BSMArt

A simple to use yet powerful scanning tool for the SARAH family and beyond

BSMArt is maintained by Mark Goodsell and Miguel Crispim Romao in collaboration with:

  • Ari Joury
  • Asesh Datta
  • Luc Darmé
  • Johannes Braathen
  • Martin Gabelmann
  • Wojciech Kotlarski
  • Farid Ibrahimov
  • Fernando Abreu de Souza
  • Werner Porod

Wanted: your scans and tools! You can now contribute them on the community github page github.com/bsmart-hep/examples.

Online documentation is found at bsmart-hep.github.io/core/

The code for version 2 and later can be perused at the mirror site github.com/bsmart-hep/core while the latest version and also legacy version 1.7 are available here via Downloads.

Included tools

  • SARAH
  • SPheno
  • FlexibleSUSY
  • MicrOmegas
  • HiggsBounds
  • HiggsSignals
  • HiggsTools
  • anyBSM Thanks Martin Gabelmann and Johannes Braathen!
  • pass_tool -- a dummy tool that does nothing, but is useful for creating parameter cards and extracting information.
  • toy_targets -- standard likelihood functions (Rosenbrock, Rastrigin, etc) implemented in python for running file-free test scans of new algorithms.
  • Vevacious++ as VevaciousPlusPlus
  • flavio
  • MadGraph
  • SModelS
  • Resummino
  • ZPEED
  • Z prime explorer as Zprime
  • MadAnalysis, both for the PAD as MadAnalysisAllPAD and expert mode analyses as MadAnalysisExpert, with automatic event generation through MadGraph.
  • HackAnalysis as HackAnalysis_LO for direct event generation in pythia, and MadGraphHackAnalysis for integrated event generation in MadGraph followed by LHE or HEPMC event analysis.

Included scans:

Scans include, among others:

  • Random.
  • Grid, which can now also run in MPI mode (distributing points across clusters).
  • Directory read (read_dir) and an MPI version (read_dir_mpi)
  • Contour2D.
  • CSV read (read_csv), handy for rerunning over previously collected points!
  • MCMC. A simple Metropolis-Hastings algorithm, that runs on parallel cores.
  • AffineMC: an affine MCMC implementing the Goodman-Weare algorithm.
  • MultiNest.
  • Diver
  • Active Learning as AL (requires pytorch)
  • ContourGP, adapted from excursion by Heinrich, Louppe and Cranmer. Requires sklearn.
  • MLS based on the version from xBit, adapted with Farid Ibrahimov.
  • MLScanner family and DLScanner.
  • CMAES and CMAES_ND scans, powerful evolutionary optimisation algorithms, also incorporating novelty detection.

Instructions on downloading are found here; BSMArt version 2 is available via pip via pip install bsmart.

Implementing your own scan is easy, check out our QuickStart

For any questions, get in touch!

If you use it, please cite:

Other relevant references include: * Active learning BSM parameter spaces * M_W in string derived Z’ models * HackAnalysis 2: A powerful and hackable recasting tool

A references.bib file is generated as part of scan running to help, which should include references for the used tools and scans!