FAQ

What is $k_y{\rho }_s$ and why does CGYRO modify ${\rho }_s$ ?

The fundamental definition of the wavenumber and unit gyroradius are given in [CBB16]. These are

$$\begin{array}{r}\begin{array}{rl}{k}_y\doteq & nq/r,\\ {\rho }_s\doteq & {\rho }_{s,\mathrm{u}\mathrm{n}\mathrm{i}\mathrm{t}}=\frac{e{B}_{\mathrm{u}\mathrm{n}\mathrm{i}\mathrm{t}}}{m_{D}c},\\ B_{\mathrm{u}\mathrm{n}\mathrm{i}\mathrm{t}}\doteq & {\displaystyle \frac{q}{r}\frac{\partial \psi }{\partial r}.}\end{array}\end{array}$$

Here, $B_{\mathrm{u}\mathrm{n}\mathrm{i}\mathrm{t}}$ is the Waltz effective magnetic field which is standard across all GACODE tools/codes, and $r$ is the midplane minor radius. For a given value of $k_y{\rho }_s$, the gyrokinetic equations are invariant to the scaling $n\to \alpha n$ and ${\rho }_s\to {\rho }_s/\alpha$, where $\alpha$ is an arbitrary scaling parameter.

The CGYRO input parameter KY is

$$\mathtt{KY}\doteq \mathrm{\Delta }(k_y{\rho }_s)=\mathrm{\Delta }n\left(\frac{q}{r}\right){\rho }_s.$$

CGYRO enforces $\mathrm{\Delta }n=1$, which sets the (artificial) value of ${\rho }_s$ to

$${\rho }_s\to \left(\frac{r}{q}\right)\mathtt{KY}.$$

To see the physical values of key parameters, you can use profiles_gen:

$ profiles_gen -i input.gacode -loc_rad 0.6
INFO: (profiles_gen) input.gacode is autodetected as GACODE.
INFO: (locpargen) Quasineutrality NOT enforced.
INFO: (locpargen) rhos/a   =+2.44577E-03
INFO: (locpargen) Te [keV] =+8.81550E-01
INFO: (locpargen) Ti [keV] =+7.58644E-01
INFO: (locpargen) Bunit    =+2.92020E+00
INFO: (locpargen) beta_*   =+4.84003E-03
INFO: ----->  n=1: ky*rhos =-7.74754E-03
INFO: (locpargen) Wrote input.*.locpargen

How does adaptive time-stepping work?

Time-stepping is controlled with the parameter DELTA_T_METHOD. Setting the parameter to 0 gives the legacy fixed timestep, whereas values greater than 0 are adaptive methods. For the adaptive methods we recommend setting DELTA_T = 0.01. Here is the recommendation:

DELTA_T_METHOD=1
DELTA_T=0.01
PRINT_STEP=100

The overall time-integration step is split between an explicit high-order step, and an implicit second-order step for collisions and trapping. When using the adaptive method, the value of DELTA_T is the size of the (large) implicit timestep. Then, the value of the explicit timestep is decreased to match the error tolerance, ERROR_TOL – the default value of which should be sufficient.