A quasi likelihood for integral data on birth and death on flows (Q1343588): Difference between revisions

The stochastic process \(M_ t\), \(t \geq 0\), taking values in the space of point measures on \(R^ d\) and describing the evolution of the countable particle system on the Brownian flow of homeomorphisms is considered. The statistical model is defined by the quadruple \((b^ \theta, c^ \theta, \pi^ \theta, k^ \theta)\), \(\theta \in \Theta \subseteq R^ m\), where \(b^ \theta\) and \(c^ \theta\) are the infinitesimal mean and covariance of the one-point motions, respectively, \(\pi^ \theta\) being the intensity measure of the births and \(k^ \theta\) is the killing rate function. For the estimation of the unknown parameter \(\theta\) observing the process \((M_ tf_ 1, \dots, M_ tf_ n)\), \(0 \leq t \leq T\), where \[ M_ t f_ j = \int_{R^ d} f_ j(x) M_ t(dx), \quad j = 1, \dots, n, \] and \(f_ 1, \dots, f_ n\) are fixed test functions, the quasi-score function is found in terms of the filtered local characteristics. The related nonlinear filtering problem is not discussed in detail.