The hippocampus is responsible for the formation of episodic memory. Recently, it was observed that atrophy of the hippocampus can lead to Alzheimer's disease. Thus, it would be important to investigate the mechanism of the formation of episodic memory in terms of the neural network model of the hippocampus. In the present study, we investigated a coding scheme of spatio-temporal inputs to CA 1 of the hippocampus. We found that Cantor sets produced in the space of membrane potentials of CA 1 neurons represent input time series in a hierarchical way. A similar behavior has been observed in experiments with rat hippocampal slices.

View full abstract

This paper summarizes current activities of numerical simulation technology applications to the power and social infrastructure systems related research and development. It is expected that the simulation-driven R & D will be achieved for the development of the power system equipments through the advances of the computer power and the simulation technology. A simulation-driven R & D process is discussed in the paper, which consists of micro-scale and/or meso-scale simulations for estimating basic engineering constants, and macro-scale simulations for determining detailed design of the power systems equipments with using the outputs of the micro-scale and the meso-scale simulations. Examples of these meso-scale and macro-scale simulations are presented. As an example of the application to the R & D for social infrastructure systems, a traffic simulator based on molecular dynamic simulation technology is presented, of which the objective is to evaluate the function of intelligent transport systems (ITS). It is considered that the role of simulation technology will expand in these engineering fields.

View full abstract

Recently "climate change" such as global warming is big issue scientifically and socially. I describe a review of statistical analysis for long term trend in meteorological data. Student's t test is most popular to test statistical significance of estimated trend. However it could not be applied to non-normal variations. By Edgeworth expansion we can obtain more accurate distribution function than the normal distribution. Nonparametric methods such as Mann-Kendall test and Sen's slope are becoming popular in such non-normal cases. Detection of trend in extreme events is often difficult. I developed a new method to test existence of trend in frequency of occurrence of extreme events.

View full abstract

[in Japanese]

View full abstract