Sequential parameter estimation with guaranteed mean-square accuracy for unstable linear stochastic systems (Q1310722)

Let \(X(t) = (X_ 1(t),\dots,X_ p(t))'\) be a \(p\)-dimensional stochastic process given by a stochastic differential equation \(d X(t) = AX(t)dt + Bd\varepsilon(t)\), where \(\varepsilon(t)\) denotes \(p\)-dimensional standard Wiener process, \(A\) and \(B\) are matrices of the form \[ A = \left \|{{a_ 1,\dots,a_ p}\atop {I_{p-1},0}}\right\|,\quad B = \left\|{{1,\dots,0}\atop {0,\dots,0}}\right\|, \] and \(I_ p\) stands for the \(p\)-dimensional identity matrix. The authors present sequential estimation, with a given mean-square accuracy, of the unknown parameters \((a_ 1,a_ 2,\dots,a_ p)'\).