LIKELIHOOD RATIO TESTS UNDER LOCAL ALTERNATIVES IN REGULAR SEMIPARAMETRIC MODELS

Size: px
Start display at page:

Download "LIKELIHOOD RATIO TESTS UNDER LOCAL ALTERNATIVES IN REGULAR SEMIPARAMETRIC MODELS"

Transcription

1 Statistica Siica 15(2005), LIKELIHOOD RATIO TESTS UNDER LOCAL ALTERNATIVES IN REGULAR SEMIPARAMETRIC MODELS Mouliath Baerjee Uiversity of Michiga Abstract: We cosider the behavior of likelihood ratio statistics for testig a fiite dimesioal parameter, or fuctioal of iterest, uder local alterative hypotheses i regular semiparametric problems. These are problems where regular estimates of the parameter/fuctioal of iterest exist ad, i particular, the MLE coverges at rate to the true value ad is asymptotically ormal ad efficiet. We show that i regular problems, the likelihood ratio statistic for testig H 0 : θ(ψ) = θ(ψ 0) = θ 0 ( where ψ 0 is a fixed poit i the ifiite dimesioal parameter space Ψ ad θ(ψ) is a fiite-dimesioal (sub)parameter or fuctioal of iterest) coverges i distributio uder local (cotiguous) alteratives of the form ψ = ψ 0 + 1/2 h + o( 1/2 ) to a o-cetral χ 2 radom variable, with o-cetrality parameter ivolvig the directio of perturbatio h ad the efficiet iformatio matrix for θ uder parameter value ψ 0. This coforms to what happes i the case of regular parametric models i classical statistics. Key words ad phrases: Asymptotic distributio, χ 2 distributio, cofidece sets, cotiguity, Cox model, least favorable submodels, likelihood ratio, local alteratives. 1. Itroductio Let X 1,..., X be a radom sample from the distributio P ψ, ψ belogig to a set Ψ. We assume a topology o the set Ψ; i applicatios Ψ is usually a subset of a Baach or Hilbert space. Let lik(ψ, x) deote the likelihood for a observatio. I cotrast to classical parametric statistics, lik(ψ, x) i may cases is a modificatio of p(ψ, x) where p(ψ, ) = dp ψ /dµ, µ beig some commo domiatig σ - fiite measure. The likelihood ratio statistic (heceforth LRS) for testig the ull hypothesis H 0 : θ(ψ) = θ 0 for some fuctioal θ is give by: ( lrt (θ 0 ) = 2 {sup ) ( )} ψ Ψ log lik(ψ, X i ) sup ψ:θ(ψ)=θ0 log lik(ψ, X i ) i=1 i=1 = 2P (log(lik( ˆψ))) 2P (log(lik( ˆψ 0 ))) ;

2 636 MOULINATH BANERJEE here ˆψ is the ucostraied maximizer of the log likelihood ad ˆψ 0 is the maximizer uder the costrait imposed by H 0, while P deotes the empirical measure of the observatios X 1,..., X. I the cases we cosider, θ takes values i R k. The behavior of lrt (θ 0 ) uder H 0, for a 1 dimesioal parameter θ, was established i Theorem 3.1 of Murphy ad Va der Vaart (1997) for a broad class of semiparametric models characterized by the requiremet that the MLE of θ coverges at the regular rate to a ormal limit ad admits a asymptotic liear represetatio. It was show that the LRS coverges to a χ 2 1 distributio. This approach exteds readily to the case of a vector valued parameter (here, the limit distributio is χ 2 k, k beig the dimesio of θ) ad was ivestigated by Murphy ad Va der Vaart (2000) i a profile likelihood settig. I this paper we focus o the behavior of the LRS i these regular semiparametric models uder a sequece of local alteratives that coverge to a poit i the ull hypothesis with icreasig sample size. More specifically, let ψ = ψ 0 +(1/ )h+o(1/ ), where θ(ψ 0 ) = θ 0. We study the limitig behavior of lrt (θ 0 ) based o i.i.d. data X 1,..., X, whe P ψ is take to be the true uderlyig distributio at stage. The alteratives {ψ } correspod to the curve ψ(t) ψ 0 + t h + o(t) with gradiet h at the poit 0. We show that the likelihood ratio statistic is asymptotically distributed as a o cetral χ 2 k radom variable with o cetrality parameter which ivolves I 0, the efficiet iformatio for the fiite dimesioal parameter θ uder parameter value ψ 0, ad the directio of perturbatio of ψ 0. I the case where the parameter ψ ca be partitioed as (θ, η) (with η beig ifiite dimesioal), the expressio for the o cetrality parameter matches that obtaied for regular parametric models. Before proceedig to the key results i the ext sectio, we impose some structural requiremets o the uderlyig model {P ψ : ψ Ψ} ad the fuctioal of iterest θ. Helliger Differetiability: Assume that the parameter ψ lies i a subset Ψ of some Hilbert space H. Cosider the path i Ψ give by ψ t = ψ + h t + o(t); thus h is the gradiet of the path at t = 0. Let s(ψ, ) = (dp ψ /dµ) 1/2, where µ is a σ-fiite measure domiatig the P ψ s. The Helliger differetiability coditio ca the be stated as: [ s(ψt, ) s(ψ, ) 1 ] 2 t 2 (Ah)s(ψ, ) dµ( ) 0 as t 0. (1.1) Here A is a bouded liear operator defied o Ḣ, the closed liear spa of the h s (the gradiets of the ψ t s) ad assumes values i L 0 2 (P ψ). I fact, Helliger differetiability implies that the rage of A is cotaied i L 0 2 (P ψ), the subset

3 LIKELIHOOD RATIO TESTS UNDER LOCAL ALTERNATIVES 637 of square itegrable fuctios uder the measure P ψ that have mea 0 (see, for example, Sectio 2.1 of Bickel, Klaasse, Ritov ad Weller (1998)). Differetiability of θ: The fuctioal of iterest θ, treated as a fuctio of ψ, i.e., θ : Ψ R k, is differetiable i the sese that ( / t)(θ(ψ t )) t=0 = L(h) where L is a bouded liear map from Ḣ Rk. Observe that by the characterizatio of real-valued fuctioals defied o Hilbert spaces, we have L(h) = ( θ 10, h,..., θ k0, h ) T for some θ 10,..., θ k0 all belogig to Ḣ. 2. The Asymptotic Distributio of the Likelihood Ratio Statistic: Mai Results We first formulate a extesio of Theorem 3.1 of Murphy ad Va der Vaart (1997) (dealig with the behavior of lrt (θ 0 ) uder the ull hypothesis) that applies to vector valued fuctioals θ of ay dimesio. This is exploited alog with results from cotiguity theory to derive the limit distributio of the likelihood ratio statistic uder a sequece of local alteratives of the type cosidered i the previous sectio. The use of cotiguity theory stems from the fact that the model P ψ is Helliger differetiable alog the curve P ψt which implies, through a LAN (local asymptotic ormality) expasio of the log-likelihood ratio log d Pψ /d Pψ 0, that the sequece of probability measures {Pψ } (the fold product of P ψ ) ad {Pψ 0 } (the -fold product of P ψ0 ) are mutually cotiguous. The Geeral Semiparametric Likelihood Ratio Statistic Theorem: Suppose that ψ 0 is the true value of the parameter with θ(ψ 0 ) = θ 0 (thus Ψ 0 H 0 ) ad deote P ψ0 by P 0. A.1 Assume that as the MLE ˆθ = θ( ˆψ) satisfies (ˆθ θ0 ) = I 1 0 P ( l) + o p (1) (2.1) uder P 0, where P 0 ( l) = 0 ad I 0 = P 0 ( l l T ). Thus l L 0 2 (P 0) k. A.2 Assume that, for every t i a eighborhood U of θ 0 ad every ψ i a eighborhood V of ψ 0, there exists a surface t ξ(t, ψ) takig values i Ψ that satisfies the followig: (a) θ(ξ(t, ψ)) = t; (b) ξ(t, ψ) t=θ(ψ) = ψ; (c) t l(x; t, ψ) = l(lik(ξ(t, ψ), x)) is twice cotiuously differetiable i t for every x with derivatives l ad l with respect to t; (d) l( ; θ 0, ψ 0 ) = l ad P ( l( ; θ, ψ)) P P 0 ( l l T ) = I 0 for ay radom θ P θ 0 ad ψ P ψ 0 uder P 0 ad P ( l( ; θ 0, ˆψ 0 ) l) P0 0. A.3 Suppose that both the ucostraied ad costraied maximizers of the likelihood, ˆψ ad ˆψ0, are cosistet uder P 0.

