Wiki

Clone wiki

gs2 / Namelists / collisions_knobs

collisions_knobs

Name Type Def CR Name Description
adjust Fortran_Bool .true. adjust
  • If adjust is .true., transform from the gyro-averaged dist. fn., g, to the non-Boltzmann part of delta f, h, when applying the collision operator
cfac Float 0.0,1.0 cfac
  • Factor multiplying FLR terms in collision operator. 1.0 by default. Set to 0.0 to turn off FLR corrections.
collision_model String default collision_model
  • Collision model used in the simulation.
    • default = pitch angle scattering and energy diffusion
    • collisionless,none = collisionless
    • lorentz = pitch angle scattering only
    • ediffuse = energy diffusion only
    • krook = use home made krook operator (no reason to use this!)
conservative Fortran_Bool .true. conservative
  • Set to .true. to guarantee exact conservation properties.
conserve_moments Fortran_Bool .true. conserve_moments
  • Set to .true. to guarantee collision operator conserves momentum and energy.
const_v Fortran_Bool .false. const_v
ediff_scheme String default ediff_scheme
ei_coll_only Fortran_Bool .false. ei_coll_only
etol Float 0.02 etol
etola Float 0.02 etola
ewindow Float 0.01 ewindow
ewindowa Float 0.01 ewindowa
heating Fortran_Bool .false. heating
  • Set to .true. to compute collisional heating.
hypermult Fortran_Bool .false. hypermult
lorentz_scheme String default lorentz_scheme
ncheck Integer 10,100 ncheck
resistivity Fortran_Bool .true. resistivity
special_wfb_lorentz Fortran_Bool .true. special_wfb_lorentz
  • If true (the default) then the wfb is treated in a special way in the lorentz collision operator. This is the standard behaviour. Setting to false has been seen to help an issue in linear flux tube simulations in which the zonal modes at large kx grow rapidly.
split_collisions Fortran_Bool .false. split_collisions
  • If true, then collisions are removed from the Kotschenreuther algorithm and evolved separately. This has the same formal accuracy as including the collisions in the Kotschenreuther advance, but requires fewer calls to the collisions module, and therefore fewer memory redistributes. This gives a speed up of between ~20% and ~60%, depending on processor count.
  • If false, the old method is used. This is the default, to prevent discrepancies with old results.
test Fortran_Bool .false. test_collisions
use_le_layout Fortran_Bool .true. use_le_layout
  • Use more efficient layouts for le grid
vary_vnew Fortran_Bool .false. vary_vnew
vnfac Float 1.1 vnfac
vnslow Float 0.9 vnslow

Updated