Use of Data Mining and Computer Vision Algorithms in Studies of Magnetic Reconnection

Details Use of Data Mining and Computer Vision Algorithms in Studies of Magnetic Reconnection Use of Data Mining and Computer Vision Algorithms in Studies of Magnetic Reconnection

Dec 2011

adsabs.harvard.edu/cgi-bin/nph-data_query?bibcode=2011agufmsm21a1986s&link_type=abstract

American Geophysical Union, Fall Meeting 2011, abstract #SM21A-1986

Physics

Plasma Physics

[2723] Magnetospheric Physics / Magnetic Reconnection, [2724] Magnetospheric Physics / Magnetopause And Boundary Layers, [7835] Space Plasma Physics / Magnetic Reconnection

Scientific paper

Knowledge discovery from large data sets collected from spacecraft measurements as well as petascale simulations remains a major obstacle to scientific progress. For example, our recent 3D kinetic simulation of reconnection included over 3 trillion particles and generated well over 200 TB of data. Similarly identification of interesting features in spacecraft data can be quite time consuming and by definition focuses on simpler features as human eye has limited capability in deciphering complex patterns and dependencies. Machine learning algorithms offer a solution to this problem. Here we present our latest results on use of machine learning algorithms in analysis of (i) 2D and 3D kinetic simulations of reconnection and (ii) reconnection events in the solar wind using Wind data. The results are quite promising and point to the power of these techniques to find hidden relationships. For example, identification of flux ropes in the solar wind remains quite controversial since unlike the magnetopause where one can search for bipolar signatures of the magnetic field component in the boundary normal coordinates, there are no generally agreed upon method of identifying them. As a preparation for this, we show results of our technique applied to time series generated from simulations of flux ropes. We find that the algorithms were not only able to detect flux ropes in the simulation data very accurately, but they were also able to distinguish crossings across a flux rope versus those along the axis of a flux rope. In case of spacecraft data, our models were able to detect crossings of the reconnection exhausts and distinguish them from non-exhausts. Finally, we use machine learning algorithms to compare the crossings of reconnection exhausts from simulations and spacecraft observations in the solar wind.

Affiliated with

Also associated with

No associations

Rating

Use of Data Mining and Computer Vision Algorithms in Studies of Magnetic Reconnection does not yet have a rating. At this time, there are no reviews or comments for this scientific paper.

If you have personal experience with Use of Data Mining and Computer Vision Algorithms in Studies of Magnetic Reconnection, we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Use of Data Mining and Computer Vision Algorithms in Studies of Magnetic Reconnection will most certainly appreciate the feedback.

Rate now

Comments { 0 }

Profile ID: LFWR-SCP-O-1889918