4 638 MOULINATH BANERJEE Theorem 2.1. If A.1, A.2 ad A.3 hold, the lrt (θ 0 ) = ) T ) (ˆθ θ 0 I0 (ˆθ θ 0 + o p (1) d Z T I 0 Z χ 2 k, where Z N k (0, I 1 0 ). The proof of this theorem is skipped. The fuctio l that appears i (2.1) is actually the efficiet score fuctio i regular semiparametric models. For the details, see Baerjee (2000, Chap.2). Power uder Local (Cotiguous) Alteratives. The followig theorem characterizes the limitig power of the likelihood ratio test uder a sequece of local alteratives covergig to the true value of the parameter. Theorem 2.2. Cosider the likelihood ratio statistic lrt (θ 0 ) for testig the ull hypothesis θ = θ 0 θ(ψ 0 ). Assume that the model is Helliger differetiable alog the curve ψ t = ψ 0 + t h + o(t) with derivative operator A. Also assume that coditios A.1 to A.3 of Theorem 2.1 hold uder P ψ0. The, uder the sequece of local alteratives P ψ0 +h/ +o(1/ ), the likelihood ratio statistic lrt (θ 0 ) coverges i distributio to L where L follows a χ 2 k distributio with o-cetrality parameter c T I 0 c; here, I 0 is the covariace matrix of l ad c is the covariace betwee A h ad l i L 0 2 (P ψ 0 ) scaled by I 0, or c = I0 1 Ah, l Pψ0. Commets: Note that A i (1.1) is take to be a bouded liear (score) operator defied o the closed liear spa of gradiets. However, the above theorem goes through without the requiremet of a score operator so log as we have Helliger differetiability alog the path ψ 0 + h t + o(t) with some g L 0 2 (P ψ 0 ) playig the role of A h i (1.1). The coclusio i the statemet of Theorem 2.2 remais valid with A h replaced by g. Proof of 2.2. Helliger differetiability of the model alog the curve ψ t implies that [ dp 1/2 ψ t dp 1/2 ψ 0 1 ] 1/2 2 (Ah)dPψ t as t 0. Now set ( dp ψ 0 + h +o( 1 ) ) Λ = log dpψ. 0 This is the log likelihood ratio based o observatios X 1,..., X at stage. By Lemma of Va der Vaart ad Weller (1996), we get a LAN expasio for the log likelihood ratio as Λ = log dp ψ 0 + h +o( 1 ) dpψ = 1 Ah(X i ) Ah 2 + o P ψ0 (1). i=1

5 LIKELIHOOD RATIO TESTS UNDER LOCAL ALTERNATIVES 639 It follows that Λ d W where W N( (1/2) Ah 2, Ah 2 ), uder P ψ 0. Therefore, uder {P ψ 0 }, exp (Λ ) dp ψ 0 + h +o( 1 ) dp ψ 0 d exp W. But E(exp (W )) = exp ( (1/2) σ 2 + (1/2) σ 2 ) = 1, usig the formula for the momet geeratig fuctio of the ormal distributio. By Le Cam s first lemma (Va der Vaart (1998, p.88)), we coclude that the sequeces of probability measures {P ψ 0 +h/ +o(1/ ) } ad {P ψ 0 } are cotiguous. Cosequetly the covergeces i probability that hold uder Pψ 0 also hold uder P ψ 0 +h/ +o(1/ ), ad vice-versa. Cosider ow the joit distributio of T = P I0 1 l ad Λ uder Pψ 0. From ( ) T Λ 1 = I 1 l(x 0 i ) 1 + o Ah(Xi ) 1 P 2 Ah 2 ψ0 (1), the CLT i cojuctio with Slutsky s theorem yields: ) ( T Λ d N k+1 (µ, Σ) uder P ψ 0 with µ = (0 k 1, (1/2) Ah 2 ) T, ad Σ = ( I 1 0 c T c Ah 2 ). Here c = Cov (Ah, I0 1 l) = I0 1 Cov (Ah, l) = I0 1 Ah, l Pψ0. Note that Ah, l Pψ0 = ( (Ah) l 1 dp ψ0,..., (Ah) l k dp ψ0 ) T. Now by Le Cam s third lemma (Va der Vaart (1998, p.90)), the limitig distributio of (T, Λ ) T uder the sequece P ψ 0 +h/ +o(1/ ) is N k+1( µ, Σ), where µ = (c, (1/2) Ah 2 ) T. This shows that T d N(c, I0 1 ) uder P ψ 0 +(h/ )+o(1/. We ow ote that coditios A.2 ) ad A.3 i Theorem 2.1 cotiue to hold uder P ψ 0 +h/ +o(1/, ad coditio ) A.1 is modified to (ˆθ θ 0 ) = T + o p (1) uder P ψ 0 +(h/ )+o(1/ ) with T covergig i distributio to N(c, I0 1 ). Cosequetly lrt (θ 0 ), which i this case is still (ˆθ θ 0 ) T I 0 (ˆθ θ0 ) + o p (1), coverges to χ 2 k (ct I 0 c). The covariace vector c is the derivative of the curve θ(ψ t ) at t = 0 whe θ is pathwise orm differetiable i the sese of Va der Vaart (1991). To see this, we briefly review the cocept of pathwise orm differetiability. Cosider θ as a fuctioal from the space of probabilities to R k ; hece write θ(ψ t ) θ(p ψt ).

6 640 MOULINATH BANERJEE The θ is called pathwise orm-differetiable if there exists a bouded liear map θ from R(A) R k such that [ ] θ(pψt ) θ(p ψ ) L(h) = lim = t 0 t θ (A h). A ecessary ad sufficiet coditio for pathwise orm differetiability due to Va der Vaart (see Va der Vaart (1991)) is give by R(L ) R(A ), ad failure of this coditio implies the o-existece of regular estimators of θ. Whe the coditio for differetiability is satisfied we have that L = θ A so that L = A θ. It is easily show that the above ecessary ad sufficiet coditio is equivalet to the followig: for each 1 j k there exists a (ecessarily uique) solutio to the equatio A x = θ j0 i R(A). We deote the uique vector of solutios by g 0 = (ġ 10,..., ġ k0 ). Sice it is also the case that for each j, θ j0 = L (e j ) = A ( θ (e j )) (where e j deotes the j th caoical basis vector) ad θ assumes values i R(A), we coclude that ġ j0 = θ (e j ). We call θ the efficiet ifluece fuctio ad idetify it with the vector g 0, its values at the caoical basis vectors. I may semiparametric situatios it is this vector that provides the liear approximatio to the cetered ad scaled MLE of θ, the parameter of iterest, i.e., g 0 = I0 1 l. Corollary 2.1. If θ is pathwise orm-differetiable ad the efficiet ifluece fuctio for the estimatio of θ at ψ 0, g 0 = I0 1 l, the c = ( / t)θ(ψt ) t=0. Proof. We have c = Cov (Ah, I 1 0 l) = Cov ((ġ 10,..., ġ k0 ) T, Ah) = ( ġ 10, Ah,..., ġ k0, Ah ) T = ( A ġ 10, h,..., A ġ k0, h ) T = ( θ 10, h,..., θ k0, h ) T = t θ(ψ t) t=0. For yet aother characterizatio of c i terms of least favorable directios, see Baerjee (2000, Chap.2). Also ote that Theorem 2.2 ad Corollary 2.1 remai valid for a Baach space valued parameter ψ (pathwise orm differetiability is characterized i exactly the same way as i the Hilbert space case if adjoits are give the proper iterpretatios). 3. Partitioed Parameters Here we use the results of the previous sectio to obtai expressios for the power of the likelihood ratio test uder local alteratives whe the parameter is aturally partitioed ito two compoets: the first, θ, which is also the parameter of iterest, belogig to a Euclidea space ad the secod, η, beig ifiite dimesioal. This situatio arises extesively i semiparametric models (ad is therefore deemed importat i its ow right), with perhaps the most commo example beig the Cox Proportioal Hazards settig where the hazard fuctio correspodig to the survival time of a idividual is modeled as

