# Hot Topic

## iTHEMS colloquium held on October 3rd

2018-10-04

Prof. Hideaki Aoyama from Kyoto University was the lecturer of the latest iTHEMS colloquium held on October 3rd. Prof. Aoyama has just joined iTHEMS as a senior visiting researcher. Welcome to iTHEMS, Prof. Aoyama!

The colloquium was entitled "Economic Networks: a Physicist's View. " In the opening of the talk, Prof. Aoyama shared his personal recollection of Richard Feynman, one of the greatest physicists of the last century, and quoted his words. He went on to explain the techniques developed to explore the correlations and hidden relations buried in the huge amount of complex data which concern the economic activities of the real world. He pointed out that a certain kind of similarities can be observed between physics and economics, for example, the time correlation of the aftershocks of earthquakes and the bankruptcies occurred after calamities. The colloquium attracted a wide range of audience including those from outside RIKEN.

Economic Networks: a Physicist's View

October 3 at 15:00 - 16:30, 2018

# Upcoming Events

## External Event

### Nerd Nite Tokyo

November 9 at 20:00 - 22:00, 2018

Dr. Ade Irma Suriajaya (Special Postdoctoral Researcher, iTHEMS)

Chacha will tell us about infinities, some bad math, and some good math.

Price: ¥1000, but speakers get in free

Food and drinks available at the event

Venue: Nagatacho GRID (2-5-3 Hirakawa-cho, Chiyoda-ku, Tokyo 102-0093)

Event Official Language: English

## Math Lecture

### Theory of Operator Algebras (4th)

October 18 at 15:30 - 17:00, 2018

Dr. Yosuke Kubota (Research Scientist, iTHEMS)

Venue: Seminar Room #160, 1F Main Research Building, RIKEN

Event Official Language: Japanese

## Math Lecture

### Theory of Operator Algebras (5th)

November 8 at 13:30 - 15:00, 2018

Dr. Yosuke Kubota (Research Scientist, iTHEMS)

Venue: Seminar Room #160, 1F Main Research Building, RIKEN

Event Official Language: Japanese

## Workshop

### Workshop on Recent Developments of Chiral Matter and Topology

December 6 - 9, 2018

The aim of this workshop is to gather researchers of high-energy and condensed-matter physics working on chiral Matter and Topology, to exchange ideas and establish collaborations to tackle unsolved issues and carry out future extensions. The workshop expects to welcome 40-60 participants who are interested in the aforementioned topics.

Organizers:

Tomoki Ozawa, Tetsuo Hatsuda (RIKEN iTHEMS)

Di-Lun Yang (RIKEN Nishina Center; YITP, Kyoto)

Chang-Tse Hsieh (Kavli IPMU / ISSP, the Univ. of Tokyo)

Jiunn-Wei Chen, Guang-Yu Guo (National Taiwan Univ.)

Hsiang-Nan Li (Academia Sinica)

Venue: National Taiwan University, Taipei, Taiwan

Event Official Language: English

## Math Lecture

### Theory of Operator Algebras (6th)

December 20 at 15:30 - 17:00, 2018

Dr. Yosuke Kubota (Research Scientist, iTHEMS)

Title: An introduction to operator algebras

Abstract: Operators are linear maps from a (usually an infinite dimensional) linear space (most frequently the Hilbert space) to itself, which is like matrices of infinite degree. Operators form an algebra by obvious addition and multiplication. Operators appear in most of the fields in mathematics, in algebra, in geometry, in analysis, ... Some of the key words at the beginning of these lectures are "spectral theory" "operator algebras" "Tomita-Takesaki theory". These lectures are for non-professional people.

Venue: Seminar Room #160, 1F Main Research Building, RIKEN

Event Official Language: Japanese

# Paper of the Week

## Self-learning Monte Carlo method with Behler-Parrinello neural networks

2018-10-01

Self Learning Monte-Carlo (SLMC) method is one of the recent promising applications of machine learning techniques to computational physics especially Marcov-Chain Monte-Carlo (MCMC) simulations in statistical physics systems. In SLMC, we prepare an effective Hamiltonian with tunable coupling constants and try to reproduce the value of the real Hamiltonian by adjusting the coupling constants. This procedure corresponds to the supervised machine learning. After the learning, we employ it for global updates in MCMC, and it drastically reduces the autocorrelation, i.e. similarity of configurations in the Marcov-Chain, thanks to the local structure of the effective Hamiltonian. In our paper, we present two novel techniques for SLMC in a quantum Monte-Carlo simulation. First technique is the use of neural networks. Based on the neural net architecture by Behler and Parrinello in the context of molecular dynamics machine learning, we develop how to construct general purposed effective Hamiltonian for SLMC. Second contribution is a new proposal for regularization in SLMC that we call batch-atom normalization. It is a generalization of the well-known technique in deep learning, and we observe it drastically improve the learning procedure for effective Hamiltonian represented by neural network.

Reference:

Yuki Nagai, Masahiko Okumura, Akinori Tanaka

"Self-learning Monte Carlo method with Behler-Parrinello neural networks"

arXiv: 1807.04955

If you would like to cancel your subscription or change your email address,

please let us know via our contact form.