Manfred Peter Fackeldey (8e180f32) at 14 Apr 15:46
add task to compute frequentist toys
Manfred Peter Fackeldey (b4413b50) at 06 Apr 10:56
hack for 2D plots for PAS
Hi @acarvalh!
Concerning many datacard manipulations in parallel, you could parallelise this with bash by sending the subprocesses in the loop in the background, e.g. something like: for i in 1 2 3; do echo $i & done
. Although this is not the best solution, it might very well speed up your manipulations as you can run some in parallel...
Not sure if this helps and/or is applicable for you, but maybe give it a try?
Best, Peter
Is there a good reason to manipulate combine outputs by hand? A reasonable file format is nice, but it might not scale for e.g. 2D tasks (just too many values). In addition, I don't think there is a compression option for e.g. CSV. In our (bbWW) analysis we produce so many outputs, that this might be a non-negligible factor as storage is sometimes limited (this statement is very hand wavy, I don't have numbers).
Manfred Peter Fackeldey (662c8329) at 25 Mar 11:13
awesome
Manfred Peter Fackeldey (63083bae) at 24 Mar 14:49
In addition, one could now end up by choosing quintic
interpolation via CLI while using unstructured data (now an Error is/has to be raise). You have to know in advance which interpolation kind is a valid choice, which not everyone knows. Personally, I'd prefer interpolation choices which work with any set of points (unstructured and structured).
I would say the choices for the interpolation kind (for the GridDataInterpolator
) is already more than enough for a user to make (maybe even via a Parameter if that's not yet possible). As far as I can see only the quintic
interpolation kind is not supported then, but I doubt that anyone ever used that one (or will use it in the future).
What do you think?
Very nice!
Maybe as a side note: Wouldn't it be possible to use the GridDataInterpolator
for structured and unstructured data?
Alright. As long as there are no funny interplays between them if you request multiple styles, I am good with that
that's true. You have to be careful not to manipulate it. But as long as one has this in mind I would argue to keep the type of parameters consistent is more worth. But that's a philosophical question... I like it how it is :)
I saw often timming
instead of timing
... typo was at a different place. nvm
ah right. ok... unused to this in case there is only one argument. But alright
Manfred Peter Fackeldey (b91e500a) at 20 Mar 08:57
enable --summary for 2d exclusion plots