Scientific information in continuous characteristics spaces

摘要

Scientific information is characterized as the flow of information created by scientists, which is evaluated by peer reviewers. A final evaluation is connected with the use of the newly generated information either by other scientists, engineers or the general scientific community. This evaluation can be estimated by measures such as the relative number of citations in the scientific literature and me relative number of citations in patents. We model the dynamics of the evolution of scientific knowledge as hill-climbing in an adaptive landscape over a continuous characteristics space. A problem is described by a large number of attributes, features or characteristics covering problem-inherent aspects and disciplinary parameters. These quantities span a characteristics space, which is a real Euclidean vector space, analogous to the phenotype space in biology. We also define a population density x(q) that describes the density of the newly generated information at time t in the point q and a real-valued multimodal fitness function/functional V(x( q ,t)) which expresses the evaluation in a qualitative way. The evolutionary dynamics including competition and mutations/innovations is modelled by reaction-diffusion equations of Fisher-Eigen type.