7 LIKELIHOOD RATIO TESTS UNDER LOCAL ALTERNATIVES 641 Λ(t Z) = e θt Z Λ(t). Here the parameter vector is ψ = (θ, Λ), with θ the Euclidea regressio parameter measurig the effect of measured covariates o the survival time, ad Λ the baselie hazard fuctio takig values i the space of all cadlag fuctios defied o a bouded iterval. Other useful models, where the parameter is aturally partitioed, iclude frailty models (for example, the Gamma frailty models cosidered i Murphy ad Va der Vaart (1997), Murphy ad Va der Vaart (2000) ad the expoetial frailty model also cosidered i these papers), case cotrol studies for missig covariates (Roeder, Carroll ad Lidsay (1996), Murphy ad Va der Vaart (2000) ad Murphy ad Va der Vaart (2001)), partially liear regressio models (see, for example Gree (1984), Egle et al. (1986) ad Va der Vaart (1998)) ad semiparametric mixture models. Let P = {P θ,η : θ Θ, η H}, where Θ is a ope subset of R k ad H is some subset of a Baach space G. Cosider a fixed set of paths i H of the form η t = η + βt + o(t) with β G. Let C deote the closed liear spa of the β s. Now cosider paths of the form (θ + th, η t ) Θ H, ad assume Helliger differetiability with respect to this set of paths. Thus [ dp 1/2 θ+th,η lim t dp 1/2 θ,η t 0 t 1 2 ( l θ T h + l η β)dp 1/2 ] 2 θ,η = 0. Now, R k C is a Baach space itself with the product topology ad i this situatio the score operator A : R k C L 0 2 (P θ,η) is give by A (h, β) = l θ T h+ l η β, where l θ L 0 2 (P θ,η) k is the score fuctio for θ ad the score operator for η, the uisace parameter, is the bouded liear map l η from C L 0 2 (P θ,η) with adjoit operator l η : L0 2 (P θ,η) C. Note that A : L 0 2 (P θ,η) (R k C), the adjoit operator of A, is give by A u (h, β) = u, A(h, β) = h T l θ, u Pθ,η + l η(u)(β). Let the fuctioal χ : Θ R m with m k be differetiable with respect to θ, ad let χ (θ) m k be its derivative. Cosider the score operator for θ, l θ = ( l θ1,..., l θk ) T. Now for each j = 1,..., k write l θj = l θj + l θj where l θj is the orthogoal projectio of l θj ito the closure of the rage of l η. Thus l θj is the orthogoal projectio of l θj ito R( l η ). We call l θ = ( l θ1,..., l θk )T the efficiet score fuctio for θ, ad the efficiet iformatio for θ is the give by I θ,η = l θ T l θ dp θ,η. Cosider ow the map κ : P R m give by κ(p θ,η ) = χ(θ). Also cosider χ(θ) as a fuctio of the full parameter; thus χ(θ) = µ(θ, η). It is ow easy to see

8 642 MOULINATH BANERJEE that t (κ(p θ+th,η t )) = t=0 t (µ(θ + th, η t)) t=0 χ(θ + th) χ(θ) = lim = χ (θ) h = µ (h, β), t 0 t µ beig a bouded liear map from R k C R m. As before, a ecessary ad sufficiet coditio for pathwise orm differetiability of κ is the existece of ġ i0 s i R(A) for i = 1,..., m such that µ (e i ) = A ġ i0 for each i. Usig argumets similar to Va der Vaart (1991) we ca easily show that the ecessary ad sufficiet coditio traslates to N (I θ,η ) N (χ (θ)), which is the case if I θ,η is ivertible. I what follows, we assume that this is the case. We also assume that χ (θ) is of full row rak (m). The efficiet ifluece fuctio g 0 = (ġ 10,..., ġ m0 ) (as before idetified with the values at the basis vectors) is the give by g 0 = χ (θ)iθ,η 1 l θ, ad the dispersio matrix for g 0, which acts as the iformatio boud for the estimatio of χ, is give by J = χ (θ)iθ,η 1 χ (θ) T ad is ivertible uder our assumptios. Deote the parameter (θ, η) by ψ. Cosider the problem of testig the ull hypothesis χ = χ 0 agaist χ χ 0. Let (θ 0, η 0 ) H 0 be the true value of the parameter. Assume that all coditios of Theorem 2.1 hold ad that (2.1) is satisfied with I0 1 l = g 0. Here χ plays the role of θ i Theorem 2.1 while (θ, η) plays the role of ψ; thus we have (ˆχ χ 0 ) = I0 1 P l + o p (1) uder ψ 0 with ( ) I 0 = χ (θ)i 1 θ,η χ (θ) T 1 = J 1 ad l = I 0 g 0. Now, cosider local alteratives of the form ψ = ψ 0 + h ( ) 1 + o = (θ 0 + h 1 /, η 0 + h 2 / + o(1/ )). Usig Theorem 2.2 we readily deduce that uder ψ, lrt (χ 0 ) d χ 2 m (ct Ic) with c = I0 1 Ah, l Pψ0 ad h = (h 1, h 2 ). Sice χ, cosidered as a fuctioal defied o the space of probabilities (the map κ), is pathwise orm differetiable at ψ 0, by the assumptio of ivertibility of I θ0,η 0 we have that c = t ( κ(p θ0 +th 1,η 0 +th 2 +o(t))) t=0 = χ (θ 0 ) h 1. Thus the fial form for the o-cetrality parameter is = h T 1 χ (θ 0 ) T ( χ (θ 0 )I 1 θ 0,η 0 χ (θ 0 ) T ) 1 χ (θ 0 )h 1. Whe χ(θ) = θ ( which happes i may cases), χ (θ 0 ) is the idetity matrix ad the expressio for the o-cetrality parameter reduces to = h T 1 I θ 0,η 0 h 1 ;

9 LIKELIHOOD RATIO TESTS UNDER LOCAL ALTERNATIVES 643 here I θ0,η 0 is the efficiet iformatio for estimatig θ at parameter value ψ 0. This expressio is exactly what oe gets i the regular parametric case, where η is fiite dimesioal, sice the discussio above icludes the special case whe η, the uisace parameter varies i a fiite dimesioal space. Applicatio to the Cox Proportioal Hazards Models. The Cox Proportioal Hazards Model is probably the most widely used of all semiparametric models. The distributio of the likelihood ratio statistic for testig H 0 : θ = θ 0 i the Cox Proportioal Hazards settig has bee studied uder various cesorig schemes. Two importat schemes are the followig: (i) Right Cesorig here we observe X = (T C, 1 {T C}, Z) where, give (the covariate) Z, the variables T (survival time) ad C (observatio time) are idepedet ad T follows the Cox Model. (ii) Iterval Cesorig here we observe X = (1 {T C}, Z); thus, we have less iformatio o the survival time distributio compared to the right cesorig sceario. For both right cesorig ad iterval cesorig, θ is estimable at rate by its MLE which is asymptotically ormal ad efficiet, ad the likelihood ratio test for testig H 0 : θ = θ 0 coverges uder the ull hypothesis to a χ 2 distributio. With right-cesorig Λ, the baselie hazard, ca be estimated at rate (by its MLE), but with iterval cesorig the rate slows to 1/3. For more details o likelihood based estimatio ad iferece o the Cox Model with right cesored data see, for example, Va der Vaart (1998), Baerjee (2000) ad Murphy ad Va der Vaart (2000). Maximum likelihood estimatio i the Cox model with iterval cesored data was first cosidered i Huag (1996), where the MLE of θ was show to asymptotically ormal ad efficiet. The likelihood ratio statistic for testig H 0 : θ = θ 0 was show to be asymptotically χ 2 d (d beig the dimesio of θ) uder the ull hypothesis by Murphy ad Va der Vaart (1997, 2000), uder appropriate regularity coditios. Natural local alteratives to cosider i the Cox PH settig with right-cesored data are of the form (θ 0 + h 1 /, Λ 0 + (1/ ) 0 h 2 d Λ 0 ), where h 2 is a bouded fuctio i L 2 (Λ 0 ). Usig the ideas of this paper, we ca show (uder appropriate assumptios) that lrt (θ 0 ), the likelihood ratio statistic for testig θ = θ 0, coverges i distributio, uder the sequece of local alteratives above, to χ 2 d (ht 1 I 0 h 1 ), where I 0 is the efficiet iformatio for θ i the right cesored problem at parameter value (θ 0, Λ 0 ). For the details, see Baerjee (2000). I the Cox PH settig with iterval cesored data, the atural local alteratives take the form (θ 0 + h 1 /, Λ 0 + h 2 / ) where h 2 is a o-egative o-decreasig cotiuous fuctio. Oce agai, it ca be show that the power of the likelihood ratio statistic for testig H 0 : θ = θ 0 coverges, uder the above sequece of local alteratives, to χ 2 d (ht 1 Ĩ0 h 1 ); here Ĩ0 is the efficiet iformatio for θ i the iterval cesorig set-up. For a explicit represetatio of Ĩ0, see Huag (1996).

