 1 > 0 B + D F 1 < 0, B D F 1 < 0. 3.1 C 1 C 1 3.13) C < 3.15) C, 3.13) C < 3.16) C, 3.14) C > 3.15) C, 3.14) C > 3.16) C β 1 = 0.8, β = 0.6, β 3 = 0.4, Ψ 1 = Ψ = Ψ 3 = 1.0 f d µ) = 1 1 0.8 exp πµi) + 1 1 0.6 exp πµi) + 1 1 0.4 exp πµi)

438 41 01 1 β 1 = 0.8, β = 0.6, β 3 = 0.4, Φ 1 = Φ = Φ 3 = 1.0 Y n 1.8Y n 1 + 1.04Y n 0.19Y n 3 = 3ɛ n 3.6ɛ n 1 + 1.04ɛ n {Y n } 39.5 C 1 0.5 3.14) 3.15) 1 3.14) 5.0 C 1.5 C 1, C ) C 1, C, C 3 ) 1.1, 1.6, 1.48) 10.35, 34.86, 8.38) 1.1, 1.6, 1.48) GCARMA3, ) f c λ) = = 1.1 πλi log 0.8 + 1.6 πλi log 0.6 + 1.48 πλi log 0.4 3.86 πλi) 4.1 πλi) 0.95 πλi) 3 + 1.65 πλi) + 0.79 πλi) + 0.10 CMA 0.34 0.73 GCARMA3, ) CMA 10.35, 34.86, 8.38) f c λ) = 10.35 πλi log 0.8 + 34.86 πλi log 0.6 + 8.38 πλi log 0.4 = 3.87 πλi) + 4.1 πλi) 0.95 πλi) 3 + 1.65 πλi) + 0.79 πλi) + 0.10 CMA 0.34 0.73 3.13)

ARMA 439 3.15) C 1 C 1 C C 3.14) 3.16) C 1, C ) C 1, C ) β 1 = exp.0), β = exp 5.0), β 3 = exp 10.0), Ψ 1 =.0, Ψ = 1.0, Ψ 3 = 0.01 3.13), 3.14), 3.15), 3.16) 8.0 < C 1 4.0 3.0 < C 1 7.0 3 3.14) 3.15) 3.16) 3.13) 4 40.0 < C 1 40.0 C 1 C 1 4 GDARMA3, ) GCARMA3, ) GCARMA3, ) 3.4 b) GCARMAP, Q), P 4 3.4 CMA ζ 1, ζ,, ζ Q Bω) = b Q ω ζ 1 )ω ζ ) ω ζ Q ) ζ 1 ζ 1 = ζ 1 + ζ 1 i ζ 1, ζ 1 ζ 1 > 0 Bω) ζ 1 ζ 1 B rev ω) = b Q ω + ζ 1 )ω ζ ) ω ζ Q ) πλi + ζ 1 = ζ 1 ) + πλ ζ 1 ) = πλi ζ 1. f c λ) = Bπλi) Aπλi) = B rev πλi) Aπλi) 1 f rev λ) = C rev.l πλi α l l=1 = f rev λ) ). C rev.l 4. GDARMAp, q) GCARMAP, Q)

440 41 01 β 1 = exp.0), β = exp 5.0), β 3 = exp 10.0), Φ 1 =.0, Φ = 1.0, Φ 3 = 0.01, C 1 < 0. 3 β 1 = exp.0), β = exp 5.0), β 3 = exp 10.0), Φ 1 =.0, Φ = 1.0, Φ 3 = 0.01, C 1 > 0. GDARMAp, q) GCARMAP, Q) DAR CAR GDARMAp, q) 1 p 3 p = p = 3 p =

ARMA 441 Huzii 006) p = 3 GDARMAp, q) GCARMAP, Q) P, Q GCARMAP, Q) Brockwell and Brockwell 1999)) GCARMA P, Q) GDARMAp, q) Peter J. Brockwell A. A.1 1.6) CARMAP, Q) DARMAp, q) CARMAP, Q) DARMAp, q) R h h h α h R h = Oexpα h )) P = 1, Q = 0 1.6) a 1 dx t dt + a 0 X t = b 0 db t dt A.1) {B t } A.1) db t /dt A.1) δ > 0 A.1) [t, t + δ] t+δ a 1 X t+δ X t ) + a 0 X u du = b 0 B t+δ B t ) t A.)

