|
[Rivet] Rivet and covariance matrices etcCorpe, Louie l.corpe at ucl.ac.ukThu May 3 12:00:34 BST 2018
Hi, > Perhaps even, if there is time, I could give a short talk about the above from the ATLAS PMG perspective: > the tools we have so far, what our recommendations would be etc... perhaps I should clarify - this would not be from the "ATLAS PMG" perspective per se (I don't want to speak for the collaboration as a whole here), but just a discussion on some of the options which I have explored in augmenting the YODA format to handle covariance info, and how submissions on HepData might be modified to make the use of covariance info more straightforward... All the best, Louie > On 3 May 2018, at 10:48, Corpe, Louie <l.corpe at ucl.ac.uk> wrote: > > Hi Hannes, Andy etc, > > > This business of using covariance information to calculate goodness of fit is something that I have been working on from the ATLAS side. > In fact, this is something which the ATLAS physics modelling group has been very interested in, and in particular we've been working on tools to help manipulate the covariance info and use it properly when making plots with YODA files. > The idea would be to eventually back-propagate some of this into YODA/rivet (as Andy mentioned) and to change the culture so that physics results published on HepData have all the required info to produce a covariance matrix. > > I'll be available during there meeting time tomorrow. I'll certainly attend. > Perhaps even, if there is time, I could give a short talk about the above from the ATLAS PMG perspective: > the tools we have so far, what our recommendations would be etc... > > Let me know what you think, > > Louie > > > > >> On 3 May 2018, at 10:20, Hannes Jung <hannes.jung at desy.de <mailto:hannes.jung at desy.de>> wrote: >> >> Dear Andy et al >> >> thanks a lot for your positive reply. >> >> Yes, sure Rivet is not using data, I meant when it comes to calculating chi2 or goodness-of-fit. >> And yes, it is exactly what we want to discuss, how to standardize the correlation matrix to be put >> on hepdata, such that it can be used easily in Rivet for calculating chi2. >> Sorry, if it was not clear, what I meant. >> So, it would be great, if someone from Rivet could participate, perhaps tomorrow is not so important, >> as we just start, but it would be great if we have this discussion in close contact. >> >> Thanks a lot >> Cheers >> Hannes >> >>> On 3. May 2018, at 11:14, Andy Buckley <andy.buckley at cern.ch <mailto:andy.buckley at cern.ch>> wrote: >>> >>> Hi Hannes, >>> >>> This is something we're very interested in, and I would be happy to participate in general although am indisposed at the critical time tomorrow (maybe someone else from the "team" can join?) We're also following this up in the context of other measurements and searches. >>> >>> But I think there's a misunderstanding in saying that Rivet uses uncorrelated data uncertainties. Other than for plotting cosmetics, Rivet doesn't use data uncertainties at all: they only come in downstream when computing goodness of fit or other likelihood measures. Do you mean that the reference data files in Rivet do not separate error classes or encode correlations?: this is true, but not fundamental. What's needed is parallel development for a) HepData to output the correlation breakdown in a standardised way which can be automatically identified with the relevant histograms, and b) for YODA's Scatter types to support multiple error bars: the latter work is underway. >>> >>> Happy to discuss this more, because it's really become critical for a lot of physics exploitation. >>> >>> Andy >>> >>> >>> On 3 May 2018 at 07:14, Hannes Jung <hannes.jung at desy.de <mailto:hannes.jung at desy.de>> wrote: >>> Dear Rivet developers, Andy et al >>> >>> in the LHC EW working group on jets and vector-bosons we are starting >>> to compare measurements from the different LHC experiments to see, >>> whether the measurements are consistent. >>> We start just doing a comparison using the Rivet plugins with one or two >>> well defined theory calculation. >>> >>> At the moment, the experimental uncertainties (stat and syst) in Rivet are >>> treated as uncorrelated uncertainties, while for a detailed comparison with theory >>> also the correlations and the different correlated systematic uncertainties >>> should be treated. >>> >>> Is there already some discussion under the Rivet developers, to include those >>> uncertainties, or if not, would you be interested to join the discussion in our >>> working group on how to consistently define the correlation matrices between the >>> experiments, such that they can be used also within Rivet ? >>> >>> We will have a first meeting of the Jets + Bosons WG on Friday, tomorrow: >>> >>> https://indico.cern.ch/event/721044/ <https://indico.cern.ch/event/721044/> >>> >>> Your participation would be very welcome >>> >>> >>> Best >>> >>> Hannes >>> >>> *********************************************************************** >>> Hannes Jung >>> Email: Hannes.Jung at desy.de <mailto:Hannes.Jung at desy.de> >>> mobile :+49 40 8998 93741 >>> http://www.desy.de/~jung <http://www.desy.de/~jung> >>> Tel: +49 (0) 40 8998 3741 >>> Fax: +49 (0) 40 8994 3741 >>> DESY, CMS 01B/02.213 >>> Notkestr.85 <https://maps.google.com/?q=Notkestr.85&entry=gmail&source=g>, 22603 Hamburg, FRG >>> *********************************************************************** >>> >>> >>> >>> >>> >>> >>> >>> >>> -- >>> Dr Andy Buckley, Lecturer / Royal Society University Research Fellow >>> Particle Physics Expt Group, University of Glasgow >> >> *********************************************************************** >> Hannes Jung >> Email: Hannes.Jung at desy.de <mailto:Hannes.Jung at desy.de> >> mobile :+49 40 8998 93741 >> http://www.desy.de/~jung <http://www.desy.de/~jung> >> Tel: +49 (0) 40 8998 3741 >> Fax: +49 (0) 40 8994 3741 >> DESY, CMS 01B/02.213 >> Notkestr.85, 22603 Hamburg, FRG >> *********************************************************************** >> >> >> >> >> >> _______________________________________________ >> Rivet mailing list >> Rivet at projects.hepforge.org <mailto:Rivet at projects.hepforge.org> >> https://www.hepforge.org/lists/listinfo/rivet <https://www.hepforge.org/lists/listinfo/rivet> > _______________________________________________ > Rivet mailing list > Rivet at projects.hepforge.org > https://www.hepforge.org/lists/listinfo/rivet -------------- next part -------------- An HTML attachment was scrubbed... URL: <https://www.hepforge.org/lists-archive/rivet/attachments/20180503/a7a3d4cb/attachment.html> -------------- next part -------------- A non-text attachment was scrubbed... Name: smime.p7s Type: application/pkcs7-signature Size: 5451 bytes Desc: not available URL: <https://www.hepforge.org/lists-archive/rivet/attachments/20180503/a7a3d4cb/attachment.bin>
More information about the Rivet mailing list |