2025-09-09
11:36 |
MEMFlow in Double Higgs search application
/CMS Collaboration
The Matrix Element Method (MEM) offers optimal statistical power for hypothesis testing in particle physics, but its application is hindered by the computationally intensive multidimensional integrals required to model detector effects. We present a novel approach that addresses this challenge by employing Transformers and generative machine learning (ML) models. [...]
CMS-DP-2025-056; CERN-CMS-DP-2025-056.-
Geneva : CERN, 2025 - 57 p.
|
Notice détaillée
|
2025-09-09
11:36 |
Notice détaillée
|
2025-09-09
11:36 |
Validation of online GPU vs. CPU reconstruction in Pixel, ECAL, HCAL, and PF
/CMS Collaboration
Since the beginning of Run 3, the CMS experiment has implemented a heterogeneous scheme for the High-Level Trigger (HLT) farm that integrates NVIDIA GPU accelerators for online event reconstruction. This note reports on the validation of GPU-based reconstruction against CPU for pixel, ECAL, and HCAL local reconstruction, as well as for pixel track and vertex reconstruction and the clustering stage of Particle Flow (PF) reconstruction. [...]
CMS-DP-2025-054; CERN-CMS-DP-2025-054.-
Geneva : CERN, 2025 - 11 p.
Fulltext: PDF;
|
Notice détaillée
|
2025-09-04
09:55 |
From bins to flows: unbinned and multivariate scale factors
/CMS Collaboration
Most CMS objects, taggers, and triggers are calibrated using likelihood fits.
A widely used technique for measuring efficiencies, particularly for electrons and muons, is the
Tag-and-Probe method. This approach exploits the clean signature of $Z\rightarrow l^{+}l^{-}$ decays: one
lepton is required to pass stringent identification and trigger criteria (the tag), while the other (the
probe) is used to study the efficiency of a given selection in both data and Monte Carlo (MC)
simulation. [...]
CMS-DP-2025-053; CERN-CMS-DP-2025-053.-
Geneva : CERN, 2025 - 25 p.
Fulltext: PDF;
|
Notice détaillée
|
2025-09-04
09:55 |
Calibration of the jet mass scale using boosted W bosons and top quarks for data taken in 2022
/CMS Collaboration
This note reports correction factors, to be applied to simulation, for soft drop jet mass for the identification of hadronically decaying top quarks or W bosons. These scale factors are calculated from and are valid for the proton--proton collision data set at $\sqrt{s} = 13.6$ TeV from the 2022 data-taking of the CMS experiment, corresponding to a total integrated luminosity of $34.65 fb ^{#-1}$..
CMS-DP-2025-052; CERN-CMS-DP-2025-052.-
Geneva : CERN, 2025 - 37 p.
Fulltext: PDF;
|
Notice détaillée
|
2025-09-01
11:27 |
Notice détaillée
|
2025-09-01
11:27 |
Debiasing Ultrafast Anomaly Detection with Posterior Agreement
/CMS Collaboration
The Level-1 Trigger system of the CMS experiment at CERN makes the final decision on which LHC collision data are stored to disk for later analysis. One algorithm used with this scope is an anomaly detection model based on an autoencoder architecture. [...]
CMS-DP-2025-050; CERN-CMS-DP-2025-050.-
Geneva : CERN, 2025 - 40 p.
Fulltext: PDF;
|
Notice détaillée
|
2025-09-01
11:27 |
Notice détaillée
|
2025-09-01
11:27 |
Notice détaillée
|
2025-09-01
11:27 |
Performance of boosted tau lepton identification with DeepTau Framework (Boosted DeepTau)
/CMS Collaboration
This note presents a dedicated identification algorithm targeting individual hadronic tau leptons ($\tau_\mathrm{h}$) within boosted ditau systems. Based on the DeepTau architecture used for resolved $\tau_\mathrm{h}$, the Boosted DeepTau algorithm achieves a factor of 2--4 improvement in the rejection of jets for individual $\tau_\mathrm{h}$ candidates with $p_\mathrm{T}$ $<$ 100 GeV, and an order of magnitude improvement at higher $p_\mathrm{T}$. [...]
CMS-DP-2025-047; CERN-CMS-DP-2025-047.-
Geneva : CERN, 2025 - 15 p.
Fulltext: PDF;
|
Notice détaillée
|