10 644 MOULINATH BANERJEE For more cocrete applicatios of the results i this paper, we refer the reader to Chapter 2 of Baerjee (2000). Ackowledgemets Thaks to Susa Murphy ad Jo Weller for helpful commets ad discussios. This research was supported partially by Natioal Sciece Foudatio grats DMS ad DMS Refereces Baerjee, M. (2000). Likelihood ratio iferece i regular ad oregular problems. Ph.D. dissertatio, Uiversity of Washigto. Bickel, P. J., Klaasse, C. A. J., Ritov, Y. ad Weller, J. A. (1998). Efficiet ad Adaptive Estimatio for Semiparametric Models. Spriger Verlag, New York. Egle, R., Grager, C., Rice, J. ad Weiss, A. (1986). Semiparametric estimatio of the relatio betwee weather ad electricity sales. J. Amer. Statist. Assoc. 81, Gree, P. (1984). Iteratively reweighted least squares for some robust ad resistat alteratives. J. Roy. Statist. Soc. Ser. B 46, Huag, J. (1996). Efficiet estimatio for the Cox model with iterval cesorig. A. Statist. 24, Murphy, S. ad Va der Vaart, A. (1997). Semiparametric likelihod ratio iferece. A. Statist. 25, Murphy, S. ad Va der Vaart, A. (2000). O profile likelihood. J. Amer. Statist. Assoc. 95, Murphy, S. ad Va der Vaart, A. (2001). Semiparametric mixtures i case cotrol studies. J. Multivariate Aal. 79, Roeder, K., Carroll, R. J. ad Lidsay, B. G. (1996). A semiparametric mixture approach to maximum likelihood estimatio i frailty models. Scad. J. Statist. 19, Va der Vaart, A. (1991). O differetiable fuctioals. A. Statist. 19, Va der Vaart, A. ad Weller, J. A. (1996). Weak Covergece ad Empirical Processes. Spriger, New York. Va der Vaart, A. (1998). Asymptotic Statistics. Cambridge Uiversity Press, Cambridge. Uiversity of Michiga, Departmet of Statistics, 439, West Hall, 1085 South Uiversity, A Arbor, MI, 48109, U.S.A. moulib@umich.edu (Received May 2003; accepted May 2004)

Empirical Processes: Glivenko Cantelli Theorems

Empirical Processes: Glivenko Cantelli Theorems Empirical Processes: Gliveko Catelli Theorems Mouliath Baerjee Jue 6, 200 Gliveko Catelli classes of fuctios The reader is referred to Chapter.6 of Weller s Torgo otes, Chapter??? of VDVW ad Chapter 8.3

More information

Lecture 27: Optimal Estimators and Functional Delta Method

Lecture 27: Optimal Estimators and Functional Delta Method Stat210B: Theoretical Statistics Lecture Date: April 19, 2007 Lecture 27: Optimal Estimators ad Fuctioal Delta Method Lecturer: Michael I. Jorda Scribe: Guilherme V. Rocha 1 Achievig Optimal Estimators

More information

Rank tests and regression rank scores tests in measurement error models

Rank tests and regression rank scores tests in measurement error models Rak tests ad regressio rak scores tests i measuremet error models J. Jurečková ad A.K.Md.E. Saleh Charles Uiversity i Prague ad Carleto Uiversity i Ottawa Abstract The rak ad regressio rak score tests

More information

Statistical Inference Based on Extremum Estimators

Statistical Inference Based on Extremum Estimators T. Rotheberg Fall, 2007 Statistical Iferece Based o Extremum Estimators Itroductio Suppose 0, the true value of a p-dimesioal parameter, is kow to lie i some subset S R p : Ofte we choose to estimate 0

More information

Efficient GMM LECTURE 12 GMM II

Efficient GMM LECTURE 12 GMM II DECEMBER 1 010 LECTURE 1 II Efficiet The estimator depeds o the choice of the weight matrix A. The efficiet estimator is the oe that has the smallest asymptotic variace amog all estimators defied by differet

More information

Regression with an Evaporating Logarithmic Trend

Regression with an Evaporating Logarithmic Trend Regressio with a Evaporatig Logarithmic Tred Peter C. B. Phillips Cowles Foudatio, Yale Uiversity, Uiversity of Aucklad & Uiversity of York ad Yixiao Su Departmet of Ecoomics Yale Uiversity October 5,

More information

LECTURE 14 NOTES. A sequence of α-level tests {ϕ n (x)} is consistent if

LECTURE 14 NOTES. A sequence of α-level tests {ϕ n (x)} is consistent if LECTURE 14 NOTES 1. Asymptotic power of tests. Defiitio 1.1. A sequece of -level tests {ϕ x)} is cosistet if β θ) := E θ [ ϕ x) ] 1 as, for ay θ Θ 1. Just like cosistecy of a sequece of estimators, Defiitio

More information

A RANK STATISTIC FOR NON-PARAMETRIC K-SAMPLE AND CHANGE POINT PROBLEMS

A RANK STATISTIC FOR NON-PARAMETRIC K-SAMPLE AND CHANGE POINT PROBLEMS J. Japa Statist. Soc. Vol. 41 No. 1 2011 67 73 A RANK STATISTIC FOR NON-PARAMETRIC K-SAMPLE AND CHANGE POINT PROBLEMS Yoichi Nishiyama* We cosider k-sample ad chage poit problems for idepedet data i a

More information

Definition 4.2. (a) A sequence {x n } in a Banach space X is a basis for X if. unique scalars a n (x) such that x = n. a n (x) x n. (4.

Definition 4.2. (a) A sequence {x n } in a Banach space X is a basis for X if. unique scalars a n (x) such that x = n. a n (x) x n. (4. 4. BASES I BAACH SPACES 39 4. BASES I BAACH SPACES Sice a Baach space X is a vector space, it must possess a Hamel, or vector space, basis, i.e., a subset {x γ } γ Γ whose fiite liear spa is all of X ad

More information

Measure and Measurable Functions

Measure and Measurable Functions 3 Measure ad Measurable Fuctios 3.1 Measure o a Arbitrary σ-algebra Recall from Chapter 2 that the set M of all Lebesgue measurable sets has the followig properties: R M, E M implies E c M, E M for N implies

More information

( θ. sup θ Θ f X (x θ) = L. sup Pr (Λ (X) < c) = α. x : Λ (x) = sup θ H 0. sup θ Θ f X (x θ) = ) < c. NH : θ 1 = θ 2 against AH : θ 1 θ 2

( θ. sup θ Θ f X (x θ) = L. sup Pr (Λ (X) < c) = α. x : Λ (x) = sup θ H 0. sup θ Θ f X (x θ) = ) < c. NH : θ 1 = θ 2 against AH : θ 1 θ 2 82 CHAPTER 4. MAXIMUM IKEIHOOD ESTIMATION Defiitio: et X be a radom sample with joit p.m/d.f. f X x θ. The geeralised likelihood ratio test g.l.r.t. of the NH : θ H 0 agaist the alterative AH : θ H 1,

More information

ON MEAN ERGODIC CONVERGENCE IN THE CALKIN ALGEBRAS

ON MEAN ERGODIC CONVERGENCE IN THE CALKIN ALGEBRAS PROCEEDINGS OF THE AMERICAN MATHEMATICAL SOCIETY Volume 00, Number 0, Pages 000 000 S 0002-9939(XX0000-0 ON MEAN ERGODIC CONVERGENCE IN THE CALKIN ALGEBRAS MARCH T. BOEDIHARDJO AND WILLIAM B. JOHNSON 2

More information

Convergence of random variables. (telegram style notes) P.J.C. Spreij

Convergence of random variables. (telegram style notes) P.J.C. Spreij Covergece of radom variables (telegram style otes).j.c. Spreij this versio: September 6, 2005 Itroductio As we kow, radom variables are by defiitio measurable fuctios o some uderlyig measurable space

More information

Inference for Alternating Time Series

Inference for Alternating Time Series Iferece for Alteratig Time Series Ursula U. Müller 1, Ato Schick 2, ad Wolfgag Wefelmeyer 3 1 Departmet of Statistics Texas A&M Uiversity College Statio, TX 77843-3143, USA (e-mail: uschi@stat.tamu.edu)

