MadVfold: accelerating NLO event generation and reducing negative weights with SIMD vectorization and GPUs
Year: 2026
Type: preprint
Abstract: NLO simulations are essential for LHC physics analyses but are expensive, as they are not only slow but also lead to negative weights, which imply the need to simulate much larger samples of events. Folding is a powerful technique to reduce negative weights but is itself expensive. In this paper I propose ``vectorized folding'' as a new idea to speed up these calculations using SIMD and GPUs, and I present its CUDACPP-based implementation for MG5aMC in MadVfold, including its extension for unfolded NLO event generation. Preliminary results show overall speedups around 6x to 9x with folding and 3x without it. This work is based on a test-centric, LLM-assisted software development process.
Source: arXiv (Cornell University)
Author: Valassi, Andrea
Institution: European Organization for Nuclear Research
Subfield: Hardware and Architecture
Field: Computer Science
Domain: Physical Sciences
Open Access status: green
OpenAlex ID: W7214045460
Cites:
Cited by:
Related to: