Shigeyoshi Watanabe

A Quantitative Method for Comparing Multi-Agent-Based Simulations in Feature Space

Comparisons of simulation results (model-to-model approach) are important for examining the validity of simulation models. One of the factors preventing the widespread application of this approach is the lack of methods for comparing multi-agent-based simulation results. In order to expand the application area of the model-to-model approach, this paper introduces a quantitative method for comparing multi-agent-based simulation models that have the following properties: (1) time series data is regarded as a simulation result and (2) simulation results are different each time the model is used due to the effect of randomness, even though the parameter setups are all the same. To evaluate the effectiveness of the proposed method, we used it for the comparison of artificial stock market simulations using two different learning algorithms. We concluded that our method is useful for (1) investigating the difference in the trends of simulation results obtained from models using different learning algorithms; and (2) identifying reliable simulation results that are minimally influenced by the learning algorithms used.

Individual Level Analysis Using Decision Making Features in Multiagent Based Simulation

We introduce a set of evaluation tools in the framework of individual level analysis for multiagent based simulations. These tools are intended to overcome the weak points of multiagent systems: that it is difficult to avoid mistakes in description and arbitrary modeling; and to find reliable explanations of causal relationships between the model and its result. We analyze some artificial market models using an evaluation process that consists of the evaluation of initial learning maturity, discrimination between models, and the visualization of attention tendency in a group. We conclude that these analytical methods are useful for the evaluation of multiagent based simulations in terms of validation and finding causal relations.

Simulation analysis of the participant nations behavior in the emissions trading

This paper proposes a decision-making model of the participant nations using the reinforcement learning under uncertainty in the international emissions trading market, aim at clarifying the success conditions of the compliance mechanism in the emissions trading market. Intensive simulations have been revealed the following implications: (1) the trade of the emissions trading tended to stop that all nations set reduction commitment than 7%, and (2) it is important to give high incentive for the improvement of the participant nations behavior.

Bayesian Analysis Method of Time Series Data in Greenhouse Gas Emissions Trading Market

This paper proposes the Bayesian analysis method (BAM) to classify the time series data which derives the complicated phenomena in the international greenhouse gas emissions trading. Our investigation compared the results using the method of Discrete Fourier transform (DFT) and BAM. Such comparisons have revealed the following implications: (1) BAM is superior to DFT in terms of classifying time series data by the different distances; and (2) the different distances in BAM show the importance of 1% influence of emission reduction targets.

Proposal of stacked type Fe-FET reconfigurable logic circuit featured with parallel processing within one silicon pillar using modified process technology of 3D NAND flash memory

In order to realize the parallel processing within one silicon pillar, a novel stacked type Fe-FET reconfigurable logic circuit featured with parallel processing within one silicon pillar using modified process technology of 3D NAND flash memory has been newly proposed. For realizing the proposed scheme plural numbers of logics are generated within one silicon pillar. Furthermore, the outputs of these logics are connected to horizontally running wires and these wires are connected to vertically running wires which act as the output signals. For realizing this structure two improvements for fabrication scheme of 3D NAND flash memory should be introduced. First is fabrication process featured by various depth trench formation technology with one fabrication step. Second is connection technology between the output of logic block generated within the silicon pillar to horizontal running wires.

Impact of FinFET with plural number of channel width using novel one step of trench formation process on fabrication cost for system LSI

Impact of FinFET with plural number of channel width using novel one step of trench formation process on fabrication cost for system LSI has been described. Novel one step of trench formation process are realized by using trench etching process for various sizes of trench openings and dummy pattern between adjacent trenches. With these technologies fabrication cost of cell library for system LSI can be reduced to 11-15% compared with that of conventional planar transistor process. Proposed technologies are promising candidate for realizing future high density system LSI design with cell library.

Analysis of emission right prices in greenhouse gas emission trading via agent-based model

This paper proposes a participant nation model for international emission trading; adaptive agents are used to explore the conditions under which an emission trading market is successful. In this study, the participation nation models with and without the compliance mechanism as prescribed in the Kyoto Protocol are compared; the simulation results for these two cases show a significant difference in both the market price and the amount of gas emissions. Intensive simulations revealed the following successful conditions for the compliance mechanism: 1 the different carbon emission reduction targets for participant nations, as prescribed in the Kyoto Protocol, contributes to reducing the emission amount, 2 there exists a critical boundary for success i.e., less than -6% emission reduction target when all participant nations set negative emission reduction targets, 3 a few developing countries with positive emission reduction targets are indispensable for maintaining market trading, and 4 the use of the compliance mechanism helps reduce emissions, while in the case of noncompliance, reduction of emission is not achieved.

MASSE: environment supporting for simulation and analysis of multiagent systems

In multiagent system research, the work in building simulation model and the effort in developing analysis methods are closely related because building multiagent models relies heavily on new effective analysis methods while justifying new analysis methods needs the simulation results of multiagent models. MASSE (Multiagent Simulation Systematic Explorer) was proposed as an integrated environment which is aimed at (a) to conduct efficient simulation by intelligent scheduling, (b) to conduct intelligent data analysis and knowledge discovery on simulation data, and (c) to develop such analysis methods themselves. Current implementation of MASSE is described in this manuscript, and its usefulness is shown in three aspects: simple accommodation of existing simulators, satisfactory performance for adopting grid computing technique, and systematic data analysis tool. Collaboration among simulation developers and analysts can be strongly supported by using MASSE to incorporate various simulators and unified method for data analysis. We conclude that the current implementation of MASSE is satisfactory