add2W and bcg auxcode

[Gary Sharpe]

For EulerEquations or ReactiveEulerEquations
can one have more than one add2W statement,
i.e. is it possible to add more than one new variable to
the solution vector?
The example i have in mind is this,
currently I have something along the lines
ReactiveEulerEquations {
space=2D
add2W=PVORT
}
and I have a procedure which after n timesteps plots the PVORT field
e.g. with statements like
minmax W::PVORT[] -> minmax.
Now I want an additonal field, NVORT say so I tried
ReactiveEulerEquations {
space=2D
add2W=PVORT
add2W=NVORT
}
without any other changes to the code
and get:
expression 'W::PVORT[]' is undefined!

Line 13 is:
minmax W::PVORT[] -> min, max

error near:
W::PVORT[] -> min, max
I am assuming the problem may be it only
accepts the last add2W statement?

On a similar vein, can one have more than one
auxcode in BCG.
e.g. I currently have
BasicCodeGenerator {
scheme = roe'$znd::model
solver = 1step
auxcode = code/pvort_trace
}
where the code pvort_trace is the routine which
updates the W::PVORT variable field.
If I wanted to evaluate a 2nd variable field
(NVORT say) is it acceptable to daisy chain
a second piece of code
BasicCodeGenerator {
scheme = roe'$znd::model
solver = 1step
auxcode = code/pvort_trace
auxcode = code/nvort_trace
}
say, or does one of the auxcode statements
just overwrite the other?

Any advice from anyone who has looked at
this sort of thing before would be appreciated...

Best Wishes
Gary

[James Quirk]

On Fri, 16 May 2008, Gary Sharpe wrote:

>
> For EulerEquations or ReactiveEulerEquations
> can one have more than one add2W statement,
> i.e. is it possible to add more than one new variable to
> the solution vector?
> The example i have in mind is this,
> currently I have something along the lines
> ReactiveEulerEquations {
> space=2D
> add2W=PVORT
> }
> and I have a procedure which after n timesteps plots the PVORT field
> e.g. with statements like
> minmax W::PVORT[] -> minmax.
> Now I want an additonal field, NVORT say so I tried
> ReactiveEulerEquations {
> space=2D
> add2W=PVORT
> add2W=NVORT
> }
> without any other changes to the code
> and get:
> expression 'W::PVORT[]' is undefined!

The above won't work as add2W is a procedure parameter
and so the second entry simply overwrites the first.
What you need to do is write:

ReactiveEulerEquations {
space = 2D
add2W = PVORT,NVORT
}

And if you take a look at:

$AMRITA/stdlib/equations/Add2W.amr

you'll see that it loops over the supplied parameter
adding named entries to the solution vector.

>
> On a similar vein, can one have more than one
> auxcode in BCG.
> e.g. I currently have
> BasicCodeGenerator {
> scheme = roe'$znd::model
> solver = 1step
> auxcode = code/pvort_trace
> }
> where the code pvort_trace is the routine which
> updates the W::PVORT variable field.
> If I wanted to evaluate a 2nd variable field
> (NVORT say) is it acceptable to daisy chain
> a second piece of code
> BasicCodeGenerator {
> scheme = roe'$znd::model
> solver = 1step
> auxcode = code/pvort_trace
> auxcode = code/nvort_trace
> }
> say, or does one of the auxcode statements
> just overwrite the other?

Again you could answer this question by delving into
the bowls of BCG. As before, the second auxcode
will overwrite the first. And if you take a look at:

$AMRITA/plugin/amr_sol/BCG/equations/ReactiveEulerEquations/AddAuxiliaryCode.amr

you'll see that it loops over supplied filenames and
so you could write:

BasicCodeGenerator {
scheme = roe'1a
solver = 1step
auxcode = code/pvort_trace,code/nvort_trace
}

James

[Gary Sharpe]

James,
ah, Isee...I assumed it was overwriting somehow. The correct way
to do it is, in hindsight, obvious.
> $AMRITA/stdlib/equations/Add2W.amr
> $AMRITA/plugin/amr_sol/BCG/equations/ReactiveEulerEquations/AddAuxiliaryCode.amr
Thank, thats useful to know where these are (again obvious from
directory structure).
Gary