Structural Parameter Space Exploration for Reinforcement Learning via a Matrix Variate Distribution

2022 ◽  
pp. 1-11
Author(s):  
Shaochen Wang ◽  
Rui Yang ◽  
Bin Li ◽  
Zhen Kan
Keyword(s):  
Reinforcement Learning ◽  
Parameter Space ◽  
Space Exploration ◽  
Structural Parameter

Rapid Characterization and Parameter Space Exploration of Perovskites Using an Automated Routine

2019 ◽  
Vol 22 (1) ◽  
pp. 6-17 ◽  
Author(s):  
Elisabeth Reinhardt ◽  
Ahmed M. Salaheldin ◽  
Monica Distaso ◽  
Doris Segets ◽  
Wolfgang Peukert
Keyword(s):  
Parameter Space ◽  
Space Exploration ◽  
Rapid Characterization

scholarly journals Parameter space exploration of the minimal SU(5) unification

2021 ◽  
Vol 104 (1) ◽  
Author(s):  
Ilja Doršner ◽  
Emina Džaferović-Mašić ◽  
Shaikh Saad
Keyword(s):  
Parameter Space ◽  
Space Exploration

scholarly journals Prediction of Agricultural Commodities Futures Prices: A DQN-LSTM Method

2021 ◽  
Author(s):  
Hanxiao Xu ◽  
Jie Liang ◽  
Wenchaun Zang
Keyword(s):  
Deep Learning ◽  
Reinforcement Learning ◽  
Prediction Accuracy ◽  
Future Application ◽  
Support Vector ◽  
Structural Parameter ◽  
Futures Prices ◽  
Agricultural Commodities ◽  
Short Term ◽  
Time Duration

Abstract This paper combines deep Q network (DQN) with long and short-term memory (LSTM) and proposes a novel hybrid deep learning method called DQN-LSTM framework. The proposed method aims to address the prediction of five Chinese agricultural commodities futures prices over different time duration. The DQN-LSTM applies the strategy enhancement of deep reinforcement learning to the structural parameter optimization of deep recurrent networks, and achieves the organic integration of two types of deep learning algorithms. The new framework has the capacity of self-optimization and learning of parameters, thus improving the performance of prediction by its own iteration, which shows great prospects for future application in financial prediction and other directions. The performance of the proposed method is evaluated by comparing the effectiveness of the DQN-LSTM method with that of traditional predicting methods such as auto-regressive integrated moving average (ARIMA), support vector machine (SVR) and LSTM. The results show that the DQN-LSTM method can effectively optimize the traditional LSTM structural parameters through policy iteration of the deep reinforcement learning algorithm, which contributes to a better long and short-term prediction accuracy. In particular, the longer the prediction period, the more obvious the advantage of prediction accuracy of a DQN-LSTM method.

scholarly journals InSituNet: Deep Image Synthesis for Parameter Space Exploration of Ensemble Simulations

2019 ◽  
pp. 1-1 ◽  
Author(s):  
Wenbin He ◽  
Junpeng Wang ◽  
Hanqi Guo ◽  
Ko-Chih Wang ◽  
Han-Wei Shen ◽  
...  
Keyword(s):  
Parameter Space ◽  
Space Exploration ◽  
Image Synthesis ◽  
Ensemble Simulations ◽  
Deep Image

scholarly journals Constraining protoplanetary discs with exoplanetary dynamics: Kepler-419 as an example

2020 ◽  
Vol 499 (1) ◽  
pp. 106-115
Author(s):  
Mohamad Ali-Dib ◽  
Cristobal Petrovich
Keyword(s):  
Parameter Space ◽  
Time Scale ◽  
Space Exploration ◽  
Semimajor Axis ◽  
Higher Order ◽  
Mean Motion Resonances ◽  
Close Encounters ◽  
Natural Result ◽  
Core Accretion ◽  
Orbital Migration

ABSTRACT We investigate the origins of Kepler-419, a peculiar system hosting two nearly coplanar and highly eccentric gas giants with apsidal orientations liberating around anti-alignment, and use this system to place constraints on the properties of their birth protoplanetary disc. We follow the proposal by Petrovich, Wu, & Ali-Dib that these planets have been placed on these orbits as a natural result of the precessional effects of a dissipating massive disc and extend it by using direct N-body simulations and models for the evolution of the gas discs, including photoevaporation. Based on a parameter space exploration, we find that in order to reproduce the system the initial disc mass had to be at least 95 MJup and dissipate on a time-scale of at least 104 yr. This mass is consistent with the upper end of the observed disc masses distribution, and the dissipation time-scale is consistent with photoevaporation models. We study the properties of such discs using simplified 1D thin-disc models and show that they are gravitationally stable, indicating that the two planets must have formed via core accretion and thus prone to disc migration. We hence finally investigate the sensitivity of this mechanism to the outer planet’s semimajor axis, and find that the nearby 7:1, 8:1, and 9:1 mean-motion resonances can completely quench this mechanism, while even higher order resonances can also significantly affect the system. Assuming the two planets avoid these high-order resonances and close encounters, the dynamics seems to be rather insensitive to planet c semimajor axis, and thus orbital migration driven by the disc.

Sign in / Sign up

Export Citation Format

Share Document