More information

Chapter 3. Strong convergence. 3.1 Definition of almost sure convergence

Chapter 3. Strong convergence. 3.1 Definition of almost sure convergence Chapter 3 Strog covergece As poited out i the Chapter 2, there are multiple ways to defie the otio of covergece of a sequece of radom variables. That chapter defied covergece i probability, covergece i

More information

Chi-Squared Tests Math 6070, Spring 2006

Chi-Squared Tests Math 6070, Spring 2006 Chi-Squared Tests Math 6070, Sprig 2006 Davar Khoshevisa Uiversity of Utah February XXX, 2006 Cotets MLE for Goodess-of Fit 2 2 The Multiomial Distributio 3 3 Applicatio to Goodess-of-Fit 6 3 Testig for

More information

Linear regression. Daniel Hsu (COMS 4771) (y i x T i β)2 2πσ. 2 2σ 2. 1 n. (x T i β y i ) 2. 1 ˆβ arg min. β R n d

Linear regression. Daniel Hsu (COMS 4771) (y i x T i β)2 2πσ. 2 2σ 2. 1 n. (x T i β y i ) 2. 1 ˆβ arg min. β R n d Liear regressio Daiel Hsu (COMS 477) Maximum likelihood estimatio Oe of the simplest liear regressio models is the followig: (X, Y ),..., (X, Y ), (X, Y ) are iid radom pairs takig values i R d R, ad Y

More information

MASSACHUSETTS INSTITUTE OF TECHNOLOGY 6.436J/15.085J Fall 2008 Lecture 19 11/17/2008 LAWS OF LARGE NUMBERS II THE STRONG LAW OF LARGE NUMBERS

MASSACHUSETTS INSTITUTE OF TECHNOLOGY 6.436J/15.085J Fall 2008 Lecture 19 11/17/2008 LAWS OF LARGE NUMBERS II THE STRONG LAW OF LARGE NUMBERS MASSACHUSTTS INSTITUT OF TCHNOLOGY 6.436J/5.085J Fall 2008 Lecture 9 /7/2008 LAWS OF LARG NUMBRS II Cotets. The strog law of large umbers 2. The Cheroff boud TH STRONG LAW OF LARG NUMBRS While the weak

More information

MATH 320: Probability and Statistics 9. Estimation and Testing of Parameters. Readings: Pruim, Chapter 4

MATH 320: Probability and Statistics 9. Estimation and Testing of Parameters. Readings: Pruim, Chapter 4 MATH 30: Probability ad Statistics 9. Estimatio ad Testig of Parameters Estimatio ad Testig of Parameters We have bee dealig situatios i which we have full kowledge of the distributio of a radom variable.

More information

ECONOMETRIC THEORY. MODULE XIII Lecture - 34 Asymptotic Theory and Stochastic Regressors

ECONOMETRIC THEORY. MODULE XIII Lecture - 34 Asymptotic Theory and Stochastic Regressors ECONOMETRIC THEORY MODULE XIII Lecture - 34 Asymptotic Theory ad Stochastic Regressors Dr. Shalabh Departmet of Mathematics ad Statistics Idia Istitute of Techology Kapur Asymptotic theory The asymptotic

More information

Kernel density estimator

Kernel density estimator Jauary, 07 NONPARAMETRIC ERNEL DENSITY ESTIMATION I this lecture, we discuss kerel estimatio of probability desity fuctios PDF Noparametric desity estimatio is oe of the cetral problems i statistics I

More information

Chapter 7 Isoperimetric problem

Chapter 7 Isoperimetric problem Chapter 7 Isoperimetric problem Recall that the isoperimetric problem (see the itroductio its coectio with ido s proble) is oe of the most classical problem of a shape optimizatio. It ca be formulated

More information

Riesz-Fischer Sequences and Lower Frame Bounds

Riesz-Fischer Sequences and Lower Frame Bounds Zeitschrift für Aalysis ud ihre Aweduge Joural for Aalysis ad its Applicatios Volume 1 (00), No., 305 314 Riesz-Fischer Sequeces ad Lower Frame Bouds P. Casazza, O. Christese, S. Li ad A. Lider Abstract.

More information

An Introduction to Asymptotic Theory

An Introduction to Asymptotic Theory A Itroductio to Asymptotic Theory Pig Yu School of Ecoomics ad Fiace The Uiversity of Hog Kog Pig Yu (HKU) Asymptotic Theory 1 / 20 Five Weapos i Asymptotic Theory Five Weapos i Asymptotic Theory Pig Yu

More information

Estimation for Complete Data

Estimation for Complete Data Estimatio for Complete Data complete data: there is o loss of iformatio durig study. complete idividual complete data= grouped data A complete idividual data is the oe i which the complete iformatio of

More information

Lecture 7: Properties of Random Samples

Lecture 7: Properties of Random Samples Lecture 7: Properties of Radom Samples 1 Cotiued From Last Class Theorem 1.1. Let X 1, X,...X be a radom sample from a populatio with mea µ ad variace σ

More information

Introductory statistics

Introductory statistics CM9S: Machie Learig for Bioiformatics Lecture - 03/3/06 Itroductory statistics Lecturer: Sriram Sakararama Scribe: Sriram Sakararama We will provide a overview of statistical iferece focussig o the key

More information

Precise Rates in Complete Moment Convergence for Negatively Associated Sequences

Precise Rates in Complete Moment Convergence for Negatively Associated Sequences Commuicatios of the Korea Statistical Society 29, Vol. 16, No. 5, 841 849 Precise Rates i Complete Momet Covergece for Negatively Associated Sequeces Dae-Hee Ryu 1,a a Departmet of Computer Sciece, ChugWoo

More information

Economics 241B Relation to Method of Moments and Maximum Likelihood OLSE as a Maximum Likelihood Estimator

Economics 241B Relation to Method of Moments and Maximum Likelihood OLSE as a Maximum Likelihood Estimator Ecoomics 24B Relatio to Method of Momets ad Maximum Likelihood OLSE as a Maximum Likelihood Estimator Uder Assumptio 5 we have speci ed the distributio of the error, so we ca estimate the model parameters

More information

Lecture 8: Convergence of transformations and law of large numbers

Lecture 8: Convergence of transformations and law of large numbers Lecture 8: Covergece of trasformatios ad law of large umbers Trasformatio ad covergece Trasformatio is a importat tool i statistics. If X coverges to X i some sese, we ofte eed to check whether g(x ) coverges

More information

7.1 Convergence of sequences of random variables

7.1 Convergence of sequences of random variables Chapter 7 Limit Theorems Throughout this sectio we will assume a probability space (, F, P), i which is defied a ifiite sequece of radom variables (X ) ad a radom variable X. The fact that for every ifiite

More information

Math Solutions to homework 6

Math Solutions to homework 6 Math 175 - Solutios to homework 6 Cédric De Groote November 16, 2017 Problem 1 (8.11 i the book): Let K be a compact Hermitia operator o a Hilbert space H ad let the kerel of K be {0}. Show that there

More information

62. Power series Definition 16. (Power series) Given a sequence {c n }, the series. c n x n = c 0 + c 1 x + c 2 x 2 + c 3 x 3 +

62. Power series Definition 16. (Power series) Given a sequence {c n }, the series. c n x n = c 0 + c 1 x + c 2 x 2 + c 3 x 3 + 62. Power series Defiitio 16. (Power series) Give a sequece {c }, the series c x = c 0 + c 1 x + c 2 x 2 + c 3 x 3 + is called a power series i the variable x. The umbers c are called the coefficiets of

More information

Journal of Multivariate Analysis. Superefficient estimation of the marginals by exploiting knowledge on the copula

Journal of Multivariate Analysis. Superefficient estimation of the marginals by exploiting knowledge on the copula Joural of Multivariate Aalysis 102 (2011) 1315 1319 Cotets lists available at ScieceDirect Joural of Multivariate Aalysis joural homepage: www.elsevier.com/locate/jmva Superefficiet estimatio of the margials

More information

Singular Continuous Measures by Michael Pejic 5/14/10

Singular Continuous Measures by Michael Pejic 5/14/10 Sigular Cotiuous Measures by Michael Peic 5/4/0 Prelimiaries Give a set X, a σ-algebra o X is a collectio of subsets of X that cotais X ad ad is closed uder complemetatio ad coutable uios hece, coutable

More information

Product measures, Tonelli s and Fubini s theorems For use in MAT3400/4400, autumn 2014 Nadia S. Larsen. Version of 13 October 2014.

