Article ID: | iaor1988767 |
Country: | Israel |
Volume: | 25 |
Issue: | 2 |
Start Page Number: | 302 |
End Page Number: | 312 |
Publication Date: | Jun 1988 |
Journal: | Journal of Applied Probability |
Authors: | Cipra Tomas |
Vector autoregresive processes of the first order are considered which are non-negative and optimize a linear objective function. These processes may be used in stochastic linear programming with a dynamic structure. By using Tweedie’s results from the theory of Markov chains, conditions for geometric rates of convergence to stationarity (i.e. so-called geometric ergodicity) and for existence and geometric convergence of moments of these processes are obtained.