Deep learning enabled Lagrangian particle trajectory simulation

Abstract We introduce a deep learning method to simulate the chaotic motion of particles trapped in the recirculation zone of a buoyancy-opposed flame. The Lagrangian trajectories of particles in flame were captured using a high-speed camera and subsequently reconstructed in a 3-dimensional cylindrical space. These experimentally determined trajectories were next used to train the two most widely adopted deep generative models – the variational autoencoder (VAE) and the generative adversarial net (GAN). The performance of both models was then benchmarked according to statistical analysis performed on both the simulated trajectories and the ground truth, regarding accuracy and generalization criteria. Our results show that the GAN model produces non-repeating trajectories precisely capturing the statistical features of those determined in experiments, as revealed in their spatial-temporal scaling relationship and linear pair-correlation values; whereas the VAE model trained with the same amount of data tends to overfit, producing inferior results grappling with the trade-off between accuracy and generalization criteria. The influence of the size of training data on model performance is also evaluated. This paper concludes with a discussion on the potential utility of the deep learning enabled trajectory simulation.

[1]  Rajan K. Chakrabarty,et al.  Fractal scaling of soot packing density across five size decades , 2017 .

[2]  Yang Wang,et al.  Kinetics of sub-2 nm TiO2 particle formation in an aerosol reactor during thermal decomposition of titanium tetraisopropoxide , 2015, Journal of Nanoparticle Research.

[3]  Yang Wang,et al.  Synthesis of Titanium Dioxide Aerosol Gels in a Buoyancy-Opposed Flame Reactor , 2015 .

[4]  Tawatchai Charinpanitkul,et al.  Particle trajectories and temperature histories of TiO2 nanoparticles synthesized in diffusion flame reactor. , 2009, Journal of nanoscience and nanotechnology.

[5]  Arun Agarwal,et al.  Recurrent neural network and a hybrid model for prediction of stock returns , 2015, Expert Syst. Appl..

[6]  S. Friedlander,et al.  Smoke, dust, and haze , 2000 .

[7]  Nicolas E. Humphries,et al.  Scaling laws of marine predator search behaviour , 2008, Nature.

[8]  Gabriel Kreiman,et al.  Unsupervised Learning of Visual Structure using Predictive Generative Networks , 2015, ArXiv.

[9]  Hassan Aref,et al.  The development of chaotic advection , 2002 .

[10]  Rajan K. Chakrabarty,et al.  Fractal scaling of coated soot aggregates , 2017 .

[11]  Andrey G. Cherstvy,et al.  Anomalous diffusion models and their properties: non-stationarity, non-ergodicity, and ageing at the centenary of single particle tracking. , 2014, Physical chemistry chemical physics : PCCP.

[12]  J. Nathan Kutz,et al.  Deep learning in fluid dynamics , 2017, Journal of Fluid Mechanics.

[13]  Yoshua Bengio,et al.  Generative Adversarial Nets , 2014, NIPS.

[14]  Frederic Bartumeus,et al.  ANIMAL SEARCH STRATEGIES: A QUANTITATIVE RANDOM‐WALK ANALYSIS , 2005 .

[15]  Milorad P. Dudukovic,et al.  A Lagrangian description of flows in stirred tanks via computer-automated radioactive particle tracking (CARPT) , 2001 .

[16]  Amir Barati Farimani,et al.  Weakly-Supervised Deep Learning of Heat Transport via Physics Informed Loss , 2018, ArXiv.

[17]  Sander,et al.  Ballistic deposition on surfaces. , 1986, Physical review. A, General physics.

[18]  Carl Doersch,et al.  Tutorial on Variational Autoencoders , 2016, ArXiv.

[19]  Josh Kohut,et al.  Trajectory prediction using HF radar surface currents: Monte Carlo simulations of prediction uncertainties , 2006 .

[20]  A. Dehbi A stochastic Langevin model of turbulent particle dispersion in the presence of thermophoresis , 2009 .

