The current density functional calculations adopt numerical integration to estimate exchange-correlation (XC) term, which brings numerical instability in case of huge molecule and massively multipoint calculations. The grid-free method developed in 1993 was capable of calculating XC term analytically and conquering brittleness but could not have given reliable computational values so far. In this paper, we clarified that the cause of this problem was in the basis set used to calculate the matrix representation of density (MRD). From this knowledge, we proposed the new grid-free method which greatly improved the computational values, where rich basis sets were applied to MRD basis. In addition, we proposed an application of Cholesky decomposition method to the grid-free method in order to reduce the computational cost safely and effectively.
In this research, a micro-fluidic device with micro heaters is fabricated to investigate cell-cell interactions under local heating stimulation. This report presents a thermal analysis based on finite element modeling, design and fabrication of a platinum micro heater device. Moreover, micro resistance thermometers, which can measure heating temperature in real time, are integrated as well. Calibration and measurement of Pt micro heater device are also presented.
To improve radiation therapy, the DNA degradation mechanisms under radiation should be studied. In this research we use silicon nanotweezers, one kind of MEMS devices for direct manipulation of DNA molecules, to detect the radiation-caused degradation. By measuring resonant frequency of tweezers capturing a DNA bundle under radiation, we monitored the damage-related stiffness change of DNA in real-time. Besides, we also performed the measurement in solution coupled with microfluidic device. We hope this research will paves the way for both fundamental and clinical studies of DNA degradation mechanisms under radiation beams for improving radiation therapy and other tumor treatment.
The blood flow information from the medical imaging data is limited due to the temporal and spatial resolution. The computational fluid dynamics (CFD) combined with the medical imaging data is considered to be a powerful tool to predict the details of blood flow for the patient-specific diagnosis. In our research, the 1D-0D multi-scale model for the cardiovascular system is developed to estimate distributions and wave forms of blood flow in the Circle of Willis (CoW). In order to achieve more realistic prediction, the patient-specific information is employed in the simulation such that the vascular geometry of CoW is extracted from the Magnetic Resonance Imaging (MRI) data and the flow information is given by the Single Photon Emission Computed Tomography (SPECT) measurement. The applicability of the present method is evaluated by comparing the simulation results to PC-MRI measurement data.
We studied the dynamics in a sparse coding-based associative memory network with dynamic synapses. The network exhibits oscillation among sparsely coded memory patterns in the depression-dominated region, while the mean activity in the network is retained at a relatively low level. The mean field model is derived using the sublattice method. We investigated the influence of the sparse coding scheme on the properties of the network dynamics. The oscillation occurs with reduced noise intensity, and the parameter region of multi-stable states on the border of the oscillation region shrinks with the increasing sparseness.
Dynamical noise in electroencephalographic (EEG) dynamical systems can be useful information to understand the dynamical behavior of neural networks. In this study, we analyzed temporal changes of the levels of dynamical noise in EEG dynamical systems. EEG data from an electrode over the visual cortex were converted into temporal changes of the levels of dynamical noise. In addition, the converted time series was quantified by the Hurst exponent. As a result, the Hurst exponent in a condition of eyes-open decreased with time. On the other hand, the Hurst exponent in a condition of eyes-closed showed an almost constant value. Therefore, we suggest that temporal changes of neural networks on the eyes-open state are on a transient state and these on the eyes-closed state are on a steady state.
Reinforcement learning, a research field of machine learning for optimizing through trial-and-error, is often considered as a narrowly-aimed technique for autonomous agents such as robots. Recently, reinforcement learning are getting applied to many practical problems in wide fields as an optimization technique for unknown environment including data collection schedule. In this review, we introduce some of these recent applications to overview a possible future areas suitable to reinforcement learning.
Recently, electricity markets have been organized in many countries. Due to the complex structure, the mechanisms of markets have been studied mathematically from various points of view. Meanwhile, many countries promote use of renewable energy sources for electricity generation as alternative to conventional exhaustible energy sources. In the present study, we analyze differences between three types of auction rules (uniform price auction, pay-as-bid auction and Vickrey auction) by numerical simulation of an agent-based market model in which renewable energy is considered. On the basis of the data of Japanese power generators, we analyzed the electricity price and effects of introducing renewable power plants into electricity market under each auction rule.
We introduce a novel in-situ underwater AFM system, which is mountable on underwater vehicles. The mission for the system is to visualize the nature of micro samples in deep sea with high-spatial-resolution down to nanometer scale. The system is composed of an underwater AFM, microfluidic devices, and the mount mechanisms for underwater vehicles. We have developed a novel sample stage, which equipped with sample filtration mechanisms using a membrane filter. The forthcoming field exploration using the in-situ underwater AFM system would reveal new findings of deep sea.