[Users] CarpetIOScalar: avoiding "reductions" for scalar variables

[Erik Schnetter]

There is only one group, called "test_scalars". It is the variables
that become arrays, and all variables are written into the same file:

test_x[0] test_x[1] test_x[2] test_y[0] test_y[1] test_y[2] test_z[0]
test_z[1] test_z[2]

-erik

On Wed, Jan 5, 2011 at 9:52 PM, Kelly, Bernard J.
(GSFC-660.0)[UNIVERSITY OF MARYLAND BALTIMORE COUNTY]
<bernard.j.kelly at nasa.gov> wrote:
> Thanks, Erik.
>
> I looked at CarpetIOASCII 0D before, but thought it couldn't be the right
> thing, as it outputs many columns of meaningless information (I.e.,
> columns 3 through 12 of the general IOASCII output, including levels,
> coordinate indices, and coordinate values). The IOScalar output has the
> advantage of being smaller & easier to read. Apart from the reduction
> issue, it's not clear to me that it's a worse choice than IOASCII/0D.
>
> Follow-on question, which applies to both thorns: I actually have three
> scalar variables -- the components of a 3-vector. Ideally, I'd like to
> output them in a single file. This works OK, as they're in a group, and
> I've activated the "one_file_per_group" option.
>
> But ... I actually have one of these scalar groups per apparent horizon.
> Here's the declaration in my interface.ccl:
>
> REAL test_scalars[num_horizons] TYPE=scalar
> {
>  test_x
>  test_y
>  test_z
> } "Testing horizon scalar"
>
> When I request scalar output of test_x[0], I get a file with test_x[0] in
> its name.
>
> When I request scalar output of the group test_scalars[0], I get a file
> with test_scalars in its name -- no "[0]".
>
> Do you know why the horizon index is dropped for group output? If I
> request scalar output of the group test_scalars[0] -and- of
> test_scalars[1], what will happen? This seems like a bug to me.
>
> Bernard
>
> On 1/5/11 3:56 PM, "Erik Schnetter" <schnetter at cct.lsu.edu> wrote:
>
>>Bernard
>>
>>CarpetIOScalar is for outputting reduction results; this does not
>>really apply to scalar variables. You can output scalar variables
>>without reduction using CarpetIOASCII; this is called "0D" output
>>there since scalars are 0-dimensional grid variables (this is also how
>>they are implemented internally).
>>
>>out0D_vars = "..."
>>out0D_every = ...
>>
>>You can also use 0D output to output multi-dimensional grid variables
>>at a certain point, e.g. the origin.
>>
>>-erik
>>
>>On Wed, Jan 5, 2011 at 7:28 PM, Kelly, Bernard J.
>>(GSFC-660.0)[UNIVERSITY OF MARYLAND BALTIMORE COUNTY]
>><bernard.j.kelly at nasa.gov> wrote:
>>> Hi all. This seems trivial, but ...
>>> How do I avoid reductions (average, minimum, maximum, etc) for scalar
>>>output
>>> of a SCALAR variable in CarpetIOScalar?
>>> CactusBase/IOBasic recognises the SCALAR type of the variable, and
>>>doesn't
>>> bother with reduction.
>>> Carpet/CarpetIOScalar seems to ignore the type, and does the
>>>reductions. At
>>> least, it outputs several files with trivial information in them. I can
>>> force it to only generate one file by setting outscalar_reductions
>>> ="average" (for instance), but then I lose the reductions for all other
>>> (non-SCALAR) variables.
>>> Thanks,
>>> Bernard [working with the Chandrasekhar release of ETK]
>>> _______________________________________________
>>> Users mailing list
>>> Users at einsteintoolkit.org
>>> http://lists.einsteintoolkit.org/mailman/listinfo/users
>>>
>>>
>>
>>
>>
>>--
>>Erik Schnetter <schnetter at cct.lsu.edu>   http://www.cct.lsu.edu/~eschnett/
>
> _______________________________________________
> Users mailing list
> Users at einsteintoolkit.org
> http://lists.einsteintoolkit.org/mailman/listinfo/users
>



-- 
Erik Schnetter <schnetter at cct.lsu.edu>   http://www.cct.lsu.edu/~eschnett/