Product measures, Tonelli s and Fubini s theorems For use in MAT3400/4400, autumn 2014 Nadia S. Larsen. Version of 13 October 2014. Product measures, Toelli s ad Fubii s theorems For use i MAT3400/4400, autum 2014 Nadia S. Larse Versio of 13 October 2014. 1. Costructio of the product measure The purpose of these otes is to preset the

More information

STAT331. Example of Martingale CLT with Cox s Model

STAT331. Example of Martingale CLT with Cox s Model STAT33 Example of Martigale CLT with Cox s Model I this uit we illustrate the Martigale Cetral Limit Theorem by applyig it to the partial likelihood score fuctio from Cox s model. For simplicity of presetatio

More information

Entropy Rates and Asymptotic Equipartition

Entropy Rates and Asymptotic Equipartition Chapter 29 Etropy Rates ad Asymptotic Equipartitio Sectio 29. itroduces the etropy rate the asymptotic etropy per time-step of a stochastic process ad shows that it is well-defied; ad similarly for iformatio,

More information

Dimension-free PAC-Bayesian bounds for the estimation of the mean of a random vector

Dimension-free PAC-Bayesian bounds for the estimation of the mean of a random vector Dimesio-free PAC-Bayesia bouds for the estimatio of the mea of a radom vector Olivier Catoi CREST CNRS UMR 9194 Uiversité Paris Saclay olivier.catoi@esae.fr Ilaria Giulii Laboratoire de Probabilités et

More information

Probability and Statistics

Probability and Statistics ICME Refresher Course: robability ad Statistics Staford Uiversity robability ad Statistics Luyag Che September 20, 2016 1 Basic robability Theory 11 robability Spaces A probability space is a triple (Ω,

More information

Chapter 6 Principles of Data Reduction

Chapter 6 Principles of Data Reduction Chapter 6 for BST 695: Special Topics i Statistical Theory. Kui Zhag, 0 Chapter 6 Priciples of Data Reductio Sectio 6. Itroductio Goal: To summarize or reduce the data X, X,, X to get iformatio about a

More information

Lecture 23: Minimal sufficiency

Lecture 23: Minimal sufficiency Lecture 23: Miimal sufficiecy Maximal reductio without loss of iformatio There are may sufficiet statistics for a give problem. I fact, X (the whole data set) is sufficiet. If T is a sufficiet statistic

More information

Random Walks on Discrete and Continuous Circles. by Jeffrey S. Rosenthal School of Mathematics, University of Minnesota, Minneapolis, MN, U.S.A.

Random Walks on Discrete and Continuous Circles. by Jeffrey S. Rosenthal School of Mathematics, University of Minnesota, Minneapolis, MN, U.S.A. Radom Walks o Discrete ad Cotiuous Circles by Jeffrey S. Rosethal School of Mathematics, Uiversity of Miesota, Mieapolis, MN, U.S.A. 55455 (Appeared i Joural of Applied Probability 30 (1993), 780 789.)

More information

Summary and Discussion on Simultaneous Analysis of Lasso and Dantzig Selector

Summary and Discussion on Simultaneous Analysis of Lasso and Dantzig Selector Summary ad Discussio o Simultaeous Aalysis of Lasso ad Datzig Selector STAT732, Sprig 28 Duzhe Wag May 4, 28 Abstract This is a discussio o the work i Bickel, Ritov ad Tsybakov (29). We begi with a short

More information

LECTURE 8: ASYMPTOTICS I

LECTURE 8: ASYMPTOTICS I LECTURE 8: ASYMPTOTICS I We are iterested i the properties of estimators as. Cosider a sequece of radom variables {, X 1}. N. M. Kiefer, Corell Uiversity, Ecoomics 60 1 Defiitio: (Weak covergece) A sequece

More information

17. Joint distributions of extreme order statistics Lehmann 5.1; Ferguson 15

17. Joint distributions of extreme order statistics Lehmann 5.1; Ferguson 15 17. Joit distributios of extreme order statistics Lehma 5.1; Ferguso 15 I Example 10., we derived the asymptotic distributio of the maximum from a radom sample from a uiform distributio. We did this usig

More information

Last Lecture. Wald Test

Last Lecture. Wald Test Last Lecture Biostatistics 602 - Statistical Iferece Lecture 22 Hyu Mi Kag April 9th, 2013 Is the exact distributio of LRT statistic typically easy to obtai? How about its asymptotic distributio? For testig

More information

On the convergence rates of Gladyshev s Hurst index estimator

On the convergence rates of Gladyshev s Hurst index estimator Noliear Aalysis: Modellig ad Cotrol, 2010, Vol 15, No 4, 445 450 O the covergece rates of Gladyshev s Hurst idex estimator K Kubilius 1, D Melichov 2 1 Istitute of Mathematics ad Iformatics, Vilius Uiversity

More information

Resampling Methods. X (1/2), i.e., Pr (X i m) = 1/2. We order the data: X (1) X (2) X (n). Define the sample median: ( n.

Resampling Methods. X (1/2), i.e., Pr (X i m) = 1/2. We order the data: X (1) X (2) X (n). Define the sample median: ( n. Jauary 1, 2019 Resamplig Methods Motivatio We have so may estimators with the property θ θ d N 0, σ 2 We ca also write θ a N θ, σ 2 /, where a meas approximately distributed as Oce we have a cosistet estimator

More information

MASSACHUSETTS INSTITUTE OF TECHNOLOGY 6.265/15.070J Fall 2013 Lecture 21 11/27/2013

MASSACHUSETTS INSTITUTE OF TECHNOLOGY 6.265/15.070J Fall 2013 Lecture 21 11/27/2013 MASSACHUSETTS INSTITUTE OF TECHNOLOGY 6.265/15.070J Fall 2013 Lecture 21 11/27/2013 Fuctioal Law of Large Numbers. Costructio of the Wieer Measure Cotet. 1. Additioal techical results o weak covergece

More information

An Empirical Likelihood Approach To Goodness of Fit Testing

An Empirical Likelihood Approach To Goodness of Fit Testing Submitted to the Beroulli A Empirical Likelihood Approach To Goodess of Fit Testig HANXIANG PENG ad ANTON SCHICK Idiaa Uiversity Purdue Uiversity at Idiaapolis, Departmet of Mathematical Scieces, Idiaapolis,

More information

Asymptotic distribution of products of sums of independent random variables

Asymptotic distribution of products of sums of independent random variables Proc. Idia Acad. Sci. Math. Sci. Vol. 3, No., May 03, pp. 83 9. c Idia Academy of Scieces Asymptotic distributio of products of sums of idepedet radom variables YANLING WANG, SUXIA YAO ad HONGXIA DU ollege

More information

TENSOR PRODUCTS AND PARTIAL TRACES

TENSOR PRODUCTS AND PARTIAL TRACES Lecture 2 TENSOR PRODUCTS AND PARTIAL TRACES Stéphae ATTAL Abstract This lecture cocers special aspects of Operator Theory which are of much use i Quatum Mechaics, i particular i the theory of Quatum Ope

More information

11 THE GMM ESTIMATION

11 THE GMM ESTIMATION Cotets THE GMM ESTIMATION 2. Cosistecy ad Asymptotic Normality..................... 3.2 Regularity Coditios ad Idetificatio..................... 4.3 The GMM Iterpretatio of the OLS Estimatio.................

More information

Summary. Recap ... Last Lecture. Summary. Theorem

Summary. Recap ... Last Lecture. Summary. Theorem Last Lecture Biostatistics 602 - Statistical Iferece Lecture 23 Hyu Mi Kag April 11th, 2013 What is p-value? What is the advatage of p-value compared to hypothesis testig procedure with size α? How ca

More information

MOMENT-METHOD ESTIMATION BASED ON CENSORED SAMPLE

MOMENT-METHOD ESTIMATION BASED ON CENSORED SAMPLE Vol. 8 o. Joural of Systems Sciece ad Complexity Apr., 5 MOMET-METHOD ESTIMATIO BASED O CESORED SAMPLE I Zhogxi Departmet of Mathematics, East Chia Uiversity of Sciece ad Techology, Shaghai 37, Chia. Email:

More information

Lecture 19: Convergence

Lecture 19: Convergence Lecture 19: Covergece Asymptotic approach I statistical aalysis or iferece, a key to the success of fidig a good procedure is beig able to fid some momets ad/or distributios of various statistics. I may

More information

Exponential Families and Bayesian Inference

Exponential Families and Bayesian Inference Computer Visio Expoetial Families ad Bayesia Iferece Lecture Expoetial Families A expoetial family of distributios is a d-parameter family f(x; havig the followig form: f(x; = h(xe g(t T (x B(, (. where

More information

Advanced Stochastic Processes.