44 41 01 {X t } 1.5) {X t } t+δ X u du t h > 0 A.) X t h a 0, a 1 X t h {B t h τ B t h τ δ ; τ 0} B t+δ B t t+δ a 1 R h+δ R h ) + a 0 R h+u t du = 0 t A.3) A.3) δ δ 0 R h h > 0 1 h = 0 a 1 dr h dh + a 0R h = 0, h > 0) A.4) h > 0 R h = G 11 expα 1 h) A.5) α 1 = a 0 /a 1 a 0 /a 1 > 0 G 11 R 0 h < 0 R h = R h R h h f c λ) = exp πhλi)r h dh = G 11 α 1 ) 1 πλi α 1 ) λ O 1/λ P =, Q = 1 a d X t dt + a 1 dx t dt + a 0 X t = b 1 d dt ) ) dbt dbt + b 0 dt dt = G 11α 1 πλ) + α 1 A.6) A.7) CAR α 1, α 0 α = ᾱ 1 B t A.7) [t, t + δ 1 ] [t, t + δ ] δ 1 > 0, δ > 0) t+δ t t+δ u +δ 1 X u+δ1 du, X u1 du 1 du, u =t u 1 =u t+δ u=t B u+δ1 B u )du P = 1, Q = 0 X t h, h > 0 δ δ 0

ARMA 443 δ 1 δ 1 0 R h h h = 0 a d R h dh + a 1 dr h dh + a 0R h = 0, h > 0) R h = G 1 expα 1 h) + Ḡ1 expᾱ 1 h) A.8) A.9) G 1 R 0 R h R 1 h < 0 R h = R h f c λ) f c λ) = = exp πhλi)r h dh G 1 πλi α 1 + f c λ) = C 1 C 1 + πλi α 1 πλi ᾱ 1 Ḡ 1 πλi ᾱ 1 + G 1 πλi α 1 + Ḡ 1 πλi ᾱ 1 = b 1 πλi) + b 0 πλi α 1 )πλi ᾱ 1 ) C 1, b 0, b 1 A.10) 1.6) {X t } {α k ; 1 k P } 1.7) 1.8) Brockwell 1995) Yaglom 1987) R h h P = 1 P = f c λ) λ P = 1 P = O 1/λ P Q)) 1.6) R h 1.5) P λ P f c λ)dλ < A.11) 1953)) P Q P Q + 3.4 {C l ; 1 l P } C l = 0, l=1 l=1 C l ) l=1 C l l 1 =1,l 1 l α l1 = 0,, α l1 α l α lp 1 [P/] = 0 A.1)

444 41 01 ) l 1, l,, l P 1 [P/] 1 P l 1 < l < < l P 1 [P/] l i l i = 1,,, P 1 [P/]) 3.3. 3.3.3 CARMAP, Q) R h A.11) 3 {C l } A.1) 1.6) 1 Brockwell 1995)) CARMA P, Q) DARMAp, q) CARMAP, Q) Brockwell, P. J. 1995). A note on the embedding of discrete-time ARMA processes, J. Time Ser. Anal., 16, 451 460. Brockwell, A. E. and Brockwell, P. J. 1999). A class of non-embeddable ARMA processes, J. Time Ser. Anal., 0, 483 486. Brockwell, P. J. and Davis, R. A. 1991). Time Series: Theory and Methods nd Ed.), Springer-Verlag, New York. Chan, K. S. and Tong, H. 1987). A note on embedding a discrete parameter ARMA model in a continuous parameter ARMA model, J. Time Ser. Anal., 8, 77 81. He, S. W. and Wang, J. G. 1989). On embedding a discrete-parameter ARMA model in a continuous-parameter ARMA model, J. Time Ser. Anal., 10, 315 33. Huzii, M. 006). Embedding a Gaussian discrete-time autoregressive moving average process in a Gaussian continuous-time autoregressive moving average process, J. Time Ser. Anal., 8, 498 50. 1953). Jones, R. H. 1981). Fitting a continuous time autoregression to discrete data, Applied Time Series Analysis II ed. D. F. Findley), Academic Press, New York, 651 68. Phillips, A. W. 1959). The estimation of parameters in systems of stochastic differential equations, Biometrika, 46, 67 76. Priestley, M. B. 1981). Spectral Analysis and Time Series, Vol. 1, Univariate Series, Vol. Multivariate Series, Prediction and Control, Academic Press. Robinson, P. M. 1978). Continuous model fitting from discrete data,directions in Time Series eds. D. R. Brillinger and G. C. Tiao), Institute of Mathematical Statistics, Hayward, California, 63 78. Yaglom, A. M. 1987). Correlation Theory of Stationary and Related Random Functions, Springer-Verlag, New York.