Multimoment convecting flux tube model of the polar wind system with return current and microprocesses

Banerjee, Supriya; Gavrishchaka, Valeriy V.

doi:10.1016/j.jastp.2007.08.004

Cited by 8 publications

(9 citation statements)

References 22 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Therefore, 3-dimensional, time-dependent modeling is needed to properly describe ionosphere-polar wind structures. There are several papers on the polar wind that provide an update on our current understanding (Barakat and Schunk, 2006;Banerjee and Gavrishchaka, 2007;Lemaire et al, 2007;Schunk, 2007;Tam et al, 2007;Yau et al, 2007). A substantial plasma structuring occurs even for the classical polar wind and even when relatively simple time-dependent plasma convection and particle precipitation patterns are adopted (cf.…”

Section: Ion Polar Windmentioning

confidence: 99%

Response of the ionosphere–thermosphere system to magnetospheric processes

Schunk

Zhu

2008

Journal of Atmospheric and Solar-Terrestrial Physics

View full text Add to dashboard Cite

Section: Ion Polar Windmentioning

confidence: 99%

Response of the ionosphere–thermosphere system to magnetospheric processes

Schunk

Zhu

2008

Journal of Atmospheric and Solar-Terrestrial Physics

View full text Add to dashboard Cite

“…However, performance of standard ML approaches could quickly deteriorate with severe data limitations, high dimensionality and non-stationarity [149,150]. Domain-expert knowledge including physical models based on deeper understanding of the considered complex system, such as the kinetic processes discussed in this article, could play a key role in applications with severe incompleteness of training data because of natural dimensionality reduction and usage of domain-specific constraints [149,150,151]. Typical practical example of the domain knowledge incorporation into ML solution is selection of model inputs and drivers using physics-based considerations [144,145,146,150].…”

Section: Comprehensive Modeling Of Space Plasma Environmentmentioning

confidence: 99%

“…For example, collection of simplified physical models with a few adjustable empirical parameters, e.g. anomalous coefficients capturing small-scale effects, could be used as base models in boosting algorithms to create ensemble of interpretable models with boosted accuracy and stability compared to a single model [149,150,151]. Alternatively, simplified physical models capturing multi-scale effects in an approximate manner can be used to generate large amounts of synthetic data for all possible regimes.…”

Section: Comprehensive Modeling Of Space Plasma Environmentmentioning

confidence: 99%

Behavior of Compressed Plasmas in Magnetic Fields

Ganguli¹,

Crabtree²,

Fletcher³

et al. 2020

Preprint

View full text Add to dashboard Cite

Plasma in the earth's magnetosphere is subjected to compression during geomagnetically active periods and relaxation in subsequent quiet times. Repeated compression and relaxation is the origin of much of the plasma dynamics and intermittency in the near-earth environment. An observable manifestation of compression is the thinning of the plasma sheet resulting in magnetic reconnection when the solar wind mass, energy, and momentum floods into the magnetosphere culminating in the spectacular auroral display. This phenomenon is rich in physics at all scale sizes, which are causally interconnected. This poses a formidable challenge in accurately modeling the physics. The large-scale processes are fluid-like and are reasonably well captured in the global magnetohydrodynamic (MHD) models, but those in the smaller scales responsible for dissipation and relaxation that feed back to the larger scale dynamics are often in the kinetic regime. The self-consistent generation of the small-scale processes and their feedback to the global plasma dynamics remains to be fully explored. Plasma compression can lead to the generation of electromagnetic fields that distort the particle orbits and introduce new features beyond the purview of the MHD framework, such as ambipolar electric fields, unequal plasma drifts and currents among species, strong spatial and velocity gradients in gyroscale layers separating plasmas of different characteristics, etc. These boundary layers are regions of intense activity characterized by emissions that are measurable. We study the behavior of such compressed plasmas and discuss the relaxation mechanisms to understand their measurable signatures as well as their feedback to influence the global scale plasma evolution.

show abstract

“…For example, the important problem of space-weather forecasting (i.e. prediction of storms and sub-storms in Earth's magnetosphere) is very challenging due to the interplay of physical processes of a vast range of time and spatial scales [50]. Besides the obvious drastic limitation of computing power, the fine-grain initial and boundary conditions are not known; it is impossible to have a satellite in every spatial location simultaneously.…”

Section: Synergy Of Physics-based Models and Machine Learning Inmentioning

confidence: 99%

“…Therefore, different kinds of model reformulations allow useful practical results to be obtained. Often, small-scale kinetic effects are introduced as anomalous coefficients into large-scale fluid simulations without running small-scale simulations [50]. One can also approximate the whole magnetosphere-ionosphere system as a giant, but a simple, electric circuit with just a few main elements having characteristics inferred from deeper physical models (analog models) [56].…”

Section: Synergy Of Physics-based Models and Machine Learning Inmentioning

confidence: 99%

Synergy of physics-based reasoning and machine learning in biomedical applications: towards unlimited deep learning with limited data

Gavrishchaka

Senyukova

Koepke

2019

Advances in Physics: X

View full text Add to dashboard Cite

Technological advancements enable collecting vast data, i. e., Big Data, in science and industry including biomedical field. Increased computational power allows expedient analysis of collected data using statistical and machine-learning approaches. Historical data incompleteness problem and curse of dimensionality diminish practical value of pure data-driven approaches, especially in biomedicine. Advancements in deep learning (DL) frameworks based on deep neural networks (DNN) improved accuracy in image recognition, natural language processing, and other applications yet severe data limitations and/or absence of transferlearning-relevant problems drastically reduce advantages of DNN-based DL. Our earlier works demonstrate that hierarchical data representation can be alternatively implemented without NN, using boosting-like algorithms for utilization of existing domain knowledge, tolerating significant data incompleteness, and boosting accuracy of low-complexity models within the classifier ensemble, as illustrated in physiological-data analysis. Beyond obvious use in initial-factor selection, existing simplified models are effectively employed for generation of realistic synthetic data for later DNN pre-training. We review existing machine learning approaches, focusing on limitations caused by trainingdata incompleteness. We outline our hybrid framework that leverages existing domain-expert models/knowledge, boosting-like model combination, DNN-based DL and other machine learning algorithms for drastic reduction of training-data requirements. Applying this framework is illustrated in context of analyzing physiological data.

show abstract

Multimoment convecting flux tube model of the polar wind system with return current and microprocesses

Cited by 8 publications

References 22 publications

Response of the ionosphere–thermosphere system to magnetospheric processes

Response of the ionosphere–thermosphere system to magnetospheric processes

Behavior of Compressed Plasmas in Magnetic Fields

Synergy of physics-based reasoning and machine learning in biomedical applications: towards unlimited deep learning with limited data

Contact Info

Product

Resources

About