[21]  Max Welling,et al.  Auto-Encoding Variational Bayes , 2013, ICLR.

[22]  Kuan-Yu Shen,et al.  Establishing the kinetics of ballistic-to-diffusive transition using directional statistics. , 2018, Physical review. E.

[23]  Christian Osendorfer,et al.  Learning Stochastic Recurrent Networks , 2014, NIPS 2014.

[24]  Ole Winther,et al.  Sequential Neural Models with Stochastic Layers , 2016, NIPS.

[25]  Nicolas E. Humphries,et al.  Environmental context explains Lévy and Brownian movement patterns of marine predators , 2010, Nature.

[26]  Paul M. Torrens,et al.  An extensible simulation environment and movement metrics for testing walking behavior in agent-based models , 2012, Comput. Environ. Urban Syst..

[27]  Bin Jiang,et al.  Agent-based simulation of human movement shaped by the underlying street structure , 2009, Int. J. Geogr. Inf. Sci..

[28]  Soumith Chintala,et al.  Unsupervised Representation Learning with Deep Convolutional Generative Adversarial Networks , 2015, ICLR.

[29]  Dibyendu Roy,et al.  Langevin dynamics in crossed magnetic and electric fields: Hall and diamagnetic fluctuations. , 2008, Physical review. E, Statistical, nonlinear, and soft matter physics.

[30]  Pearn P. Niiler,et al.  A Lagrangian description of motion in Northern California coastal transition filaments , 1990 .

[31]  Solomon,et al.  Observation of anomalous diffusion and Lévy flights in a two-dimensional rotating flow. , 1993, Physical review letters.

[32]  Sebastian Johann Wetzel,et al.  Unsupervised learning of phase transitions: from principal component analysis to variational autoencoders , 2017, Physical review. E.

[33]  Pratim Biswas,et al.  Numerical modeling of the performance of high flow DMAs to classify sub-2 nm particles , 2018, Aerosol Science and Technology.

[34]  P. Meakin Formation of fractal clusters and networks by irreversible diffusion-limited aggregation , 1983 .

[35]  Markus Raffel,et al.  Particle Image Velocimetry: A Practical Guide , 2002 .

[36]  Geoffrey E. Hinton,et al.  Deep Learning , 2015, Nature.

[37]  Geoffrey E. Hinton,et al.  ImageNet classification with deep convolutional neural networks , 2012, Commun. ACM.

[38]  Nick Collings,et al.  New method for particle mass classification : the Couette centrifugal particle mass analyzer , 2005 .

[39]  Yoshua Bengio,et al.  Gradient-based learning applied to document recognition , 1998, Proc. IEEE.

[40]  H. Aref Chaotic advection of fluid particles , 1990, Philosophical Transactions of the Royal Society of London. Series A: Physical and Engineering Sciences.

[41]  Chenyang Lu,et al.  Spatiotemporal distribution of indoor particulate matter concentration with a low-cost sensor network , 2018 .

[42]  Çagdas Hakan Aladag,et al.  Recurrent Multiplicative Neuron Model Artificial Neural Network for Non-linear Time Series Forecasting , 2014, Neural Processing Letters.

[43]  Rajan K. Chakrabarty,et al.  Trapping and aerogelation of nanoparticles in negative gravity hydrocarbon flames , 2014 .

[44]  Garrison W. Cottrell,et al.  A Dual-Stage Attention-Based Recurrent Neural Network for Time Series Prediction , 2017, IJCAI.

[45]  Rajan K. Chakrabarty,et al.  Breakdown of fractal dimension invariance in high monomer-volume-fraction aerosol gels , 2018, Aerosol Science and Technology.

[46]  Ian J. Goodfellow,et al.  NIPS 2016 Tutorial: Generative Adversarial Networks , 2016, ArXiv.

[47]  Mohammad M. Sultan,et al.  Variational encoding of complex dynamics. , 2017, Physical review. E.