Particle Clustering in Turbulence: Prediction of Spatial and Statistical Properties with Deep Learning
We investigate the utility of deep learning for modeling the clustering of particles that are aerodynamically coupled to turbulent fluids. Using a Lagrangian particle module within the Athena++ hydrodynamics code, we simulate the dynamics of particles in the Epstein drag regime within a periodic dom...
| Main Authors: | , , , , , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
IOP Publishing
2023-01-01
|
| Series: | The Astrophysical Journal |
| Subjects: | |
| Online Access: | https://doi.org/10.3847/1538-4357/ad088c |
| _version_ | 1797385898574217216 |
|---|---|
| author | Yan-Mong Chan Natascha Manger Yin Li Chao-Chin Yang Zhaohuan Zhu Philip J. Armitage Shirley Ho |
| author_facet | Yan-Mong Chan Natascha Manger Yin Li Chao-Chin Yang Zhaohuan Zhu Philip J. Armitage Shirley Ho |
| author_sort | Yan-Mong Chan |
| collection | DOAJ |
| description | We investigate the utility of deep learning for modeling the clustering of particles that are aerodynamically coupled to turbulent fluids. Using a Lagrangian particle module within the Athena++ hydrodynamics code, we simulate the dynamics of particles in the Epstein drag regime within a periodic domain of isotropic forced hydrodynamic turbulence. This setup is an idealized model relevant to the collisional growth of micron- to millimeter-sized dust particles in early-stage planet formation. The simulation data are used to train a U-Net deep-learning model to predict gridded three-dimensional representations of the particle density and velocity fields, given as input the corresponding fluid fields. The trained model qualitatively captures the filamentary structure of clustered particles in a highly nonlinear regime. We assess model fidelity by calculating metrics of the density field (the radial distribution function) and of the velocity field (the relative velocity and the relative radial velocity between particles). Although trained only on the spatial fields, the model predicts these statistical quantities with errors that are typically <10%. Our results suggest that, given appropriately expanded training data, deep learning could complement direct numerical simulations in predicting particle clustering within turbulent flows. |
| first_indexed | 2024-03-08T22:00:58Z |
| format | Article |
| id | doaj.art-5003a5b7966d44e0b33447787eb252eb |
| institution | Directory Open Access Journal |
| issn | 1538-4357 |
| language | English |
| last_indexed | 2024-03-08T22:00:58Z |
| publishDate | 2023-01-01 |
| publisher | IOP Publishing |
| record_format | Article |
| series | The Astrophysical Journal |
| spelling | doaj.art-5003a5b7966d44e0b33447787eb252eb2023-12-19T14:38:04ZengIOP PublishingThe Astrophysical Journal1538-43572023-01-0196011910.3847/1538-4357/ad088cParticle Clustering in Turbulence: Prediction of Spatial and Statistical Properties with Deep LearningYan-Mong Chan0https://orcid.org/0000-0003-4469-8829Natascha Manger1https://orcid.org/0000-0001-6552-8605Yin Li2https://orcid.org/0000-0002-0701-1410Chao-Chin Yang3https://orcid.org/0000-0003-2589-5034Zhaohuan Zhu4https://orcid.org/0000-0003-3616-6822Philip J. Armitage5https://orcid.org/0000-0001-5032-1396Shirley Ho6https://orcid.org/0000-0002-1068-160XCenter for Computational Astrophysics, Flatiron Institute , 162 Fifth Avenue, New York, NY 10010, USA; Department of Physics, The Chinese University of Hong Kong , Shatin, NT, Hong KongCenter for Computational Astrophysics, Flatiron Institute , 162 Fifth Avenue, New York, NY 10010, USACenter for Computational Astrophysics, Flatiron Institute , 162 Fifth Avenue, New York, NY 10010, USADepartment of Physics and Astronomy, University of Alabama , Box 870324, Tuscaloosa, AL 35487-0324, USA; Department of Physics and Astronomy, University of Nevada , Las Vegas, 4505 S. Maryland Parkway, Box 454002, Las Vegas, NV 89154, USADepartment of Physics and Astronomy, University