1 Introduction to Stochastic Models GSLM 54100
2 Outline limiting distribution connectivity types of states and of irreducible DTMCs transient, recurrent, positive recurrent, null recurrent periodicity limiting behavior of irreducible chains
3 Connectivity
4 Connectivity of a DTMC connectivity: one factor that determines the limiting behavior of a DTMC A B 0.01
5 Connectivity of a DTMC A
6 Connectivity of a DTMC B 0.01
7 Connectivity of a DTMC B A
8 Connectivity of a DTMC C
9 Connectivity of a DTMC A D 0.2
10 Connectivity of a DTMC rows of P ( ) may not be the same as in the previous examples
11 Connectivity of a DTMC F 1 limit of P (m) may not exist
12 Connectivity of a DTMC
13 Types of States and of Irreducible DTMCs
14 Limiting Results for a DTMC depending on the type of states and the chain type: transient, positive recurrent, null recurrent connectivity and periodicity
15 Transient State A
16 Recurrent State state i is recurrent if P(return to i|X 0 = i) = 1
17 Recurrent State two types of recurrent states positive recurrent: E(# of transitions to return to i|X 0 = i) < null recurrent: E(# of transitions to return to i|X 0 = i) =
18 Periodicity
19 Periodicity state i is of period d if X n can return to state i in multiples of d states 1, 2, 3 are of period 2 state i of period d
20 Periodicity period of states 1, 2, 3, and 4 = ? state 4 of period 2 state 4 can return to itself in 2 steps
21 Communicating States communicating states are of the same type transient, positive recurrent, null recurrent at the same time of the same period states in an irreducible chain are of the same type transient, positive recurrent, null recurrent at the same time of the same period
22 Limiting Behavior of Irreducible Chains
23 Limiting Behavior of a Positive Irreducible Chain j = fraction of time at state j N: a very large positive integer # of periods at state j j N balance of flow j N i ( i N)p ij j = i i p ij
24 Limiting Behavior of a Positive Irreducible Chain j = fraction of time at state j j = i i p ij 1 = 0.9 2 2 = 0.1 2 linearly dependent normalization equation: 1 + 2 = 1 solving: 1 = 2/3, 2 = 1/ C 0.2
25 Limiting Behavior of a Positive Irreducible Chain 1 = 0.75 3 3 = 0.25 2 1 + 2 + 3 = 1 1 = 301/801, 2 = 400/801, 3 = 100/
26 Limiting Behavior of a Positive Irreducible Chain an irreducible DTMC {X n } is positive there exists a unique nonnegative solution to j : stationary (steady-state) distribution of {X n }
27 Limiting Behavior of a Positive Irreducible Chain j = fraction of time at state j j = fraction of expected time at state j average cost c j for each visit at state j random i.i.d. C j for each visit at state j for aperiodic chain:
28 Limiting Behavior of a Positive Irreducible Chain 1 = 301/801, 2 = 400/801, 3 = 100/801 profit per state: c 1 = 4, c 2 = 8, c 3 = -2 average profit
29 Limiting Behavior of a Positive Irreducible Chain 1 = 301/801, 2 = 400/801, 3 = 100/801 C 1 ~ unif[0, 8], C 2 ~ Geo(1/8), C 3 = -4 w.p. 0.5; and = 0 w.p. 0.5 E(C 1 ) = 4, E(C 2 ) = 8, E(C 3 ) = -2 average profit
30 Different Interpretations of balance equations: balance of rates total rate into a group of states = total rate out of a group of states {0}: 0 = q 1 {0, 1}: p 1 = q 2 {0, 1, …, n}: p n = q n+1, n q 2 3 p q p q p 1 … q
31 Example: Condition for the Following Chain to be Positive {0}: 0 = q 1 {0, 1}: p 1 = q 2 1 {0, 1, …, n}: p n = q n+1, n 1 1 positive { j } exists 0 + 1 + 2 + … = 1 has solution .. p < q 0 1 q 2 3 p q p q p 1 … q
32 Example 4.24 of Ross four-state production process states {1, 2, 3, 4} up states {3, 4}, down states {1, 2} find E(up time) & E(down time) time 1 state time down state up
33 Example 4.24 of Ross 1 = 3/16, 2 = 1/4, 3 = 7/24, 4 = 13/48 how to find E(up time) E(down time)
34 Example 4.24 of Ross fraction of up time = 3 + 4 = rate of turning from up to down = (p 31 +p 32 ) 3 + (p 41 +p 42 ) 4 = rate of turning from down to up = p 13 1 + (p 23 +p 24 ) 2 =