Research Overview
Key features of the big data era are increase in the volume of data and necessity to handle completely different types of data. Establishing techniques for both simulation and big data analysis is the long-cherished dream of every scientific field, and the effect of their spread will reach every field that uses simulation.
This project aims to develop and spread methods for integrating simulation and data analysis. To this end, we will use collaborative research, aiming to solve specific problems in geoscience and life science, and to expand data assimilation methods in various fields. (Project director: Junji Nakano (The Institute of Statistical Mathematics).
Project Objective
In geoscience field, we aim to recreate the aurora by using a simulation model with observation data in the earth’s magnetosphere. We also use the data from the Program of the Antarctic Syowa MST/IS Radar (PANSY) to attack problems of high speed, high resolution, multidimensional data obtained in real time at the observation post. In life sciences, we apply data assimilation methods to problems in developmental cell biology (such as segmentation and sperm formation in mammals, animal cell division, and cytoplasmic streaming), with the goal of establishing it as a powerful method in the field.
Through these collaborative research activities, we aim to nurture a new type of researcher who can solve problems not only with a forward problem solving approach, but also with an understanding of how to work backward from data to theory in a truly inverse problem solving approach.
Project Framework
The project is organized into the following three teams, which collaborate with one another as they perform their research.
Mathematical Theory and Computing Team
This team’s main research objective is to develop new data assimilation techniques and to apply them to new fields of science. They are developing a hybrid data assimilation algorithm that uses an optimal sequence of Bayesian filtering aiming to apply data assimilation to a hierarchical simulation with a realistic computation time. There are also developing algorithms for successive assimilation of large-scale real-time data such as sensor data or traffic in a social network, as well as a program to enable full utilization of high performance computing.
Modeling Team
Using the distribution of electric potential in the ionosphere, obtained from the Super Dual Aurora Radar Network (SuperDARN) of the National Institute of Polar Research for data assimilation, the team confirms the changes in the ionospheric electric potential distribution and magnetospheric convection with respect to the Global magnetohydrodynamic (MHD) simulation model boundary parameters. The team studies the similarity between the model and the observed data for the changes in boundary parameters, not by comparing them with absolute values of local physical quantities, but by comparing them with shape patterns for the convection cells observed in the ionospheric electrical potential distribution. Through data assimilation experiments, the team is estimating the optimal values for the model parameters of the internal boundary of the simulation (magnetosphere-ionosphere boundary). For the atmospheric radar, the team is using both the Middle and Upper Atmospheric (MU) Radar (Shiga Prefecture) that has been used in observational experiments to date and the Program of the Antarctic Syowa MST/IS Radar (PANSY) to construct a general theory and method incorporating the special factors of each radar’s observation target and system. The team will also work to construct an optimal method for estimating physical quantities, including by fusing estimation algorithms from multichannel signal processing and smoothers.
Data Design Team
The team applies data assimilation and related statistical methodologies to biological questions. Experimental and theoretical biologists in National Institute of Genetics (NIG) collaborate to model the formation of somites during mouse development. Experimental biologists in NIG collaborate with statisticians in Institute of Statistical Mathematics (ISM) to develop new analysis methods to invesitigate the regulatory mechanisms of germ cells using RNA molecules. For the research on cytoplasmic streaming in the nematode C. elegans, biologists in NIG utilize the latest computing environment and other resources in ISM.
Research View 021
Supporting Structures for Building Great Teams
[Data assimilation and simulation support technologies] Tomoyuki Higuchi (Director-General, The Institute of Statistical Mathematics)
Research View 010
Understanding the cell through the viewpoint of architectonics
[Data assimilation and simulation support technologies] Akatsuki Kimura (Associate Professor, National Institute of Genetics)
Research View 008
Where do auroras fall?
[Data assimilation and simulation support technologies] Satoko Saita (Specially appointed researcher at the Institute of Statistical Mathematics)
Research View 007
How Will Infectious Diseases Spread in Cities?
[Data assimilation and simulation support technologies] Masaya Saito (Special Assistant Professor, Institute of Statistical Mathematics)
Research View 006
Realistic Modeling by using Data
[Data assimilation and simulation support technologies] Junji Nakano (Professor, The Institute of Statistical Mathematics)