Consider a continuous time Markov chain $(X_t)_{t \geq 0}$ on some state space $S$ with transition matrix (P-matrix) $p_t(x,y)$, the probability density of jumping from $x$ to $y$ in time $t$.

The associated Q-matrix is defined as

$$q(x,y) = \frac{d}{d t}p_t(x,y)\vert_{t=0}$$

and offers an infinitesimal description of the chain.

What are advantages and disadvantages of investigating the Markov with either the Q- or the P-matrix?

Are the two descriptions any different in the case of $S$ being finite or do the methods only differ for $S$ being countably infinite?


Those are two equivalent descriptions of the same semi-group. You mentioned how to deduce $q$ from $(p_t)$, note that, conversely, $p_t=\mathrm e^{tq}$ for every $t\geqslant0$.

  • $\begingroup$ Do there have to be assumptions on $Q$ for the exponential series to converge? (which would imply that there is no one-to-one correspondence) Is one description favorable over the other from a practical point of view when working with Markov chains? $\endgroup$ – madison54 Jul 20 '14 at 10:06
  • $\begingroup$ The exponential series always converges, for the same reason that in the scalar case, plus the submultiplicativity of the norm. "More or less favorable": this all depends on what one wants to do with these. $\endgroup$ – Did Jul 20 '14 at 10:11

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.