Adding a new measurement¶
If a new measurement already exists in your installed version Rivet and you want to make use of it in Contur, you
“simply” need to add it to the Contur database, which can be found under your local contur directory
This is an sqlite database compiled when you type
make in your
local contur directory.
The changes you’ll need to make are:
Check if the new measurement fits in an existing analysis pool. If it does not, then you need to create a new analysis pool with an appropriate name, and add it to the
analysis_pooltable with a line like:
INSERT INTO analysis_pool VALUES('ATLAS_7_JETS','7TeV','Inclusive hadronic final states');. The first entry is the name, the second the beam descriptor, the third a brief text description of the pool.
Once you have done this, or have identified an existing pool in which the analysis belongs, add the analysis to the
analysistable with a line like:
INSERT INTO analysis VALUES('ATLAS_2014_I1325553',4500.0,'ATLAS_7_JETS');where the first field is the Rivet analysis name, the second is the integrated luminosity (in the same units in which the plots in the paper are displayed) and the third is the analysis pool name.
For simple cases, that’s all! Next time you type
make, the database will be remade and the
.ana files for steering rivet (see the running instructions) etc will be remade to include your new data. If you use those then next time you run Herwig/Rivet/Contur, you should see comparisons to your new data.
If the plots in the paper are area normalised, Contur needs to know the absolute normalisation. This should be provided in the
analysis.sql, for example:
INSERT INTO normalization VALUES('ATLAS_2012_I1203852','(d03|d05|d07)',0.0254,1);. The first field is the analysis name, the second a regular expression, the third is the integrated cross section for a plot matching the regexp, and the fourth is a flag saying whether the MC plot is normalised by Rivet or not (should usually be 1).
If there are a few plots you don’t want to use in the comparison, add them to the
blacklisttable. They will be ignored by contur.
Alternatively, if there are only a few plots you do want to use in the comparison, add them to the
whitelisttable. All other plots from the analysis will be ignored by Contur.
If there are several statistically independent plots in your analysis, you can add them to a subpool in the
subpooltable to maximise sensitivity.
If the analysis is a search analysis, add it to the
Check the other tables:
bveto. These flag up various special cases. If you just want to run locally you can probably ignore them, but if you want to contribute your new work to the Contur repository (encouraged!) please check them.
If you have a new local Rivet routine you want to use then just add it (the usual
.cc, .yoda, .plot and
.info files to the
modified_analyses directory of your local Contur installation, and do all the above steps in the database as well. Next time you
make, the new routne should be compiled along with the database.