Closed geriJana closed 2 months ago
Thanks @geriJana. This mostly affects source/init/mus_initFluid_module.f90, I think.
To handle several variants of the scheme relaxations, we have decided to extend the relaxation option in scheme identify table as a table with additional variant
option.
identify = {
kind = 'fluid',
relaxation = {
name = 'bgk',
variant = 'improved'
},
layout = 'd3q27'
}
If that variant requires addditional parameters then they should be inside the relaxation table like
identify = {
kind = 'fluid',
relaxation = {
name = 'bgk',
variant = 'hybrid_recursive_regularized',
regularization_omega=0.98
},
layout = 'd3q27'
}
Standard variant which is a default can be defined by any of these forms:
relaxation='bgk'
relaxation={ name = 'bgk'}
relaxation={name='bgk', variant='standard'}
Here is the new naming convection of the kernels.
mus_advRel_k<kind>_r<relaxation>_v<variant>_l<layout>
Examples:
For non-specific implementation leave names out. So we do not use generic keyword in the kernel names anymore. Examples:
Here is the new naming convection of the kernels.
mus_advRel_k<kind>_r<relaxation>_v<variant>_l<layout>
Examples:* mus_advRel_kFluid_rBGK_v_ld3q19 * mus_advRel_kFluidIncomp_rBGK_ld3q19 * mus_advRel_kFluid_rBGK_vHRR_ld3q19
mus_advRel_kFluidIncomp_rBGK_ld3q19
does not seem to follow the naming convention outlined above (doesn't have a _v
part). Is this on purpose to show that some parts might be left out? Or is this a typo?
mus_advRel_kFluidIncomp_rBGK_ld3q19
does not seem to follow the naming convention outlined above (doesn't have a_v
part). Is this on purpose to show that some parts might be left out? Or is this a typo?
Yes, it was a typo. Corrected it now.
@geriJana I believe we can close this issue with the merge of #14. Do you open points from this issue that still need to be addressed?
Thanks! We can close this issue now.
It seems that inconsistencies have been introduced into the kernel naming convention of the routines as well as some configuration options (see also #9).
In general
E.g. what does ... mean or when to use ...
Decide on
Correction routines
Besides, how should we name the correction routine _bgkHybridRecursiveRegularizedCorr ? It only differs in the blending parameter sigma but is now called for different configure options as listed below:
Suggestion:
bgk_blendedRecursiveRegularizedCorrected
Please note also the change of word ordering and removed trailing capital later in routine name.Decide on
Control
Once the naming convention is set for the routine names as well as the configuration options, we should check all routines for discrepancies and to identify need for renaming.
Harald asked @KannanMasilamani to decide on the names.