Multi-step time series generator for molecular dynamics

Katsuhiro Endo, Katsufumi Tomobe, Kenji Yasuoka

Research output: Chapter in Book/Report/Conference proceedingConference contribution

1 Citation (Scopus)

Abstract

Molecular dynamics (MD) is a powerful computational method for simulating molecular behavior. Deep neural networks provide a novel method of generating MD data efficiently, but there is no architecture that mitigates the well-known exposure bias accumulated by multi-step generations. In this paper, we propose a multi-step time series generator using a deep neural network based on Wasserstein generative adversarial nets. Instead of sparse real data, our model evolves a latent variable z that is densely distributed in a low-dimensional space. This novel framework successfully mitigates the exposure bias. Moreover, our model can evolve part of the system (Feature extraction) with any time step (Step skip), which accelerates the efficient generation of MD data. The applicability of this model is evaluated through three different systems: harmonic oscillator, bulk water, and polymer melts. The experimental results demonstrate that our model can generate time series of the MD data with sufficient accuracy to calculate the physical and important dynamical statistics.

Original languageEnglish
Title of host publication32nd AAAI Conference on Artificial Intelligence, AAAI 2018
PublisherAAAI press
Pages2192-2199
Number of pages8
ISBN (Electronic)9781577358008
Publication statusPublished - 2018 Jan 1
Event32nd AAAI Conference on Artificial Intelligence, AAAI 2018 - New Orleans, United States
Duration: 2018 Feb 22018 Feb 7

Other

Other32nd AAAI Conference on Artificial Intelligence, AAAI 2018
CountryUnited States
CityNew Orleans
Period18/2/218/2/7

Fingerprint

Molecular dynamics
Time series
Polymer melts
Computational methods
Feature extraction
Statistics
Water
Deep neural networks

ASJC Scopus subject areas

  • Artificial Intelligence

Cite this

Endo, K., Tomobe, K., & Yasuoka, K. (2018). Multi-step time series generator for molecular dynamics. In 32nd AAAI Conference on Artificial Intelligence, AAAI 2018 (pp. 2192-2199). AAAI press.

Multi-step time series generator for molecular dynamics. / Endo, Katsuhiro; Tomobe, Katsufumi; Yasuoka, Kenji.

32nd AAAI Conference on Artificial Intelligence, AAAI 2018. AAAI press, 2018. p. 2192-2199.

Research output: Chapter in Book/Report/Conference proceedingConference contribution

Endo, K, Tomobe, K & Yasuoka, K 2018, Multi-step time series generator for molecular dynamics. in 32nd AAAI Conference on Artificial Intelligence, AAAI 2018. AAAI press, pp. 2192-2199, 32nd AAAI Conference on Artificial Intelligence, AAAI 2018, New Orleans, United States, 18/2/2.
Endo K, Tomobe K, Yasuoka K. Multi-step time series generator for molecular dynamics. In 32nd AAAI Conference on Artificial Intelligence, AAAI 2018. AAAI press. 2018. p. 2192-2199
Endo, Katsuhiro ; Tomobe, Katsufumi ; Yasuoka, Kenji. / Multi-step time series generator for molecular dynamics. 32nd AAAI Conference on Artificial Intelligence, AAAI 2018. AAAI press, 2018. pp. 2192-2199
@inproceedings{0c3446db424c4f5a982138b14e181d35,
title = "Multi-step time series generator for molecular dynamics",
abstract = "Molecular dynamics (MD) is a powerful computational method for simulating molecular behavior. Deep neural networks provide a novel method of generating MD data efficiently, but there is no architecture that mitigates the well-known exposure bias accumulated by multi-step generations. In this paper, we propose a multi-step time series generator using a deep neural network based on Wasserstein generative adversarial nets. Instead of sparse real data, our model evolves a latent variable z that is densely distributed in a low-dimensional space. This novel framework successfully mitigates the exposure bias. Moreover, our model can evolve part of the system (Feature extraction) with any time step (Step skip), which accelerates the efficient generation of MD data. The applicability of this model is evaluated through three different systems: harmonic oscillator, bulk water, and polymer melts. The experimental results demonstrate that our model can generate time series of the MD data with sufficient accuracy to calculate the physical and important dynamical statistics.",
author = "Katsuhiro Endo and Katsufumi Tomobe and Kenji Yasuoka",
year = "2018",
month = "1",
day = "1",
language = "English",
pages = "2192--2199",
booktitle = "32nd AAAI Conference on Artificial Intelligence, AAAI 2018",
publisher = "AAAI press",

}

TY - GEN

T1 - Multi-step time series generator for molecular dynamics

AU - Endo, Katsuhiro

AU - Tomobe, Katsufumi

AU - Yasuoka, Kenji

PY - 2018/1/1

Y1 - 2018/1/1

N2 - Molecular dynamics (MD) is a powerful computational method for simulating molecular behavior. Deep neural networks provide a novel method of generating MD data efficiently, but there is no architecture that mitigates the well-known exposure bias accumulated by multi-step generations. In this paper, we propose a multi-step time series generator using a deep neural network based on Wasserstein generative adversarial nets. Instead of sparse real data, our model evolves a latent variable z that is densely distributed in a low-dimensional space. This novel framework successfully mitigates the exposure bias. Moreover, our model can evolve part of the system (Feature extraction) with any time step (Step skip), which accelerates the efficient generation of MD data. The applicability of this model is evaluated through three different systems: harmonic oscillator, bulk water, and polymer melts. The experimental results demonstrate that our model can generate time series of the MD data with sufficient accuracy to calculate the physical and important dynamical statistics.

AB - Molecular dynamics (MD) is a powerful computational method for simulating molecular behavior. Deep neural networks provide a novel method of generating MD data efficiently, but there is no architecture that mitigates the well-known exposure bias accumulated by multi-step generations. In this paper, we propose a multi-step time series generator using a deep neural network based on Wasserstein generative adversarial nets. Instead of sparse real data, our model evolves a latent variable z that is densely distributed in a low-dimensional space. This novel framework successfully mitigates the exposure bias. Moreover, our model can evolve part of the system (Feature extraction) with any time step (Step skip), which accelerates the efficient generation of MD data. The applicability of this model is evaluated through three different systems: harmonic oscillator, bulk water, and polymer melts. The experimental results demonstrate that our model can generate time series of the MD data with sufficient accuracy to calculate the physical and important dynamical statistics.

UR - http://www.scopus.com/inward/record.url?scp=85055432338&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=85055432338&partnerID=8YFLogxK

M3 - Conference contribution

SP - 2192

EP - 2199

BT - 32nd AAAI Conference on Artificial Intelligence, AAAI 2018

PB - AAAI press

ER -