English (United Kingdom)

https://curated-unify.zendy.io/wp-json/zendy-region/v1/featured_content/oa?rat=en

https://curated-unify.zendy.io/wp-json/zendy-region/v1/highlighted_journal/

Global: Zendy Plus annual

Presents the access of premium content as premium feature

Premium Content

Presents the keyphrase highlighting as premium feature

Keyphrase Highlighting

Presents the summarisation as premium feature

Summarisation

Insights

Presents the pdf analysis as premium feature

PDF Analysis

Presents the zaia usage as premium feature

ZAIA

Global: Zendy Tools annual

Global: Zendy Free Plan

Global: Zendy Tools monthly

Global: Zendy Plus monthly

The interstellar medium (ISM) of galaxies is composed of a turbulentmagnetized plasma. In order to quantitatively measure relevant turbulentparameters of the ISM, a wide variety of statistical techniques and metricshave been developed that are often tested using numerical simulations andanalytic formalism. These metrics are typically based on the Fourier powerspectrum, which does not capture the Fourier phase information that carries themorphological characteristics of images. In this work we use density slices ofmagnetohydrodyanmic turbulence simulations to demonstrate that a modern tool,convolutional neural networks, can capture significant information encoded inthe Fourier phases. We train the neural network to distinguish between twosimulations with different levels of magnetization. We find that, even given atiny slice of simulation data, a relatively simple network can distinguishsub-Alfv\'enic (strong magnetic field) and super-Alfv\'enic (weak magneticfield) turbulence >98% of the time, even when all spectral amplitudeinformation is stripped from the images. In order to better understand how theneural network is picking out differences betweem the two classes ofsimulations we apply a neural network analysis method called "saliency maps".The saliency map analysis shows that sharp ridge-like features are adistinguishing morphological characteristic in such simulations. Our analysisprovides a way forward for deeper understanding of the relationship betweenmagnetohydrodyanmic turbulence and gas morphology and motivates furtherapplications of neural networks for studies of turbulence. We make publiclyavailable all data and software needed to reproduce our results.

Do Androids Dream of Magnetic Fields? Using Neural Networks to Interpret the Turbulent Interstellar Medium