Notation in the stochastic derivatives in the mean square sense

693 Views Asked by At

The stochastic limit $X$ in the mean square sense is given the definition: For a row (sequence?) of stochastic variables $X_n$ if $\displaystyle\lim_{n\to\infty}E\{(X_n-X)^2\}$ = 0 and we write $\text{l.i.m.} X_n = X$. This I understand as the limit of a sequence argument akin to usual analysis.

When we talk about the derivative of a stochastic process $X_t^\prime$, we define

$X_t' = \frac{dX_t}{dt} = \text{l.i.m} \frac{X_{t+h} - X_t}{h}$

where $X_t, t\geq 0$ is defined to be stochastic process. When I see the $\text{l.i.m}$ operator, I expect to see a row (sequence?) of stochastic processes. However, in the reference I have, only the index $t$ is written.

Shouldn't an $'n'$ be also written to depict that the derivative is the square mean limit of a row of stochastic processes? Or have I got the idea wrong that the limit is in the sequence index $n$ and the derivative is defined for the stochastic process index $t$? To me, the following seems the clearer definition of the derivative

$X_t' = \frac{dX_t}{dt} = \text{l.i.m} \frac{X_{n,t+h} - X_{n,t}}{h}$

which is sort of a pointwise argument for the index $t$ of the stochastic process. I guess my confusion is mainly from the notation used to express the two concepts where one index seems to be omitted for the clarity sake.

1

There are 1 best solutions below

4
On

It is like the notation for limit of a sequence and that in continuous time. Specifically, the derivative is defined by a process $\{X_t', t \ge 0\}$ that satisfies \begin{align*} \lim_{h\rightarrow0}E\left(\left(\frac{X_{t+h}-X_t}{h} -X_t'\right)^2 \right) = 0. \end{align*}