The purpose here is to make explicit the sense in which two dynamic processes, due to Malinvaud and others (whose solutions determine an efficient allocation for a given economy), are related to the gradient projection method known in the nonlinear optimization literature. The connections we establish derive from simple observations on first order characterizations of efficient allocations; they also lead to the formulation of another process, that applies to a classical welfare maximization problem; finally they provide a common basis for an a priori justification of each of the three processes involved, which supplements the intrinsic properties that they can be shown to have.