An approach to reasoning in which an inference engine endeavors to find a value for an overall goal by recursively finding values for subgoals. At any point in the recursion, the effort of finding a value for the immediate goal involves examining rule conclusions to identify those rules that could possibly establish a value for that goal. An unknown variable in the premise of one of these candidate rules becomes a new subgoal for recursion purposes.
See
▶ Expert Systems
Backward Kolmogorov EquationsIn a continuous-time Markov chain with state X(t) at time t, define p ij (t) as the probability that X(t + s) ¼ j, given that X(s) ¼ i, s, t ! 0, and r ij as the transition rate out of state i to state j. Then Kolmogorov's backward equations say that, for all states i, j and times t ! 0, the derivatives dp ij (t)/dt ¼ P k6 ¼i r ik p kj (t) À v i p ij (t), where v i is the transition rate out of state i, v i ¼ P j r ij .
See
▶ Markov Chains ▶ Markov Processes
Backward-Recurrence TimeSuppose events occur at times T 1 , T 2 , . . . such that the interevent times T k À T kÀ1 are mutually independent, positive random variables with a common cumulative distribution function. Choose an arbitrary time t.The backward recurrence time at t is the elapsed time since the most recent occurrence of an event prior to t.
Balance Equations(1) In probability modeling, steady-state systems of equations for the state probabilities of a stochastic process found by equating transition rates. For Markov chains, such equations can be derived from the Kolmogorov differential equations or from the fact that the flow rate into a system state or level must equal the rate out of that state or level for steady state to be achieved. (2) In linear programming (usually referring to a production process model), constraints that express the equality of inflows and outflows of material.
See
IntroductionOR/MS techniques find applications in numerous and diverse areas of operation in a banking institution. Applications include the use of data-driven models to measure the operating efficiency of bank branches through data envelopment analysis, the use of image recognition techniques for check processing, the use of artificial neural networks for evaluating loan applications, and the use of facility location theory for opening new branches and placing automatic teller machines (e.g., Harker and Zenios 1999). A primary area of application is that of financial risk control in developing broad asset/liability management strategies. Papers that summarize these areas are Zenios (1993), Jarrow et al. (1994), and Ziemba and Mulvey (1998). This work can be classified into three categories: (1) pricing contingent cashflows, (2) portfolio immunization, and (3) portfolio diversification.
Pricing Contingent CashflowsThe fundamental pricing equation computes the price of a contingent cashflow as the expected net present value of the cashflows, discounted by an appropriate discount rate. In discrete time the pricing equation takes the formwhere E denotes expectation over...