Advanced Stochastic Processes. Advaced Stochastic Processes. David Gamarik LECTURE 2 Radom variables ad measurable fuctios. Strog Law of Large Numbers (SLLN). Scary stuff cotiued... Outlie of Lecture Radom variables ad measurable fuctios.

More information

Appendix to: Hypothesis Testing for Multiple Mean and Correlation Curves with Functional Data

Appendix to: Hypothesis Testing for Multiple Mean and Correlation Curves with Functional Data Appedix to: Hypothesis Testig for Multiple Mea ad Correlatio Curves with Fuctioal Data Ao Yua 1, Hog-Bi Fag 1, Haiou Li 1, Coli O. Wu, Mig T. Ta 1, 1 Departmet of Biostatistics, Bioiformatics ad Biomathematics,

More information

Lecture 33: Bootstrap

Lecture 33: Bootstrap Lecture 33: ootstrap Motivatio To evaluate ad compare differet estimators, we eed cosistet estimators of variaces or asymptotic variaces of estimators. This is also importat for hypothesis testig ad cofidece

More information

Chi-squared tests Math 6070, Spring 2014

Chi-squared tests Math 6070, Spring 2014 Chi-squared tests Math 6070, Sprig 204 Davar Khoshevisa Uiversity of Utah March, 204 Cotets MLE for goodess-of fit 2 2 The Multivariate ormal distributio 3 3 Cetral limit theorems 5 4 Applicatio to goodess-of-fit

More information

A Note on the Kolmogorov-Feller Weak Law of Large Numbers

A Note on the Kolmogorov-Feller Weak Law of Large Numbers Joural of Mathematical Research with Applicatios Mar., 015, Vol. 35, No., pp. 3 8 DOI:10.3770/j.iss:095-651.015.0.013 Http://jmre.dlut.edu.c A Note o the Kolmogorov-Feller Weak Law of Large Numbers Yachu

More information

It should be unbiased, or approximately unbiased. Variance of the variance estimator should be small. That is, the variance estimator is stable.

It should be unbiased, or approximately unbiased. Variance of the variance estimator should be small. That is, the variance estimator is stable. Chapter 10 Variace Estimatio 10.1 Itroductio Variace estimatio is a importat practical problem i survey samplig. Variace estimates are used i two purposes. Oe is the aalytic purpose such as costructig

More information

A NOTE ON SPECTRAL CONTINUITY. In Ho Jeon and In Hyoun Kim

A NOTE ON SPECTRAL CONTINUITY. In Ho Jeon and In Hyoun Kim Korea J. Math. 23 (2015), No. 4, pp. 601 605 http://dx.doi.org/10.11568/kjm.2015.23.4.601 A NOTE ON SPECTRAL CONTINUITY I Ho Jeo ad I Hyou Kim Abstract. I the preset ote, provided T L (H ) is biquasitriagular

More information

Berry-Esseen bounds for self-normalized martingales

Berry-Esseen bounds for self-normalized martingales Berry-Essee bouds for self-ormalized martigales Xiequa Fa a, Qi-Ma Shao b a Ceter for Applied Mathematics, Tiaji Uiversity, Tiaji 30007, Chia b Departmet of Statistics, The Chiese Uiversity of Hog Kog,

More information

Slide Set 13 Linear Model with Endogenous Regressors and the GMM estimator

Slide Set 13 Linear Model with Endogenous Regressors and the GMM estimator Slide Set 13 Liear Model with Edogeous Regressors ad the GMM estimator Pietro Coretto pcoretto@uisa.it Ecoometrics Master i Ecoomics ad Fiace (MEF) Uiversità degli Studi di Napoli Federico II Versio: Friday

More information

MASSACHUSETTS INSTITUTE OF TECHNOLOGY 6.265/15.070J Fall 2013 Lecture 3 9/11/2013. Large deviations Theory. Cramér s Theorem

MASSACHUSETTS INSTITUTE OF TECHNOLOGY 6.265/15.070J Fall 2013 Lecture 3 9/11/2013. Large deviations Theory. Cramér s Theorem MASSACHUSETTS INSTITUTE OF TECHNOLOGY 6.265/5.070J Fall 203 Lecture 3 9//203 Large deviatios Theory. Cramér s Theorem Cotet.. Cramér s Theorem. 2. Rate fuctio ad properties. 3. Chage of measure techique.

More information

5 : Exponential Family and Generalized Linear Models

5 : Exponential Family and Generalized Linear Models 0-708: Probabilistic Graphical Models 0-708, Sprig 206 5 : Expoetial Family ad Geeralized Liear Models Lecturer: Matthew Gormley Scribes: Yua Li, Yichog Xu, Silu Wag Expoetial Family Probability desity

More information

The random version of Dvoretzky s theorem in l n

The random version of Dvoretzky s theorem in l n The radom versio of Dvoretzky s theorem i l Gideo Schechtma Abstract We show that with high probability a sectio of the l ball of dimesio k cε log c > 0 a uiversal costat) is ε close to a multiple of the

More information

Mi-Hwa Ko and Tae-Sung Kim

Mi-Hwa Ko and Tae-Sung Kim J. Korea Math. Soc. 42 2005), No. 5, pp. 949 957 ALMOST SURE CONVERGENCE FOR WEIGHTED SUMS OF NEGATIVELY ORTHANT DEPENDENT RANDOM VARIABLES Mi-Hwa Ko ad Tae-Sug Kim Abstract. For weighted sum of a sequece

More information

Sequences and Series of Functions

Sequences and Series of Functions Chapter 6 Sequeces ad Series of Fuctios 6.1. Covergece of a Sequece of Fuctios Poitwise Covergece. Defiitio 6.1. Let, for each N, fuctio f : A R be defied. If, for each x A, the sequece (f (x)) coverges

More information

Probabilistic and Average Linear Widths in L -Norm with Respect to r-fold Wiener Measure

Probabilistic and Average Linear Widths in L -Norm with Respect to r-fold Wiener Measure joural of approximatio theory 84, 3140 (1996) Article No. 0003 Probabilistic ad Average Liear Widths i L -Norm with Respect to r-fold Wieer Measure V. E. Maiorov Departmet of Mathematics, Techio, Haifa,

More information

MATHEMATICAL SCIENCES PAPER-II

MATHEMATICAL SCIENCES PAPER-II MATHEMATICAL SCIENCES PAPER-II. Let {x } ad {y } be two sequeces of real umbers. Prove or disprove each of the statemets :. If {x y } coverges, ad if {y } is coverget, the {x } is coverget.. {x + y } coverges

More information

Lecture 3 The Lebesgue Integral

Lecture 3 The Lebesgue Integral Lecture 3: The Lebesgue Itegral 1 of 14 Course: Theory of Probability I Term: Fall 2013 Istructor: Gorda Zitkovic Lecture 3 The Lebesgue Itegral The costructio of the itegral Uless expressly specified

More information

Mathematical Statistics - MS

Mathematical Statistics - MS Paper Specific Istructios. The examiatio is of hours duratio. There are a total of 60 questios carryig 00 marks. The etire paper is divided ito three sectios, A, B ad C. All sectios are compulsory. Questios

More information

Kolmogorov-Smirnov type Tests for Local Gaussianity in High-Frequency Data

Kolmogorov-Smirnov type Tests for Local Gaussianity in High-Frequency Data Proceedigs 59th ISI World Statistics Cogress, 5-30 August 013, Hog Kog (Sessio STS046) p.09 Kolmogorov-Smirov type Tests for Local Gaussiaity i High-Frequecy Data George Tauche, Duke Uiversity Viktor Todorov,

More information

Notes 5 : More on the a.s. convergence of sums

Notes 5 : More on the a.s. convergence of sums Notes 5 : More o the a.s. covergece of sums Math 733-734: Theory of Probability Lecturer: Sebastie Roch Refereces: Dur0, Sectios.5; Wil9, Sectio 4.7, Shi96, Sectio IV.4, Dur0, Sectio.. Radom series. Three-series

