aru is hosted by Hepforge, IPPP Durham


This is the development page of the Automatic Regularized Unfolding (ARU) algorithm.

ARU is a non-parametric algorithm that unfolds detector effects from one-dimensional data distributions. The unfolded solution minimizes the mean integrated squard error (MISE). ARU uses a completely unbinned analysis and a powerful regularization based on the information gain in the solution. ARU requires no manual tuning of the unfolding process and can be used as a black box if necessary.

Get the latest software release.

If you publish distributions that have been unfolded with ARU, please cite us:

H.P. Dembinski and M. Roth,
ARU - towards automatic unfolding of detector effects,
Proceedings of PHYSTAT 2011, CERN (2011).


ARU's documentation is released together with the source code. We try to make interaction with the ARU software as simple as possible.

The default front end is a Python interface. A pure C++ interface will supplied in the future.

picture of unfolded distribution

ARU applied to the toy example from V. Blobel, Unfolding methods in high energy physics experiments, Proceedings of the 1984 CERN School of Computing (1984).
b(x) = unfolded solution with statistical uncertainty band
t(x) = true distribution