Nathanaël Schaeffer

Efficient spherical harmonic transforms aimed at pseudospectral numerical simulations

In this paper, we report on very efficient algorithms for the spherical harmonic transform (SHT). Explicitly vectorized variations of the algorithm based on the Gauss-Legendre quadrature are discussed and implemented in the SHTns library which includes scalar and vector transforms. The main breakthrough is to achieve very efficient on-the-fly computations of the Legendre associated functions, even for very high resolutions, by taking advantage of the specific properties of the SHT and the advanced capabilities of current and future computers. This allows us to simultaneously and significantly reduce memory usage and computation time of the SHT. We measure the performance and accuracy of our algorithms. Even though the complexity of the algorithms implemented in SHTns are in

Turbulent geodynamo simulations: a leap towards Earth’s core

O(N^3)

Stability of a pair of co-rotating vortices with axial flow

(where N is the maximum harmonic degree of the transform), they perform much better than any third party implementation, including lower complexity algorithms, even for truncations as high as N=1023. SHTns is available at this https URL as open source software.

Performance benchmarks for a next generation numerical dynamo model

We present an attempt to reach realistic turbulent regime in direct numerical simulations of the geodynamo. We rely on a sequence of three convection-driven simulations in a rapidly rotating spherical shell. The most extreme case reaches towards the Earths core regime by lowering viscosity (magnetic Prandtl number Pm=0.1) while maintaining vigorous convection (magnetic Reynolds number Rm>500) and rapid rotation (Ekman number E=1e-7), at the limit of what is feasible on todays supercomputers. A detailed and comprehensive analysis highlights several key features matching geomagnetic observations or dynamo theory predictions – all present together in the same simulation – but it also unveils interesting insights relevant for Earths core dynamics. In this strong-field, dipole-dominated dynamo simulation, the magnetic energy is one order of magnitude larger than the kinetic energy. The spatial distribution of magnetic intensity is highly heterogeneous, and a stark dynamical contrast exists between the interior and the exterior of the tangent cylinder (the cylinder parallel to the axis of rotation that circumscribes the inner core). In the interior, the magnetic field is strongest, and is associated with a vigorous twisted polar vortex, whose dynamics may occasionally lead to the formation of a reverse polar flux patch at the surface of the shell. Furthermore, the strong magnetic field also allows accumulation of light material within the tangent cylinder, leading to stable stratification there. Torsional Alfven waves are frequently triggered in the vicinity of the tangent cylinder and propagate towards the equator. Outside the tangent cylinder, the magnetic field inhibits the growth of zonal winds and the kinetic energy is mostly non-zonal. Spatio-temporal analysis indicates that the low-frequency, non-zonal flow is quite geostrophic (columnar) and predominantly large-scale: an m=1 eddy spontaneously emerges in our most extreme simulations, without any heterogeneous boundary forcing. Our spatio-temporal analysis further reveals that (i) the low-frequency, large-scale flow is governed by a balance between Coriolis and buoyancy forces – magnetic field and flow tend to align, minimizing the Lorentz force; (ii) the high-frequency flow obeys a balance between magnetic and Coriolis forces; (iii) the convective plumes mostly live at an intermediate scale, whose dynamics is driven by a 3-term 1 MAC balance – involving Coriolis, Lorentz and buoyancy forces. However, small-scale (E^{1/3}) quasi-geostrophic convection is still observed in the regions of low magnetic intensity.

On the reflection of Alfvén waves and its implication for Earth's core modelling

The three-dimensional linear temporal stability properties of a flow composed of two corotating q-vortices (also called Batchelor vortices) are predicted by numerical stability analysis. As for the corresponding counter-rotating case, when the axial flow parameter is increased, different instability modes are observed and identified as a combination of resonant Kelvin modes of azimuthal wavenumbers m and m+2 within each vortex. In particular, we show that the sinuous mode, which is the dominant instability mode without axial flow, is stabilized in the presence of a moderate axial flow. Different types of mode with a large amplitude in the critical layer are also identified. For small separation distances (above the merging threshold), unstable eigenmodes, corresponding to axial wavenumbers that cannot be easily identified with simple resonant interactions of Kelvin modes, are also observed. Their growth rate is a substantial fraction of the growth rates of low-order resonant modes. The effects of the Reyn...

Variability modes in core flows inverted from geomagnetic field models

Numerical simulations of the geodynamo have successfully represented many observable characteristics of the geomagnetic field, yielding insight into the fundamental processes that generate magnetic fields in the Earths core. Because of limited spatial resolution, however, the diffusivities in numerical dynamo models are much larger than those in the Earths core, and consequently, questions remain about how realistic these models are. The typical strategy used to address this issue has been to continue to increase the resolution of these quasi-laminar models with increasing computational resources, thus pushing them toward more realistic parameter regimes. We assess which methods are most promising for the next generation of supercomputers, which will offer access to O(106) processor cores for large problems. Here we report performance and accuracy benchmarks from 15 dynamo codes that employ a range of numerical and parallelization methods. Computational performance is assessed on the basis of weak and strong scaling behavior up to 16,384 processor cores. Extrapolations of our weak-scaling results indicate that dynamo codes that employ two-dimensional or three-dimensional domain decompositions can perform efficiently on up to ∼106 processor cores, paving the way for more realistic simulations in the next model generation.

