On the importance of learning non-local dynamics for stable data-driven climate modeling: A 1D gravity wave-QBO testbed
Authors:
Hamid A. Pahlavan,
Pedram Hassanzadeh,
M. Joan Alexander
Abstract:
Machine learning (ML) techniques, especially neural networks (NNs), have shown promise in learning subgrid-scale parameterizations for climate models. However, a major problem with data-driven parameterizations, particularly those learned with supervised algorithms, is model instability. Current remedies are often ad-hoc and lack a theoretical foundation. Here, we combine ML theory and climate phy…
▽ More
Machine learning (ML) techniques, especially neural networks (NNs), have shown promise in learning subgrid-scale parameterizations for climate models. However, a major problem with data-driven parameterizations, particularly those learned with supervised algorithms, is model instability. Current remedies are often ad-hoc and lack a theoretical foundation. Here, we combine ML theory and climate physics to address a source of instability in NN-based parameterization. We demonstrate the importance of learning spatially $\textit{non-local}$ dynamics using a 1D model of the quasi-biennial oscillation (QBO) with gravity wave (GW) parameterization as a testbed. While common offline metrics fail to identify shortcomings in learning non-local dynamics, we show that the concept of receptive field (RF) can identify instability a-priori. We find that NN-based parameterizations that seem to accurately predict GW forcings from wind profiles ($\mathbf{R^2 \approx 0.99}$) cause unstable simulations when RF is too small to capture the non-local dynamics, while NNs of the same size but large-enough RF are stable. We examine three broad classes of architectures, namely convolutional NNs, Fourier neural operators, and fully-connected NNs; the latter two have inherently large RFs. We also demonstrate that learning non-local dynamics is crucial for the stability and accuracy of a data-driven spatiotemporal emulator of the zonal wind field. Given the ubiquity of non-local dynamics in the climate system, we expect the use of effective RF, which can be computed for any NN architecture, to be important for many applications. This work highlights the necessity of integrating ML theory with physics to design and analyze data-driven algorithms for weather and climate modeling.
△ Less
Submitted 15 July, 2024; v1 submitted 6 July, 2024;
originally announced July 2024.
Virtual Machine Consolidation for Datacenter Energy Improvement
Authors:
Sina Esfandiarpoor,
Ali Pahlavan,
Maziar Goudarzi
Abstract:
Rapid growth and proliferation of cloud computing services around the world has increased the necessity and significance of improving the energy efficiency of could implementations. Virtual machines (VM) comprise the backend of most, if not all, cloud computing services. Several VMs are often consolidated on a physical machine to better utilize its resources. We take into account the cooling and n…
▽ More
Rapid growth and proliferation of cloud computing services around the world has increased the necessity and significance of improving the energy efficiency of could implementations. Virtual machines (VM) comprise the backend of most, if not all, cloud computing services. Several VMs are often consolidated on a physical machine to better utilize its resources. We take into account the cooling and network structure of the datacenter hosting the physical machines when consolidating the VMs so that fewer racks and routers are employed, without compromising the service-level agreements, so that unused routing and cooling equipment can be turned off to reduce energy consumption. Our experimental results on four benchmarks shows that our technique improves energy consumption of servers, network equipment, and cooling systems by 2.5%, 18.8%, and 28.2% respectively, resulting in a total of 14.7% improvement on average in the entire datacenter.
△ Less
Submitted 9 February, 2013;
originally announced February 2013.
