Abstract
Techniques for analysis of quasi coherent multi-channel time series data are explained, with examples from mode and profile analysis in plasma physics, using the open source pyfusion python library and the statistics package R. Spectral and coherence analysis are discussed in relation to pre-processing the raw data, and an application of singular value decomposition is described which assists in extracting the important information from these big data sets, which may be 1- 10s of gigabytes per experimental day. Various clustering algorithms are used to group and identify magnetic fluctuation signatures, collapsing the vast and high dimensional data space to a much more compact form, and providing insight into the physics of the fluctuations.
| Original language | English |
|---|---|
| Pages (from-to) | 342-346pp |
| Journal | Journal of Plasma and Fusion Research |
| Volume | 92 |
| Issue number | 5 |
| Publication status | Published - 2016 |