More information

Solutions: Homework 3

Solutions: Homework 3 Solutios: Homework 3 Suppose that the radom variables Y,...,Y satisfy Y i = x i + " i : i =,..., IID where x,...,x R are fixed values ad ",...," Normal(0, )with R + kow. Fid ˆ = MLE( ). IND Solutio: Observe

More information

CEE 522 Autumn Uncertainty Concepts for Geotechnical Engineering

CEE 522 Autumn Uncertainty Concepts for Geotechnical Engineering CEE 5 Autum 005 Ucertaity Cocepts for Geotechical Egieerig Basic Termiology Set A set is a collectio of (mutually exclusive) objects or evets. The sample space is the (collectively exhaustive) collectio

More information

Distribution of Random Samples & Limit theorems

Distribution of Random Samples & Limit theorems STAT/MATH 395 A - PROBABILITY II UW Witer Quarter 2017 Néhémy Lim Distributio of Radom Samples & Limit theorems 1 Distributio of i.i.d. Samples Motivatig example. Assume that the goal of a study is to

More information

A note on self-normalized Dickey-Fuller test for unit root in autoregressive time series with GARCH errors

A note on self-normalized Dickey-Fuller test for unit root in autoregressive time series with GARCH errors Appl. Math. J. Chiese Uiv. 008, 3(): 97-0 A ote o self-ormalized Dickey-Fuller test for uit root i autoregressive time series with GARCH errors YANG Xiao-rog ZHANG Li-xi Abstract. I this article, the uit

More information

Random Variables, Sampling and Estimation

Random Variables, Sampling and Estimation Chapter 1 Radom Variables, Samplig ad Estimatio 1.1 Itroductio This chapter will cover the most importat basic statistical theory you eed i order to uderstad the ecoometric material that will be comig

More information

NEW FAST CONVERGENT SEQUENCES OF EULER-MASCHERONI TYPE

NEW FAST CONVERGENT SEQUENCES OF EULER-MASCHERONI TYPE UPB Sci Bull, Series A, Vol 79, Iss, 207 ISSN 22-7027 NEW FAST CONVERGENT SEQUENCES OF EULER-MASCHERONI TYPE Gabriel Bercu We itroduce two ew sequeces of Euler-Mascheroi type which have fast covergece

More information

4.1 Data processing inequality

4.1 Data processing inequality ECE598: Iformatio-theoretic methods i high-dimesioal statistics Sprig 206 Lecture 4: Total variatio/iequalities betwee f-divergeces Lecturer: Yihog Wu Scribe: Matthew Tsao, Feb 8, 206 [Ed. Mar 22] Recall

More information

A statistical method to determine sample size to estimate characteristic value of soil parameters

A statistical method to determine sample size to estimate characteristic value of soil parameters A statistical method to determie sample size to estimate characteristic value of soil parameters Y. Hojo, B. Setiawa 2 ad M. Suzuki 3 Abstract Sample size is a importat factor to be cosidered i determiig

More information

EFFECTIVE WLLN, SLLN, AND CLT IN STATISTICAL MODELS

EFFECTIVE WLLN, SLLN, AND CLT IN STATISTICAL MODELS EFFECTIVE WLLN, SLLN, AND CLT IN STATISTICAL MODELS Ryszard Zieliński Ist Math Polish Acad Sc POBox 21, 00-956 Warszawa 10, Polad e-mail: rziel@impagovpl ABSTRACT Weak laws of large umbers (W LLN), strog

More information

arxiv: v1 [math.pr] 4 Dec 2013

arxiv: v1 [math.pr] 4 Dec 2013 Squared-Norm Empirical Process i Baach Space arxiv:32005v [mathpr] 4 Dec 203 Vicet Q Vu Departmet of Statistics The Ohio State Uiversity Columbus, OH vqv@statosuedu Abstract Jig Lei Departmet of Statistics

More information

5. Likelihood Ratio Tests

5. Likelihood Ratio Tests 1 of 5 7/29/2009 3:16 PM Virtual Laboratories > 9. Hy pothesis Testig > 1 2 3 4 5 6 7 5. Likelihood Ratio Tests Prelimiaries As usual, our startig poit is a radom experimet with a uderlyig sample space,

More information

Self-normalized deviation inequalities with application to t-statistic

Self-normalized deviation inequalities with application to t-statistic Self-ormalized deviatio iequalities with applicatio to t-statistic Xiequa Fa Ceter for Applied Mathematics, Tiaji Uiversity, 30007 Tiaji, Chia Abstract Let ξ i i 1 be a sequece of idepedet ad symmetric

More information

Common Coupled Fixed Point of Mappings Satisfying Rational Inequalities in Ordered Complex Valued Generalized Metric Spaces

Common Coupled Fixed Point of Mappings Satisfying Rational Inequalities in Ordered Complex Valued Generalized Metric Spaces IOSR Joural of Mathematics (IOSR-JM) e-issn: 78-578, p-issn:319-765x Volume 10, Issue 3 Ver II (May-Ju 014), PP 69-77 Commo Coupled Fixed Poit of Mappigs Satisfyig Ratioal Iequalities i Ordered Complex

More information

On forward improvement iteration for stopping problems

On forward improvement iteration for stopping problems O forward improvemet iteratio for stoppig problems Mathematical Istitute, Uiversity of Kiel, Ludewig-Mey-Str. 4, D-24098 Kiel, Germay irle@math.ui-iel.de Albrecht Irle Abstract. We cosider the optimal

More information

Properties of Fuzzy Length on Fuzzy Set

Properties of Fuzzy Length on Fuzzy Set Ope Access Library Joural 206, Volume 3, e3068 ISSN Olie: 2333-972 ISSN Prit: 2333-9705 Properties of Fuzzy Legth o Fuzzy Set Jehad R Kider, Jaafar Imra Mousa Departmet of Mathematics ad Computer Applicatios,

More information

Lecture 20: Multivariate convergence and the Central Limit Theorem

Lecture 20: Multivariate convergence and the Central Limit Theorem Lecture 20: Multivariate covergece ad the Cetral Limit Theorem Covergece i distributio for radom vectors Let Z,Z 1,Z 2,... be radom vectors o R k. If the cdf of Z is cotiuous, the we ca defie covergece

More information

Advanced Analysis. Min Yan Department of Mathematics Hong Kong University of Science and Technology

Advanced Analysis. Min Yan Department of Mathematics Hong Kong University of Science and Technology Advaced Aalysis Mi Ya Departmet of Mathematics Hog Kog Uiversity of Sciece ad Techology September 3, 009 Cotets Limit ad Cotiuity 7 Limit of Sequece 8 Defiitio 8 Property 3 3 Ifiity ad Ifiitesimal 8 4

More information

Lecture Stat Maximum Likelihood Estimation

Lecture Stat Maximum Likelihood Estimation Lecture Stat 461-561 Maximum Likelihood Estimatio A.D. Jauary 2008 A.D. () Jauary 2008 1 / 63 Maximum Likelihood Estimatio Ivariace Cosistecy E ciecy Nuisace Parameters A.D. () Jauary 2008 2 / 63 Parametric

More information

First Year Quantitative Comp Exam Spring, Part I - 203A. f X (x) = 0 otherwise

First Year Quantitative Comp Exam Spring, Part I - 203A. f X (x) = 0 otherwise First Year Quatitative Comp Exam Sprig, 2012 Istructio: There are three parts. Aswer every questio i every part. Questio I-1 Part I - 203A A radom variable X is distributed with the margial desity: >

More information

The log-behavior of n p(n) and n p(n)/n

The log-behavior of n p(n) and n p(n)/n Ramauja J. 44 017, 81-99 The log-behavior of p ad p/ William Y.C. Che 1 ad Ke Y. Zheg 1 Ceter for Applied Mathematics Tiaji Uiversity Tiaji 0007, P. R. Chia Ceter for Combiatorics, LPMC Nakai Uivercity

More information

7.1 Convergence of sequences of random variables

7.1 Convergence of sequences of random variables Chapter 7 Limit theorems Throughout this sectio we will assume a probability space (Ω, F, P), i which is defied a ifiite sequece of radom variables (X ) ad a radom variable X. The fact that for every ifiite

More information

Asymptotic Results for the Linear Regression Model

Asymptotic Results for the Linear Regression Model Asymptotic Results for the Liear Regressio Model C. Fli November 29, 2000 1. Asymptotic Results uder Classical Assumptios The followig results apply to the liear regressio model y = Xβ + ε, where X is

More information