Nonlinear dynamics of the elliptic instability

SUMMARY Alfven waves propagate in electrically conducting fluids in the presence of a magnetic field. Their reflection properties depend on the ratio between the kinematic viscosity and the magnetic diffusivity of the fluid, also known as the magnetic Prandtl number Pm. In the special case, Pm = 1, there is no reflection on an insulating, no-slip boundary, and the incoming wave energy is entirely dissipated in the boundary layer. We investigate the consequences of this remarkable behaviour for the numerical modelling of torsional Alfven waves (also known as torsional oscillations), which represent a special class of Alfven waves, in rapidly rotating spherical shells. They consist of geostrophic motions and are thought to exist in the fluid cores of planets with internal magnetic field. In the geophysical limit Pm ≪ 1, these waves are reflected at the core equator, but they are entirely absorbed for Pm = 1. Our numerical calculations show that the reflection coefficient at the equator of these waves remains below 0.2 for Pm ≥ 0.3, which is the range of values for which geodynamo numerical models operate. As a result, geodynamo models with no-slip boundary conditions cannot exhibit torsional oscillation normal modes.

Turbulence reduces magnetic diffusivity in a liquid sodium experiment.

The flow of liquid metal inside the Earths core produces the geomagnetic field and its time variations. Understanding the variability of those deep currents is crucial to improve the forecast of geomagnetic field variations, which affect human spacial and aeronautic activities. Moreover, it may provide relevant information on the core dynamics. The main goal of this study is to extract and characterize the leading variability modes of core flows over centennial periods, and to assess their statistical robustness. To this end, we use flows that we invert from two geomagnetic field models (gufm1 and COV-OBS), and apply Principal Component Analysis and Singular Value Decomposition of coupled fields. The quasi geostrophic (QG) flows inverted from both geomagnetic field models show similar features. However, COV-OBS has a less energetic mean and larger time variability. The statistical significance of flow components is tested from analyses performed on subareas of the whole domain. Bootstrapping methods are also used to extract significant flow features required by both gufm1 and COV-OBS. Three main empirical circulation modes emerge, simultaneously constrained by both geomagnetic field models and expected to be robust against the particular a priori used to build them (large scale QG dynamics). Mode 1 exhibits three large vortices at medium/high latitudes, with opposite circulation under the Atlantic and the Pacific hemispheres. Mode 2 interestingly accounts for most of the variations of the Earths core angular momentum. In this mode, the regions close to the tangent cylinder and to the equator are correlated, and oscillate with a period between 80 and 90 years. Each of these two modes is energetic enough to alter the mean flow, sometimes reinforcing the eccentric gyre, and other times breaking it up into smaller circulations. The three main circulation modes added together to the mean flow account for about 70% of the flows variability, 90% of the root mean square total velocities, and 95% of the secular variation induced by the total flows. Direct physical interpretation of the computed modes is not straightforward. Nonethe-less, similarities found between the two first modes and time/spatial features identified in different studies of core dynamics, suggest that our approach can help to pinpoint the relevant physical processes inside the core on centennial timescales.

A dynamo driven by zonal jets at the upper surface : applications to giant planets

In this paper, we analyse by numerical simulations the nonlinear dynamics of the elliptic instability in the configurations of a single strained vortex and a system of two counter-rotating vortices. We show that although a weakly nonlinear regime associated with a limit cycle is possible, the nonlinear evolution far from the instability threshold is, in general, much more catastrophic for the vortex. In both configurations, we put forward some evidence of a universal nonlinear transition involving shear layer formation and vortex loop ejection, leading to a strong alteration and attenuation of the vortex, and a rapid growth of the vortex core size.

Magnetic induction and diffusion mechanisms in a liquid sodium spherical Couette experiment.

The contribution of small scale turbulent fluctuations to the induction of a mean magnetic field is investigated in our liquid sodium spherical Couette experiment with an imposed magnetic field. An inversion technique is applied to a large number of measurements at Rm≈100 to obtain radial profiles of the α and β effects and maps of the mean flow. It appears that the small scale turbulent fluctuations can be modeled as a strong contribution to the magnetic diffusivity that is negative in the interior region and positive close to the outer shell. Direct numerical simulations of our experiment support these results. The lowering of the effective magnetic diffusivity by small scale fluctuations implies that turbulence can actually help to achieve self-generation of large scale magnetic fields.