[CIG-MC] Questions about CitcomS input parameters

Thu Mar 10 11:23:36 PST 2016

Thorsten W Becker - USC
(twb at jsg.utexas.edu from 06/2016)

On Thu, Mar 10, 2016 at 4:16 AM, ELISA FIERRO <elisa.fierro at unicam.it>
wrote:

> Hi All,
> I'm an italian PhD student in Geodynamics and the topic of my PhD is
> modelling the rifting and rift to drift transition phase of the Red Sea by
> CitcomS 3.3.1. I'm new of this software (and of the geodynamic modelling in
> general) so I'm writing to you because I'm reading the manual but I'm not
> able to understand the meaning of a lot of input parameters and I would
> need a help.
>
> I'm sorry in advance for the large number of the questions!!!!
>
>
>
>    - side_sbcs=off/on  Do I have to set it on for a mantle convection
>    modelling? and after having set it on, do I need to set or change other
>    conditions related to it?
>
>
>
not for global, might for regional

>
>    -
>
>
>    - the* mgunit* parameters gathered in the mesh grid information (*mgunitx,
>    mgunity, mgunitz *and* levels*). What are they for?
>
>
>    -
>
>
check the manual, those are the core multigrid stencils which are refined
something levels times to something like 2^(levels-1)*mgunit

>
>    - I'm not able to understand how I have to set the *lith_age_files* (*lith.dat).
>    *I know that the format of these files has to be *'n age'*, but *n *is
>    for node? and if so, do I have to set an age for every node of the initial
>    input mesh (like for the vx and vy inside the bvel.dat)?
>
>
yes, check the source code where this gets read in. i've never used those.

>
>    - When lith_age parameter is on I have to add the followed additional
>    parameters for temperature boundary conditions:
>
>               temperature_bound_adj=off
>               depth_bound_adj=0.157
>               width_bound_adj=0.08727
>               lith_age_depth=0.0314
>
>            What is their meaning? When do I have to set the
> temperature_bound_adj off and on? Are the parameters depth_bound_adj and
> width_bound_adj
>
           dimensionless like lith_age_depth? and if so, which equations do
> I have to use to give them a dimension?
>

again, best to check the source code file if this is not in the manual

for example

grep -i lith_age_depth *.c

shows you that the variable is E->control.lith_age_depth internally, grep
that, find the file it's in, and go from there.

>
>    - A lot of Momentum Equation Solver Parameters:
>
>               node_assemble=on
>               mg_cycle=1
>               down_heavy=3
>               up_heavy=3
>               vlowstep=1000
>               vhighstep=3
>               max_mg_cycles=50
>               uzawa=cg
>               compress_iter_maxstep=100
>               precond=on
>               aug_lagr=on
>               aug_number=2.0e3
>
>             I didn't understand their meaning on the manual; What are V
> cycle and W cycle?; What is the downward/upward smoothing?; what is the
> preconditer? what is
>             the meaning of the abbreviations *cg* and *bicg* related to
> the uzawa algorithm?
>
>
those are multi grid solver parameters, you might want to leave those as
in the default for now, and perhaps increas the vhighstep. V and W are
names for the ways the mutligrid solver goes "up" and "down"

>
>    - Energy Equation Solver Parameters
>
>               fixed_timestep=0.0         If it is equal to 0, the time
> step is variable and it is determined dynamically. What does it mean?
>
>

it means that the courant criterion might not be fulfilled upon advection,
don't set it to 1

>               finetunedt=0.9                What is this parameter for? I
> didn't understand the explanation on the manual
>

a prefactor for the courant criterion to be on the safe side. if anything,
adjust down.

>
>               filter_temp=off/on           I didn't understand what the
> overshoots and undershoots of the temperature field are
>
>               monitor_max_T=on/off    I didn't understand when I have to
> set it on/off. The manual says that if it set on the maximum value of the
> current and previous
>                                                    temperature fields are
> compared but I don't understand the meaning of this sentence.
>
>
advection dominated problems or other instabilities can lead to overshoot
where the temperature goes outside the physical range, e.g. [0;1] in the
absence of internal heating if you have 0 and 1 temp BCs. you can use this
as a check for robustness

>               inputdiffusivity=1            diffusivity is for thermal
> diffusivity? In my case will do I have to change this parameter since my
> problem is integrated backward in
>
                                                   time? and how can I do
> it?
>

you can set it to something small but not zero (because of the algorithm)
if you want to backward advect

and hope things don't blow up.

>
>    -  Dimensional information
>
>              density=3340
>              density_above=1030.0
>              density_below=6600.0
>              thermexp=3.0e-5
>              refvisc=1.0e+21
>
>             What do these parameters refer to exactly?
>

again, check the source code by grepping. usually, dimensional values
aren't used internally (but there are exceptions), but they are used to
scale output.

>
>
> THANKS A LOT!!!
>
> Elisa Fierro
>
>
>
> _______________________________________________
> CIG-MC mailing list
> CIG-MC at geodynamics.org
> http://lists.geodynamics.org/cgi-bin/mailman/listinfo/cig-mc
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.geodynamics.org/pipermail/cig-mc/attachments/20160310/d0155b7e/attachment.html>