2026-05-20
13:21 |
GNN based Primary Vertex Reconstruction with Slot Attention
/CMS Collaboration
A new approach for primary vertex reconstruction using graph neural networks (GNNs) is explored for the Phase-2 upgrade of CMS. In the high-luminosity LHC era, the average number of simultaneous $pp$ interactions per bunch crossing is expected to reach up to 200 at $\sqrt{s} = 14$\,TeV, making vertex reconstruction a challenging clustering problem: $\sim$6000 tracks must be grouped into $\sim$200 vertices in 4D $(z, t)$ space.
The MIP Timing Detector (MTD) adds a crucial fourth dimension to this problem. [...]
CMS-DP-2026-042; CERN-CMS-DP-2026-042.-
Geneva : CERN, 2026 - 22 p.
Fulltext: PDF;
|
Details des Eintrags
|
2026-05-20
13:19 |
Measurement of the Lorentz Angle in the CMS Silicon Strip Tracker during Run 3
/CMS Collaboration
The CMS silicon strip tracker consists of inner barrel (TIB) layers, inner discs (TID), outer barrel
(TOB) layers, and end-caps (TEC), which close off the tracker on either end. It has been
successfully recording data during both Run 1 and Run 2 data taking at the Large Hadron Collider
(LHC). [...]
CMS-DP-2026-041; CERN-CMS-DP-2026-041.-
Geneva : CERN, 2026 - 19 p.
Fulltext: PDF;
|
Details des Eintrags
|
2026-05-20
13:18 |
Phase II CMS Outer Tracker Unpacker Port to alpaka
/CMS Collaboration
Modern high-performance computing centers increasingly opt for heterogeneous system designs, integrating
general-purpose computing cores with accelerators, to deliver high computing performance and high efficiency.
Scientific computing codes need to adapt to this change in cluster architecture, which introduces the challenge of
hardware portability. This concept has great relevance for the high-energy physics research at CERN, where
different computer accelerators from diverse vendors are used.
In the context of the phase-2 upgrade of the CMS Silicon Tracker [1], a new detector is being built, populated by
so-called Double-Strip (2S) and Pixel-Strip (PS) modules [...]
CMS-DP-2026-040; CERN-CMS-DP-2026-040.-
Geneva : CERN, 2026 - 8 p.
Fulltext: PDF;
|
Details des Eintrags
|
2026-05-20
13:15 |
Level-1 Trigger Performance in 2025 Proton-Proton Collisions at $\sqrt{s} =$ 13.6 TeV
/CMS Collaboration
In 2025, the CMS experiment collected a record-large data set of proton-proton collisions at $\sqrt{s} =$ 13.6 TeV, corresponding to an integrated luminosity of 110.73 $fb^{-1}$ certified as good for physics.
This note summarizes the performance of the CMS Level-1 Trigger system to identify and select muons, jets, transverse energy sums, electrons/photons, and hadronically decaying tau leptons in this data set..
CMS-DP-2026-039; CERN-CMS-DP-2026-039.-
Geneva : CERN, 2026 - 40 p.
Fulltext: PDF;
|
Details des Eintrags
|
2026-05-20
13:14 |
Details des Eintrags
|
2026-05-20
13:13 |
Details des Eintrags
|
2026-05-20
13:11 |
Neural network-based topological trigger for HH events in the CMS L1T during the 2026 pp data-taking
/CMS Collaboration
This note presents a novel machine learning-based trigger algorithm developed for the Level-1 Trigger (L1T) system of the CMS experiment. The algorithm, referred to as L1 TOPO, is a topological trigger based on a shallow neural network (NN), and it is designed to enhance the trigger acceptance, and thus the sensitivity, to the soft phase-space of Higgs boson pair (HH) production in the single-muon final state of the bbWW decay channel. [...]
CMS-DP-2026-036; CERN-CMS-DP-2026-036.-
Geneva : CERN, 2026 - 29 p.
Fulltext: PDF;
|
Details des Eintrags
|
2026-05-19
13:19 |
Full simulation performance for Run 3 and Run 4
/CMS Collaboration
In this note, several results of recent development for CMS simulation are presented. The historical trends in CPU performance for the CMS detector in Full and Fast simulation are shown from Run 2 to Run 3 and for the Run 4 configuration. [...]
CMS-DP-2026-035; CERN-CMS-DP-2026-035.-
Geneva : CERN, 2026 - 20 p.
Fulltext: PDF;
|
Details des Eintrags
|
2026-05-19
13:19 |
Details des Eintrags
|
2026-05-19
13:19 |
Towards the Usage of FlashSim End-to-End Simulation to Physics Analysis
/CMS Collaboration
FlashSim is a deep learning-based end-to-end simulation framework developed by the CMS Collaboration to provide fast, accurate, and analysis-agnostic simulation of NanoAOD-level variables from generator-level inputs. This note presents the recent advancements made to the FlashSim framework. [...]
CMS-DP-2026-033; CERN-CMS-DP-2026-033.-
Geneva : CERN, 2026 - 15 p.
Fulltext: PDF;
|
Details des Eintrags
|
|
|