further reading

where to go after the demo · 2D app · 3D app · how to use this page

some papers, books, and lecture notes that might be interesting. the first four sections are the "basics", in order of distance from the simulations performed by this page. then, a long list of topics that I (Alfred) find particularly interesting. just dip into it as you like, probably reading from top to bottom is not a good idea.

the basics

ordered by distance from what you just watched on the other page.

the equations this page solves

reduced magnetohydrodynamics (RMHD), the equation set solved in the demos, is a simplified description of a plasma in a system with a strong background magnetic field (there are many restrictions on when it is appropriate, see if you can figure out when it would break down). it was originally derived in the context of the fusion program, independently by US and Soviet researchers. it turns out to also be really useful in the context of astrophysical turbulence!

learning plasma physics

I (Alfred) owe a great deal to Alex Schekochihin's teaching, but if you want a gentler introduction, use Fitzpatrick (free and online!) or (one of my favourite books) Choudhuri. Matt Kunz notes are also extremely useful and complementary to Alex's course. you probably need to know some classical mechanics and E&M.

learning turbulence

a warning: turbulence in neutral fluids ("hydro") is usually taught as a completely different subject to MHD turbulence, despite many similarities. the Schekochihin paper here is a complete, modern, guided tour to the MHD side, while the Frisch and Davidson books are for the hydro side, in different but complementary styles.

heliophysics

the Sun is our closest star, and now with NASA's Parker Solar Probe, we have actually visited its outer atmosphere, the solar corona. the solar system generally is a great laboratory for plasma turbulence: throughout the space age, humanity has explored the solar system with spacecraft, taking increasingly detailed measurements of the plasma. here are some reviews.

deep lore (subjective)

a list of more specialized topics that I (Alfred) find particularly interesting, or have been personally involved in. these references are woefully incomplete, but hopefully at least give a flavour of the topics involved.

a brief history of (MHD) turbulence

from the mathematical foundation by Kolmogorov up to critical balance in MHD by Goldreich & Sridhar.

dynamic alignment

numerical simulations of 3D MHD turbulence often stubbornly refuse to measure −5/3 spectra; instead typically obtaining −3/2. Boldyrev explained this with a very elegant argument: locally, velocity and magnetic field vectors dynamically align, causing the nonlinearity to deplete in a scale-dependent way. this was rather controversial, with others (e.g. Andrey Beresnyak) objecting, and claiming that alignment saturates and the asymptotic slope is −5/3. this was followed by nearly a decade of each group answering the other with ever-bigger simulations (an arms race up to resolutions of 4096 cubed!), and increasingly snarky replies. note that this resolution corresponds to inertial ranges still grossly insufficient for real astrophysical purposes. I (Alfred) was put to work on dynamic alignment as a short project at the start of my career; it ended up taking 7 years for us to figure out a way to resolve this.

reconnection takes over: the disruption scale

dynamically aligned eddies get more sheet-like as they cascade, which means that at some critical scale, they are thin enough for the tearing mode to break them up (see the tearing mode demo!). in other words, given enough resolution, the aligning cascade disrupts, forcing a transition to a new kind of turbulence: reconnecting turbulence, with a predicted spectral index of −11/5. this was predicted in 2017, simultaneously by two groups: Mallet, Schekochihin & Chandran, and Loureiro & Boldyrev: we discovered close to publication that we were working along the same lines, and agreed to put our preprints on arXiv on the same day. at the time I thought this theory was completely untestable numerically, but soon enough I was proved wrong by 10000²×5000 simulations by Dong et al. 2022, who largely confirmed our theory!

imbalance: an unsolved problem

turbulence in the solar wind and corona is strongly "imbalanced": z+>>z-, with Alfvén waves preferentially travelling away from the Sun. this allows an extra degree of freedom that has never been satisfactorily closed off: a dirty secret when it comes to applying the theory to the real world! these are some references, from the usual suspects. you can check out what happens in this case (at modest resolution) in the 3D forced turbulence demo by modifying epsilon+ and epsilon-.

reflection-driven turbulence

what sets the imbalance? waves launched from the Sun partially reflect from the gradients in density and magnetic field, driving the turbulent cascade. this means the turbulence is no longer homogeneous! taking this into account will be a major goal of turbulence theory for the next few years.

