There are of course many different ways of generating probabilities. In many classical statistical experiments we can assume all outcomes are equally likely to occur. This chapter introduces different ways to produce the required probability distribution. Generally the probability of each of measured outcome would depend on the state of the system as well as that particular observable in question. A simple counting of number of outcomes without reference to the state cannot possibly lead to the correct probability distribution. The theory presented can be extended to lower as well as higher dimensional real vector spaces. For a model theory to have any chance of success it is necessary for the model to satisfy the following additional requirements: There must be enough unit vectors and selfadjoint operators to correspond to all the states and all the observables of the system; and the probability distribution for every observable in every state must agree with experimental results.
Many selfadjoint operators important to physical applications, e.g., number operators and some first order and second order differential operators, possess a discrete spectrum. For a selfadjoint operator with a discrete spectrum the spectral function reduces to a sum in the form. An arbitrary symmetric operator does not possess a spectral function and the spectral theorem does not apply to such an operator. Hence the relation between selfadjoint operators and probability distributions presented above does not apply to symmetric operators. This is the reason that the probabilistic nature of quantum mechanics is formulated in terms of selfadjoint operators, not symmetric operators. In the momentum representation the momentum appears as a multiplication operator in the same way that the position acts as a multiplication operator in the coordinate representation. It follows that we can treat the momentum in the momentum representation in the same way as position in the coordinate representation.