30th Jan 2012: We (BG, AJB, CGL) regret to annouce that a bug was found today in the way the statistical benefits of mTStar and mTBound were evaluated in the two 2011 papers that applied them to the Higgs to WW problem (arXiv:1106.2322 and arXiv:1108.3468). We are in the process of providing a full correction / statement, however in the interim it is fair to say that the conclusion for mTStar almost certainly changes from "mTStar is better than mTTrue for higgs masses between mW and 2 mW" to "mTStar is neither better nor worse than mTTrue for any higgs mass". The same mistake is also present in the mTBound paper, and also reduces the gain which mTBound has over mTTrue (though initial investigations suggest that it is still an improvement, albeit a smaller one).
On the bright side: the kinematics described in the papers all stands, the mathematical and physics properties of the bounding variables described in both papers remains correct, and the code (below) implementing the variables is OK. The problem is confined to the statistical calculation of the benefits provided by these variables.
We are in the process of attempting to contact any persons we know who may be using these variables to apologise for this mistake, and hope to have the arxiv papers corrected as soon as possible. We extend our apologies to those who may have been inconvenienced by this error. [The rest of the text of this web-page remains unaltered from its state prior to the discovery of this bug.]
| mTBound | : | arXiv:1106.2322 |
| mTStar | : | arXiv:1108.3468 |
| mTTrue | : | arXiv:0902.4864 (mTTrue) and arXiv:1105.2977 (M1T) |
If you wish to compute any of the above variables, you will need to download the appropriate header file(s). The library consists of header files only, and does not need to be built prior to use. It is sufficient top copy the relevant header file into your code. There are no external dependencies but you will have to make one simple change to FlatRandom.h before you can compute mTBound or mTStar! See instructions below!
The files you will need for each variable are as follows:
| mTBound | : | mTBound.h | and | FlatRandom.h |
| mTStar | : | mTStar.h | and | FlatRandom.h |
| mTTrue | : | mTTrue.h | only. |
You must replace the implementation of the function double FlatRandom() with one that uses your preferred source of high quality random numbers distributed uniformly between 0 and 1. Our reference implementation uses CLHEP::RandFlat::shoot(), but you could use many other possibilities, for example TRandom::Uniform() from ROOT's math core. We factored out our random number dependency in this way as different users have different requirements regarding reproducability.
Documentation explaining how to use the implementations may be found in large C++ comments at the top of each header. Please klick on the links to the headers above to read this documentation.