how to measure anisotropy

it turns out that actually measuring critical balance is not trivial: because of the strong wandering of the field lines, the relevant mean field is the local one, including the fluctuations, so a standard Fourier transform would hide it. this was solved in simulations by Cho & Vishniac and Maron & Goldreich by using structure functions, and in the solar wind data by using wavelet transforms by Horbury et al. in simulations, we can also carefully step along the field lines and take Fourier transforms, which is what is done in the demos.

weak turbulence, and the transition to strong

if the nonlinear time is much longer than the Alfvén propagation time, one can do formal perturbation theory. this beautiful but slightly arcane theory was worked out around the turn of the millennium. worth noting that most observed Alfvénic turbulence is, in fact, strong, and even if weak at large scales, it becomes strong: see Meyrand et al. (numerics) and Zhao et al. (observations) below. what happens to the spectrum & anisotropy in the 3D forced demo if you make the energy injection rate as small as possible?

intermittency

the spectrum isn't the whole story! most of the dissipation is concentrated in a small part of the volume, in large coherent structures. this breaks the self-similarity of Kolmogorov 1941: the fix is a more complex statistical model, which in MHD has to include the effects of critical balance and dynamic alignment.

collisionless turbulence and gyrokinetics

at small enough scales, the ion's gyration around the magnetic field has a larger radius than the size of the eddies, and RMHD breaks down. what happens then? a good framework for this is gyrokinetics: another model derived first in the fusion world and later applied very successfully to astrophysics. more recently, the discovery of the helicity barrier by Meyrand et al. has revolutionized this field: showing that small-scale effects can break even the most fundamental rules of turbulence.

collisionless high-beta dynamics: firehose and mirror

when the magnetic field is relatively weak (compared to the thermal pressure), the plasma pressure can be anisotropic: this drives a whole zoo of instabilities, altering the picture of Alfvénic turbulence based on the RMHD equations. interestingly, this is the case in many astrophysical environments, e.g. the intracluster medium, accretion disks, and (some of) the solar wind.

how plasma gets so hot

the solar coronal plasma is orders of magnitude hotter than the photosphere: we think this is because of turbulence (or reconnection, according to some). what is difficult is to explain why the ions are heated: because of the anisotropic cascade, the fluctuations were expected to be unable to access the cyclotron resonance: the helicity barrier (see above) may be a way out. recently, Parker Solar Probe observations have allowed us to make serious progress in how this heating occurs.

switchbacks

one of Parker Solar Probe's most interesting discoveries was that the magnetic field near the Sun is dominated by sharp, quasi-discontinuous rotations of the magnetic field, dubbed switchbacks since they can even reverse its direction completely. whether this should have been surprising is unclear (see Goldstein et al. 1974 below). nevertheless, this has spawned a frenzy of theoretical explanations. my own point of view: the switchbacks are simply large-amplitude Alfvén waves (Alfvén tsunamis?), that grow in normalized amplitude due to the gradients in the background as they propagate away from the Sun, in the same way that water waves steepen and break as they approach the beach.

magnetic reconnection: another 50-year problem

without dissipation, in MHD the magnetic field lines are frozen into the flow. with resistivity (or kinetic effects!) added, they can "reconnect", driving heating and fast flows: the classic Sweet–Parker theory, however, predicts rates that are far too slow. this can be resolved either with kinetic effects (Drake et al.) or by taking the tearing instability of the sheet into account (Loureiro's plasmoid instability).

turbulent dynamo

everything else on this page assumes a strong background magnetic field. where did it come from? turbulence amplifies a seed field exponentially, which is a different problem with a different literature — and in a collisionless plasma it is a genuinely open one. outside my area of expertise, but here is a review and a couple of recent papers on the collisionless problem.

turbulence, everywhere

a few more examples of turbulence throughout the universe.

disclaimer: simulations lie

a caveat on all numerical simulations. we are limited in the length of the inertial range by the resolution available on our machines; even with massively parallelized codes, this is still far smaller than even the solar wind inertial range, let alone the interstellar medium! this means we cannot expect to rigorously get the intermittency and heating correct: we have to use our intuition and experience to argue for the broader applicability of our tools and theories. this causes serious arguments that last decades (see dynamic alignment, for example). moreover, we also use highly simplified models (like RMHD), with simplified driving, and simplified dissipation. this is a blessing and a curse: we can switch effects on and off with ease, but we have to be careful not to over-interpret our results: the test is always ultimately in nature.

Every link on this page is external and will open in a new tab. If one has rotted, the DOI is in the URL and will usually still resolve.