Comparing Gyrokinetic Codes
Cross-verification and benchmarking of gyrokinetic codes builds confidence that turbulence predictions reflect physics rather than numerical artifacts.
Why compare codes
Turbulence simulation is complex enough that a single code's result is not self-validating. Different codes make different numerical choices, particle-in-cell versus continuum, local versus global, various collision operators, and can disagree. Systematic benchmarking against shared, precisely specified test cases is how the community separates real physics from method-dependent artifacts.
A benchmark specifies the geometry, profiles, gradients, species, and physics options exactly, then compares outputs: linear growth rates and real frequencies, nonlinear saturated fluxes, and fluctuation spectra.
Linear before nonlinear
Comparison starts with linear eigenvalues because they are cheap and unambiguous. If two codes agree on the growth rate and frequency of the dominant instability across a wavenumber scan, their linear physics and geometry handling match. Only then does nonlinear flux comparison make sense, since nonlinear results depend on saturation mechanisms that are harder to pin down.
Sources of disagreement
Typical culprits are the collision operator, the treatment of the parallel boundary, the electron model (kinetic versus adiabatic), and resolution. A disciplined benchmark isolates each by turning options on one at a time, which is a form of code verification.
Design relevance
Because the Hyperion breeder is a spherical tokamak where the local approximation is stressed, its turbulence predictions rest on agreement between independent codes rather than any single run. Cross-code agreement is part of the honest, simulation-based evidence assembled before construction.
- Shared test cases isolate method dependence
- Linear eigenvalues compared before nonlinear fluxes
- Collision operator and electron model are common culprits
- Cross-code agreement underpins turbulence claims