Unfolding meeting

→ Europe/Berlin
EVO (at DESY VC room 1c/134)

EVO (at DESY VC room 1c/134)
Gero Flucke (DESY), Thomas Schoerner-Sadenius (DESY)
Description
Title: AnalysisCentre Description: EVO meetings of the Terascale Analysis Centre Community: Terascale Password: anacen Meeting URL: http://evo.caltech.edu/evoGate/koala.jnlp?meeting=eneueMvDvtasaaIBalIn Phone Bridge: ID: 997106 Password: 8525
Hide

Unfolding Meeting on February 9, 2011

Participants: Natalie Mielke, Tim Adye, Nikolai Gagunashvili, Pedro Cipriano, Gero Flucke

Excused: Katharina Bierwagen

 

These minutes by Gero...

 

Welcome/News:

Gero is back at work, but mainly busy with other tasks...

 

Report from PHYSTAT 2011

Tim gives an oral report:

  1. Blobel and Zech showed quite some Dos and Don'ts
  2. Unfolding methods news: new errors for bayesian (presented by Katahrina), development from SVD side (included in ROOT 5.28), and a "dynamically stabilised iterative" method (bayesian like, but stabilisation against background)
  3. a talk of a statistician focused a lot on unbinned unfolding
  4. reports from experiments showed the extensive use of bin-by-bin corrections, now also some use of bayesian - but experts really recommend better methods (than bin-by-bin)
  5. discussion at the end about a probable unfolding challenge for future PHYSTAT

Nikolai asks whether someone propsed a method to test biases in methods, but no answer from PHYSTAT.

RooUnfold news

See Tim's slides. Nikolai mentioned that it is better first to test and tell people about caveats of methods before intgerating into ROOT.

TRUEE news

Just minor bug fixes.

Blind test proposal discussion

Gero raised again the issue of bin width (30 bins from 4 to 16, i.e. width 0.4) vs resolution (sigma 1.5) that seems a too ambitious relation.Nikolai mentions that the relation is similar to the one in one of Blobel's articles (GF: which one?), but he is happy with other relations as well - we just have to agree.

Pedro will participate with TUnfold, Natalie wants to try with TRUEE (nothing against an author using her/his algorithm). Natalie will try to do some studies about resolution and bin width until next neeting, details to be agreed upon offline with Gero.

Our relation with the proposed PHYSTAT challenge is probably that we develop and find out about the figure of merit...

AOB

The man power issue was again discussed, seems that we lack of it. Kerstin Tackmann (CERN, SVD) might get a summer student (Tim to contact her about that). The alliance could fund three month studentships at DESY for work on this topic.

Nikolai mentions that his last article is accepted by NIM, he will add to the list of literature.

As next date we suggest just one month from this meeting if nobody of those not connected today complain (then we doodle). One reason to meet that soon again is that  later on there is a good chance that Gero is on a one month parental leave (exact date depends on the birth of the child...) and Tim will not be around end March/beginning of April.

 

There are minutes attached to this event. Show them.