一种基于多维尾条件期望的扩散模型检验方法

A Test for Diffusion Model Based on Multi-Dimensional Tail Condition Expectations

为了解决多维扩散模型的设定检验问题,我们基于多维尾条件期望(CTE)制定了一个检验统计量.虽然不能直接估计出多维扩散模型的转移密度矩阵,但是可以估计出每个分量的转移密度,再通过CTE将每个分量联合起来,建立了一个真正的多维统计量.最后通过仿真模拟评估了该检验方法的检验的性能.

In order to solve the problem of testing multidimensional diffusion models,we develop a test statistic based on Multi-Dimensional Tail Condition Expectations(CTEs).Although it is almost impossible to estimate the transition density matrix of a multidimensional diffusion model directly,the transition density of each component can be estimated and each component can be combined by the CTE to establish a true multidimensional statistics.Finally,the performance of the test is evaluated through simulation.