TAMM Simulated di-Higgs Dataset

Simulated signal (di-Higgs to 4b) and background (other 4b) jets used in the paper "Many Wrongs Make a Right: Leveraging Biased Simulations Towards Unbiased Parameter Inference". The "_sd" files correspond to the TDs described in the paper, and the "_ssd" files to zipped archives containing the 500 MSD populations.

The jets are generated using ${\tt MadGraph5\_aMC@NLO}$, and Higgs decay is performed with ${\tt MadSpin}$. We use ${\tt Pythia~8}$ for parton showering and hadronization, followed by ${\tt Delphes 3}$ for fast detector simulation. We use a modified CMS card for the detector simulation, reconstructing jets with the anti-$k_T$ algorithm with $R=0.8$ and demanding $p_{T_j} > 8\;{\rm GeV}$.

The background generation is biased to enhance efficiency by requiring greater than a $90$ GeV mass for each $b \bar{b}$ pair.

Each of the jets is then subjected to further kinematic cuts, $p_T>25$ GeV and $|\eta|<2.5$, and events with at least four surviving jets are selected. Higgs candidate dijets are then constructed by minimizing the metric

$\chi^2 = \frac{(m_1-125\,\mathrm{GeV})^2}{\sigma^2} + \frac{(m_2-125\,\mathrm{GeV})^2}{\sigma^2}$

as a function of the dijet masses $m_1$ and $m_2$. The event is accepted if both dijet masses are within $[100,150]$ GeV, and the dataset consists of these dijet masses for each accepted event.

The TD and MSD datasets differ through the Jet Energy Scale (JES) in the detector simulation, with the TD setting the CMS card JES parameters to their default values and the MSDs modifying these parameters in the manner described in the article.

The code used to produce the datasets can be found in the GitHub repository associated with the paper. The file results.tar.gz are results for Bᴀʏᴇꜱɪᴀɴ Tᴏᴘɪᴄ Mᴏᴅᴇʟɪɴɢ, to be used with the code in the GitHub repository.