前 言(译) V种推广。这些具体的估计方程比以往需要大量计算的似然方程更容易计算和求解。它的思想是去逼近似然方程,而且在某些情况下,估计函数可以提供完全有效的估计。作为一种特殊情形,第 1 章还讨论了极大似然估计。
第 2 章由 Per Mykland 和 Lan Zhang 撰写。讨论了金融资产价格中高频数据的建模问题。考虑的模型被假设为一个带有所谓微结构噪声的误差的半鞅。微结构噪声对于估计的影响可能比模型参数对于估计的影响还大,因此会造成估计上的困难。这里,利用多尺度已实现波动,给出了一个克服这些困难的办法。
第 3 章由 Jean Jacod 撰写,考虑了带有一般性跳跃点的基于高频数据的扩散过程的推断问题。这意味着在 0 到 T 的时间间隔内以等距的时间节点观测随机过程,其中相邻的两个观测时间节点对应的区间很小,且趋于 0。这样的模型有很多应用,特别是在金融领域中,常常对估计整合波动率感兴趣。主要基于二次变分的变体,本章给出了很多对于这些模型的估计方法,也阐明了相应的极限理论。
第 4 章由 Omiros Papaspiliopoulos 和 Gareth Roberts 撰写,集中考虑了实现扩散模型的基于相似度的推断的计算方法。在详细讲述了扩散的各种模拟方法之后,本章给出了一个确切的特别强调条件扩散模拟的模拟方法。不同于使用欧拉逼近格式,该方法精确地模拟了条件扩散的路径,而不带有任何离散化误差。与蒙特卡罗方法相结合,该方法有效地计算了过程的极大似然估计和贝叶斯估计。
第 5 章由 Fabienne Comte、Valentine Genon-Catalot 和 Yves Rozenholc撰写,提供了随机微分方程模型的几个非参数估计方法,考虑了相应的收敛速度,还通过几个例子来解释所列方法的效果。
第 6 章由 Peter Brockwell 和 Alexander Lindner 撰写,讨论了一些最新的随机波动模型,其中的驱动过程是带有跳跃点的 Lévy 过程。本章在列出了这种模型的出发点和性质之后,描述了一些估计方法。
最后,第 7 章由 Grigorios Pavliotis、Yvo Pokern 和 Andrew Stuart撰写,处理了数据中所表现的多尺度特征的建模问题,描述了可以用来找到一个有用的扩散逼近的方法,给出了物理上和分子动力学上的一些例子。
PrefaceThe chapters of this volume represent the revised versions of the main papersgiven at the seventh S′eminaire Europ′een de Statistique on “Statistics forStochastic Differential Equations Models,” held at La Manga del Mar Menor,Cartagena, Spain, May 7th–12th, 2007. The aim of the S ′eminaire Europ ′eende Statistique is to provide talented young researchers with an opportunity toget quickly to the forefront of knowledge and research in areas of statisticalscience which are of major current interest. As a consequence, this volume istutorial, following the tradition of the books based on the previous seminars inthe series entitled:.NetworksandChaos–StatisticalandProbabilisticAspects.TimeSeriesModelsinEconometrics,FinanceandOtherFields.StochasticGeometry:LikelihoodandComputation.ComplexStochasticSystems.ExtremeValuesinFinance,TelecommunicationsandtheEnvironment.StatisticsofSpatio-TemporalSystemsAbout 40 young scientists from 15 different nationalities mainly from Europeancountries participated. More than half presented their recent work in shortcommunications; an additional poster session was organized, all contributionsbeing of high quality.The importance of stochastic differential equations as the modeling basis forphenomena ranging from finance to neurosciences has increased dramaticallyin recent years. Effective and well behaved statistical methods for these modelsare therefore of great interest. However, the mathematical complexity ofthe involved objects raises theoretical but also computational challenges. TheS′eminaire and the present book present recent developments that address, onone hand, properties of the statistical structure of the corresponding modelsand, on the other hand, relevant implementation issues, thus providing a valuableand updated overview of the field.The first chapter of the book, written byMichael S.rensen, describes the applicationof estimating functions to diffusion-type models. Estimating functions原书前言PrefaceThe chapters of this volume represent the revised versions of the main papersgiven at the seventh S′eminaire Europ′een de Statistique on “Statistics forStochastic Differential Equations Models,” held at La Manga del Mar Menor,Cartagena, Spain, May 7th–12th, 2007. The aim of the S ′eminaire Europ ′eende Statistique