Conservation laws in a neural network architecture: enforcing the atom balance of a Julia-based photochemical model (v0.2.0)
<p>Models of atmospheric phenomena provide insight into climate, air quality, and meteorology and provide a mechanism for understanding the effect of future emissions scenarios. To accurately represent atmospheric phenomena, these models consume vast quantities of computational resources. Mach...
Main Authors: | , |
---|---|
Format: | Article |
Language: | English |
Published: |
Copernicus Publications
2022-04-01
|
Series: | Geoscientific Model Development |
Online Access: | https://gmd.copernicus.org/articles/15/3417/2022/gmd-15-3417-2022.pdf |
_version_ | 1811303482035011584 |
---|---|
author | P. O. Sturm A. S. Wexler A. S. Wexler A. S. Wexler A. S. Wexler |
author_facet | P. O. Sturm A. S. Wexler A. S. Wexler A. S. Wexler A. S. Wexler |
author_sort | P. O. Sturm |
collection | DOAJ |
description | <p>Models of atmospheric phenomena provide insight into
climate, air quality, and meteorology and provide a mechanism for
understanding the effect of future emissions scenarios. To accurately
represent atmospheric phenomena, these models consume vast quantities of
computational resources. Machine learning (ML) techniques such as neural
networks have the potential to emulate computationally intensive components of these
models to reduce their computational burden. However, such ML surrogate
models may lead to nonphysical predictions that are difficult to uncover.
Here we present a neural network architecture that enforces conservation
laws to numerical precision. Instead of simply predicting properties of
interest, a physically interpretable hidden layer within the network
predicts fluxes between properties which are subsequently related to the
properties of interest. This approach is readily generalizable to physical
processes where flux continuity is an essential governing equation. As an
example application, we demonstrate our approach on a neural network
surrogate model of photochemistry, trained to emulate a reference model that
simulates formation and reaction of ozone. We design a physics-constrained
neural network surrogate model of photochemistry using this approach and
find that it conserves atoms as they flow between molecules while
outperforming two other neural network architectures in terms of accuracy,
physical consistency, and non-negativity of concentrations.</p> |
first_indexed | 2024-04-13T07:47:42Z |
format | Article |
id | doaj.art-2b833cce3f2c40dc8c4a6369629a735a |
institution | Directory Open Access Journal |
issn | 1991-959X 1991-9603 |
language | English |
last_indexed | 2024-04-13T07:47:42Z |
publishDate | 2022-04-01 |
publisher | Copernicus Publications |
record_format | Article |
series | Geoscientific Model Development |
spelling | doaj.art-2b833cce3f2c40dc8c4a6369629a735a2022-12-22T02:55:37ZengCopernicus PublicationsGeoscientific Model Development1991-959X1991-96032022-04-01153417343110.5194/gmd-15-3417-2022Conservation laws in a neural network architecture: enforcing the atom balance of a Julia-based photochemical model (v0.2.0)P. O. Sturm0A. S. Wexler1A. S. Wexler2A. S. Wexler3A. S. Wexler4Air Quality Research Center, University of California, Davis, Davis, California 95616, USAAir Quality Research Center, University of California, Davis, Davis, California 95616, USADepartment of Mechanical and Aerospace Engineering, University of California, Davis, Davis, California 95616, USADepartment of Civil and Environmental Engineering, University of California, Davis, Davis, California 95616, USADepartment of Land, Air and Water Resources, University of California, Davis, Davis, California 95616, USA<p>Models of atmospheric phenomena provide insight into climate, air quality, and meteorology and provide a mechanism for understanding the effect of future emissions scenarios. To accurately represent atmospheric phenomena, these models consume vast quantities of computational resources. Machine learning (ML) techniques such as neural networks have the potential to emulate computationally intensive components of these models to reduce their computational burden. However, such ML surrogate models may lead to nonphysical predictions that are difficult to uncover. Here we present a neural network architecture that enforces conservation laws to numerical precision. Instead of simply predicting properties of interest, a physically interpretable hidden layer within the network predicts fluxes between properties which are subsequently related to the properties of interest. This approach is readily generalizable to physical processes where flux continuity is an essential governing equation. As an example application, we demonstrate our approach on a neural network surrogate model of photochemistry, trained to emulate a reference model that simulates formation and reaction of ozone. We design a physics-constrained neural network surrogate model of photochemistry using this approach and find that it conserves atoms as they flow between molecules while outperforming two other neural network architectures in terms of accuracy, physical consistency, and non-negativity of concentrations.</p>https://gmd.copernicus.org/articles/15/3417/2022/gmd-15-3417-2022.pdf |
spellingShingle | P. O. Sturm A. S. Wexler A. S. Wexler A. S. Wexler A. S. Wexler Conservation laws in a neural network architecture: enforcing the atom balance of a Julia-based photochemical model (v0.2.0) Geoscientific Model Development |
title | Conservation laws in a neural network architecture: enforcing the atom balance of a Julia-based photochemical model (v0.2.0) |
title_full | Conservation laws in a neural network architecture: enforcing the atom balance of a Julia-based photochemical model (v0.2.0) |
title_fullStr | Conservation laws in a neural network architecture: enforcing the atom balance of a Julia-based photochemical model (v0.2.0) |
title_full_unstemmed | Conservation laws in a neural network architecture: enforcing the atom balance of a Julia-based photochemical model (v0.2.0) |
title_short | Conservation laws in a neural network architecture: enforcing the atom balance of a Julia-based photochemical model (v0.2.0) |
title_sort | conservation laws in a neural network architecture enforcing the atom balance of a julia based photochemical model v0 2 0 |
url | https://gmd.copernicus.org/articles/15/3417/2022/gmd-15-3417-2022.pdf |
work_keys_str_mv | AT posturm conservationlawsinaneuralnetworkarchitectureenforcingtheatombalanceofajuliabasedphotochemicalmodelv020 AT aswexler conservationlawsinaneuralnetworkarchitectureenforcingtheatombalanceofajuliabasedphotochemicalmodelv020 AT aswexler conservationlawsinaneuralnetworkarchitectureenforcingtheatombalanceofajuliabasedphotochemicalmodelv020 AT aswexler conservationlawsinaneuralnetworkarchitectureenforcingtheatombalanceofajuliabasedphotochemicalmodelv020 AT aswexler conservationlawsinaneuralnetworkarchitectureenforcingtheatombalanceofajuliabasedphotochemicalmodelv020 |