Abstract
We present MadFlow, a first general multi-purpose framework for Monte Carlo (MC) event simulation of particle physics processes designed to take full advantage of hardware accelerators, in particular, graphics processing units (GPUs). The automation process of generating all the required components for MC simulation of a generic physics process and its deployment on hardware accelerator is still a big challenge nowadays. In order to solve this challenge, we design a workflow and code library which provides to the user the possibility to simulate custom processes through the MadGraph5_aMC@NLO framework and a plugin for the generation and exporting of specialized code in a GPU-like format. The exported code includes analytic expressions for matrix elements and phase space. The simulation is performed using the VegasFlow and PDFFlow libraries which deploy automatically the full simulation on systems with different hardware acceleration capabilities, such as multi-threading CPU, single-GPU and multi-GPU setups. The package also provides an asynchronous unweighted events procedure to store simulation results. Crucially, although only Leading Order is automatized, the library provides all ingredients necessary to build full complex Monte Carlo simulators in a modern, extensible and maintainable way. We show simulation results at leading-order for multiple processes on different hardware configurations.
Highlights
Ators have been designed in order to simplify the implementation of algorithms and models in particular in the context of Artificial Intelligence applications
In order to write a competitive graphics processing units (GPUs)-capable full parton-level Monte Carlo (MC) by any measure with existing tools, there are at least five required ingredients: (i) an integrator, able to parallelize over the number of events; (ii) a GPU-capable parton distribution function (PDF) interpolation tool; (iii) an efficient phase space generator, which should generate valid phase space points on GPU, apply any fiducial cuts; (iv) evaluate the matrix element squared for the target processes; (v) an efficient asynchronous output storage system for observables, such as histograms and Les Houches event files
We have developed open-source tools that provide the ground basis for the implementation of an automatic Monte Carlo simulation framework for HEP addressing some of the aforementioned issues: VegasFlow [13,14] and PDFFlow [15,16]
Summary
If we consider the research domain of High Energy Physics, we can observe several examples of applications that could benefit from the conversion or systematic implementation of existing algorithms and code libraries on GPU. PDFFlow is a library which provides fast evaluation of parton distribution functions (PDFs) designed for platforms with hardware accelerators following the design idea inspired from VegasFlow. The availability of both packages completes respectively points (i) and (ii) above. We call MadFlow [18] the open-source software library which implements this automatic pipeline for GPU deployment of Monte Carlo simulation It combines the matrix elements expressions generated by the MadGraph5_aMC@NLO (MG5_aMC) [5,6] framework with the VegasFlow and PDFFlow efficient simulation tool for hardware accelerators.
Sign-in/Register to view highlights & summary Sign-in/Register to access full text options 
Free) 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call
