An absorbing state is a state of a Markov chain that cannot be left. For a time-independent stochastic
matrix
under the row convention, state
is absorbing iff
. The other entries in row
are then 0 (Grinstead and Snell 1997, p. 416).
The presence of an absorbing state does not require other states to reach it. That additional reachability condition is part of the definition of an absorbing Markov chain.