Simulations Encounter with Data Science
Data Assimilation, Emulators, Rare Events and Design

March 9-11, 2017

The Institute of Statistical Mathematics

Tachikawa, Tokyo

ACCESS

Room
Seminar Room 1 / 2 (3F)

Organizers
Yukito Iba (ISM)
Hiroshi Fujisaki (Nippon Medical School)
Hiromichi Suetani (Oita University)

Aim and Scope

In this era, advances in computer hardware enable massive simulation in everyday life, and a combination of such simulations with machine learning and modern statistical methods becomes an attractive strategy to open up novel fields of science and engineering. This workshop is intended at discussing related topics and facilitating cross-disciplinary activities among researchers from various fields.

The scope of the workshop includes the following subjects:

1. Data assimilation and statistical inference on dynamical systems. They are basic methodologies for creating a bridge between simulations and the real world.
2. Construction of emulators or surrogate models. They are useful tools for saving computational time and bridging time scales in simulations of complicated systems.
3. Sampling of rare events in simulations. It enables finding the local minima, transition paths, unstable orbits, and other interesting structures in simulations. Unbiased estimation of probabilities of such structures is also important.
4. Designing interesting objects through simulations. Such a designing process can be improved by the efficient use of machine learning and modern statistical methodologies, as well as effective sampling methods.

Keynote Speakers (confirmed)

• David J. Wales, FRS (Cambridge)
  Exploring energy landscapes: from molecules to nanodevices


• Robert B. Gramacy (Virginia Tech)
  Local approximate Gaussian processes for large computer experiments

Confirmed Speakers (some titles are tentative)

• Koji Shimoyama (Tohoku Univ)
  Surrogate modeling for numerical simulation and design
• Shinji Nishiwaki (Kyoto Univ)
  Topology optimization for the innovative structural designs of devices and materials
• Takashi Misaka (Tohoku Univ)
  Accelerating engineering data assimilation by reduced order modeling
• Wataru Kurebayashi (Aomori Univ)
  Kernel dynamic mode decomposition
• Hiroshi Takano (Keio Univ)
  Relaxation mode analysis of biopolymer systems
• Yasuhiro Matsunaga (RIKENCPRESTO)
  Markov state modeling of protein folding dynamics by combining single-molecule experiments and simulations
• Hiromichi Suetani (Oita Univ)
  Chaotic itinerancy in multiple patterns generation using FORCE-based neural network learning
• Ryo Yoshida (ISM)
  Inverse design of functional materials through computer experiments and machine learning
• Stephen Wu (ISM)
  A hierarchical Bayesian approach to force-field calibration using heterogeneous data
• Genta Ueno (ISM)
  Data assimilation and optimal error covariance
• Hiroshi Fujisaki (Nippon Medical School, ISM)
  Molecular dynamics simulations and dimensional reduction techniques
• Yukito Iba (ISM) and Shinichi Takayanagi (SOKEN DAI)
  Rare event sampling using multicanonical MCMC and time-reversed simulation

Programme

March 9 (Tur)

13:00-13:20 opening

13:20-14:00 Koji Shimoyama (40min)
Surrogate modeling for numerical simulation and design

14:00-15:30 Robert B. Gramacy (90min/keynote)
Local approximate Gaussian processes for large computer experiments

... (20min break)

15:50-16:50 Shinji Nishiwaki (60min/long)
Topology optimization for the innovative structural designs of devices and materials

16:50-17:30 Yukito Iba (ISM) and Shinichi Takayanagi (SOKEN DAI) (40min)
Rare event sampling using multicanonical MCMC and time-reversed simulation

March 10 (Fri)

10:30-11:10 Ryo Yoshida (40min)
Inverse design of functional materials through computer experiments and machine learning

11:10-11:50 Stephen Wu (40min)
A hierarchical Bayesian approach to force-field calibration using heterogeneous data

... (70min lunch break)

13:00-14:30 David J. Wales (90min/keynote)
Exploring energy landscapes: from molecules to nanodevices

... (40 min long break with discussions)

15:10-15:50 Hiroshi Fujisaki (40min)
Molecular dynamics simulations and dimensional reduction techniques

15:50-16:30 Yasuhiro Matsunaga (40min)
Markov state modeling of protein folding dynamics by combining single-molecule experiments and simulations

... (20 min break)

16:50-17:30 Genta Ueno (40min)
Data assimilation and optimal error covariance

March 11 (Sat)

10:30-11:10 Wataru Kurebayashi (40min)
Kernel dynamic mode decomposition

11:10-11:50 Hiroshi Takano (40min)
Relaxation mode analysis of biopolymer systems

... (70min lunch break)

13:00-13:40 Takashi Misaka (40min)
Accelerating engineering data assimilation by reduced order modeling

13:40-14:20 Hiromichi Suetani (40min)
Chaotic itinerancy in multiple patterns generation using FORCE-based neural network learning

14:20-14:30 closing

iba at ism.ac.jp