[Rivet] a few issues/problems with Rivet and its tools

Andy Buckley andy.buckley at cern.ch
Tue Jan 27 10:49:30 GMT 2015


On 26/01/15 21:42, roman lysak wrote:
> 
>> Yes, 2D histograms are not fully implemented at present. The next
>> version of YODA should add overflows for 2D histos and fix 2D I/O, at
>> which point we can start supporting them properly. I'll make sure that
>> we provide an Analysis::scale() function for them at the same time.
> Do you have time estimate for this? If it's more than 1-2 months, we
> will probably have to find another solution.

I think we can do that pretty easily. It's always hard to find time to
work on YODA as well as everything else, but adding "proper" outflows is
not so hard. I've been holding back because it should fit in with the
very long-term plan for how the interface should look. In the meantime
we can improve the Rivet support, with the caveat that outflows and
slicing will not be fully supported. As far as I can tell this is anyway
the case for ROOT 2D histograms, so ~no-one will miss any features ;-)

>> If just calling SetBinContent + SetBinError would be sufficient (and
>> would work), please let me know... it's an easy change to make!
> my understanding is that this would not work.
> However, according
> https://root.cern.ch/root/html/TProfile.html#TProfile:Fill@4 and
> references  therein  it seems to me that this is minimally needed to set
> for each bin:
> 
> - for sum(y*y):  TProfile::GetSumw2() //right now you set here sum(w*w),
> if I understand correctly
> - for sum(y):  TProfile::GetArray()
> - for sum(w):     TProfile::SetBinEntries(bin, w)
> 
> Clearly, the names of accessors/methods are confusing...

So I see! Thanks for that -- I think I must have worked out the sum(w)
part, hence the existing code. The others are not at all obvious, so
this helps.

I guess I can also provide Profile1D -> TH1D / TGraph converters which
just set the bin content and error -- you will be able to use these if
the TProfile still proves elusive.

Best wishes,
Andy

-- 
Dr Andy Buckley, Royal Society University Research Fellow
Particle Physics Expt Group, University of Glasgow / PH Dept, CERN


More information about the Rivet mailing list