Information theory for experiments with known conditional probabilities: Application to the prediction of branching ratios

Information theory is used, under conditions of incomplete data, to provide a maximally conservative prior expectation of a population distribution. This is done by systematically refining the estimates of the full distribution via the incorporation of fragmentary data in the form of conditional probabilities. The net effect is to reduce the available phase space volume of the distribution as more data are incorporated. The above refinements are then applied to calculate estimates of branching ratios for chemical reactions with more than one possible product. Two new branching ratio expressions are derived: one for the case where each product's internal state distribution is known for reactions originating from various specific reactant state?, and when the joint state-to-state reactive probability is known.