nuisance is hosted by Hepforge, IPPP Durham


NUISANCE provides a coherent framework for comparing neutrino generators to each other and external data. NUISANCE can also tune cross-section parameters to available data.

We have over 350 cross-section distributions publicly available for GENIE, NuWro, NEUT and GiBUU. For the full list of distributions, please see here.

For more information, see a recent talk by us.

The NUISANCE slack: If you want access without a email address, please email us.

Authors: Luke Pickering, Patrick Stowell, Callum Wilkinson, Clarence Wret.

Installing NUISANCE

Documentation on the NUISANCE installation procedure is available here.

We also provide a container with GENIE, NuWro, NEUT and NUISANCE prebuilt, with a detailed set of tutorials, from the NuInt 2024 school. The slides are here, and the tutorial readme is on our github. For a tutorial on using the flat trees, see this talk.

For a list of tutorials, see the tutorials page and the tutorials part of the talks page.


If you're finding NUISANCE helpful and publish results using it, please cite it accordingly. The page citing NUISANCE has the citation in different formats.


Multiple Generator Inputs

NUISANCE can currently handle inputs from the following generators:

  • NEUT
  • NuWro
  • GiBUU (with third party event converter)
  • NUANCE (Shape-only)
If there is a different generator you would like us to implement please let us know.

Raw Generator Comparisons

The structure of the NUISANCE core classes promotes consistent comparison of different generators by converting each format into a common NUISANCE event format. Tools are included to convert these events into simple 'flat tree' formats that can be analysed with ROOT alone.

Figure 1: Raw Generator comparisons with NUISANCE for the MiniBooNE flux with CC1pi+ final state.
(left) Shape comparison. (right) Normalised cross-section comparison.

Data/MC Comparisons

The NUISANCE tool "nuiscomp" can be used to generate comparisons of any of the different generators supported and any dataset class already implemented into the framework.

Figure 2: Model comparisons with NUISANCE.
(left) MiniBooNE CCQE numu. (right) MINERvA CC1pi numu.

Automated Parameter Tuning

Generator ReWeight dials can be provided to NUISANCE to produce modified cross-section predictions. The "nuismin" application interfaces these reweight dials with ROOT's minimizer libraries to support automated model parameter tuning using various possible parameter estimation routines.

Figure 3: ANL Automated Tuning of the CCQE Axial Mass Parameter in NEUT.
(left) Likelihood Scan. (right) Enu ANL Cross-section Data.

Cross-section Systematic Tools

The NUISANCE tool "nuissyst" is provided to support the study of cross-section systematics for neutrino experiments. Different routines are implemented to validate reweight parameter responses and generate systematic error bands from arbritrary covariance matrices for those parameters.

Figure 4: GENIE MC Toy Throws used to create systematic error bands.
(left) Example 1-sigma reweight dial variations. (right) Likelihood distribution for all toy throws.

Figure 5: GENIE MC Systematic Error Bands for ArgoNeuT anti-neutrino data.
(left) Muon Momentum. (right) Muon Angle.