Let N=(N_s^l,…, N_5^d)be a multivariate counting process with intensities of the form λ(α)=(λ_s^lα_a^l,…,λ_s^dα_S^d). We consider the model where some of the α's are parametrized by the Euclidean parameter θ∈Θ and the rest of α's are r...
http://chineseinput.net/에서 pinyin(병음)방식으로 중국어를 변환할 수 있습니다.
변환된 중국어를 복사하여 사용하시면 됩니다.
https://www.riss.kr/link?id=A30042539
So, Beong-Soo (Department of Statistics, Ewha Womens University)
1992
English
335.000
학술저널
19-30(12쪽)
0
상세조회0
다운로드다국어 초록 (Multilingual Abstract)
Let N=(N_s^l,…, N_5^d)be a multivariate counting process with intensities of the form λ(α)=(λ_s^lα_a^l,…,λ_s^dα_S^d). We consider the model where some of the α's are parametrized by the Euclidean parameter θ∈Θ and the rest of α's are r...
Let N=(N_s^l,…, N_5^d)be a multivariate counting process with intensities of the form λ(α)=(λ_s^lα_a^l,…,λ_s^dα_S^d). We consider the model where some of the α's are parametrized by the Euclidean parameter θ∈Θ and the rest of α's are regarded as nuisance parameters. In this framework we will derive the asymptotic distribution of the minimum distance type test statistics : ??|??-F_θ|which is based on the procuct limit estimator(PLE)?? and show that minimum distance estimator(MDE)?? of θ defined by : ??|??-F_θ|=|??-??| is asymptotically normal and has a locally asymptotic minimax(LAM) optimality. We also propose computationally tractable bootstrap versions of these quantities and show that they have the same asymptotic properties as desired.
본 논문에서는 일반적으로 강도함수가 λ(α)=(α_iλ_i)=l,…,d인 형태로 주어지는 다변량 계수확률과정 N(t)=(N_i(t))i=l,…,d을 고려한다. 이 모형에서 α는 확률과정의 강도함수를 지정하는 미지의 함수로서 그 일부는 유한차원의 관심모수벡타를 포함하여 나머지는 미지의 부수적인 함수들로 간주된다. 본 연구는 위의 모형에서의 승법극한추정치에 근거한 최소거리 검정통계량의 점근분포를 유도하고 최소거리 추정량의 점근분포가 정규분포임을 보이고 또한 국소점근 minimax최적성을 가짐을 증명한다. 아울러 이들 검정 통계량과 측정량들의 Bootstrap 추정치를 고려하여 대표본에서 본래의 절차들과 동일한 성질을 가짐을 보인다.
다국어 초록 (Multilingual Abstract)
Let N_=(N_s^1,…,N_s^d)be a multivariate counting process with intensities of the form λ(α)=(λ_s^1c_a^1,…,λ_s^dα_s^d).We consider the model where some of the α's are parametrized by the Euclidean parameter θ∈Θ and the rest of α's are reg...
Let N_=(N_s^1,…,N_s^d)be a multivariate counting process with intensities of the form λ(α)=(λ_s^1c_a^1,…,λ_s^dα_s^d).We consider the model where some of the α's are parametrized by the Euclidean parameter θ∈Θ and the rest of α's are regarded as nuisance parameters. In this framework we will derive the asymptotic distribution of the minimum distance type test statistics : inf|F△_n-F_ θ|which is based on the product limit estimator(PLE) F△_n and show that minimum distance estimator(MIDE) θ^△_n of θ definde by:inf|F△_n-F_ θ|=|F△_n-F_ θ^|is asymptotically normal and has a locally asymptotic minimax(LAM) optimality. We also propose computationally tractable bootstrap versions of these quantities and show that they have the same asymptotic properties as desired.
A Note on Duality of Drinfel Modules