of Nevada , Las Vegas, 4505 S. Maryland Parkway, Box 454002, Las Vegas, NV 89154, USACenter for Computational Astrophysics, Flatiron Institute , 162 Fifth Avenue, New York, NY 10010, USA; Department of Physics and Astronomy, Stony Brook University , Stony Brook, NY 11794, USACenter for Computational Astrophysics, Flatiron Institute , 162 Fifth Avenue, New York, NY 10010, USA; Department of Astrophysical Sciences, Princeton University , Peyton Hall, Princeton, NJ 08544, USA; Department of Physics, Carnegie Mellon University , Pittsburgh, PA 15213, USAWe investigate the utility of deep learning for modeling the clustering of particles that are aerodynamically coupled to turbulent fluids. Using a Lagrangian particle module within the Athena++ hydrodynamics code, we simulate the dynamics of particles in the Epstein drag regime within a periodic domain of isotropic forced hydrodynamic turbulence. This setup is an idealized model relevant to the collisional growth of micron- to millimeter-sized dust particles in early-stage planet formation. The simulation data are used to train a U-Net deep-learning model to predict gridded three-dimensional representations of the particle density and velocity fields, given as input the corresponding fluid fields. The trained model qualitatively captures the filamentary structure of clustered particles in a highly nonlinear regime. We assess model fidelity by calculating metrics of the density field (the radial distribution function) and of the velocity field (the relative velocity and the relative radial velocity between particles). Although trained only on the spatial fields, the model predicts these statistical quantities with errors that are typically <10%. Our results suggest that, given appropriately expanded training data, deep learning could complement direct numerical simulations in predicting particle clustering within turbulent flows.https://doi.org/10.3847/1538-4357/ad088cHydrodynamical simulationsProtoplanetary disksNeural networksPlanet formation |
| spellingShingle | Yan-Mong Chan Natascha Manger Yin Li Chao-Chin Yang Zhaohuan Zhu Philip J. Armitage Shirley Ho Particle Clustering in Turbulence: Prediction of Spatial and Statistical Properties with Deep Learning The Astrophysical Journal Hydrodynamical simulations Protoplanetary disks Neural networks Planet formation |
| title | Particle Clustering in Turbulence: Prediction of Spatial and Statistical Properties with Deep Learning |
| title_full | Particle Clustering in Turbulence: Prediction of Spatial and Statistical Properties with Deep Learning |
| title_fullStr | Particle Clustering in Turbulence: Prediction of Spatial and Statistical Properties with Deep Learning |
| title_full_unstemmed | Particle Clustering in Turbulence: Prediction of Spatial and Statistical Properties with Deep Learning |
| title_short | Particle Clustering in Turbulence: Prediction of Spatial and Statistical Properties with Deep Learning |
| title_sort | particle clustering in turbulence prediction of spatial and statistical properties with deep learning |
| topic | Hydrodynamical simulations Protoplanetary disks Neural networks Planet formation |
| url | https://doi.org/10.3847/1538-4357/ad088c |
| work_keys_str_mv | AT yanmongchan particleclusteringinturbulencepredictionofspatialandstatisticalpropertieswithdeeplearning AT nataschamanger particleclusteringinturbulencepredictionofspatialandstatisticalpropertieswithdeeplearning AT yinli particleclusteringinturbulencepredictionofspatialandstatisticalpropertieswithdeeplearning AT chaochinyang particleclusteringinturbulencepredictionofspatialandstatisticalpropertieswithdeeplearning AT zhaohuanzhu particleclusteringinturbulencepredictionofspatialandstatisticalpropertieswithdeeplearning AT philipjarmitage particleclusteringinturbulencepredictionofspatialandstatisticalpropertieswithdeeplearning AT shirleyho particleclusteringinturbulencepredictionofspatialandstatisticalpropertieswithdeeplearning |