[Rivet] mcplots

[Hendrik Hoeth]

Thus spake Peter Skands (peter.skands at cern.ch):

> >- The error bands in the ratio plots on
> >   http://mcplots.cern.ch/?query=plots,pp%2Fppbar,uemb-soft,pt,Pythia%208,
> >   are asymmetric, while the reference data has symmetric errors.
> 
> Please look at the error bars again, in particular for the last few
> bins, which are the ones in question.

Okay, you're right. After searching for the histogram ids for a while
(the aesthetical thing, you know) I see that these are indeed
asymmetric.

> >- Can you plot the tick marks in the ratio plots on top of the error
> >   bands, rather than below?
> 
> Yes, if ROOT had been better coded, that would be possible. As it
> is, I don't know how to do that. If anyone stumbles on a solution to
> this, I'd be happy to know.

Well, the choice of ROOT was yours. We have a simple and very
configurable plotting tool that gets this right.  ;-)   *SCNR*

Anyway, if I remember correctly one can as ROOT to draw an empty
coordinate frame. If you do that after plotting everything else, you get
the tick marks on top. Disclaimer: Luckily I didn't have to use ROOT
recently, so that memory dates back to my time at D0.

Here's the first google hit for "root filled histogram tickmarks":
http://root.cern.ch/root/roottalk/roottalk04/0014.html
I guess Draw("sameaxis") is what you need.

> >- Luckily my horizontal screen resolution is 1920 pixel. But on anything
> >   smaller four plots next to each other aren't nice to navigate. Maybe
> >   you can let them float?
> 
> That would indeed be nice. We did look into that, but were not able
> to do it so far. We made a compromise between horizontal and
> vertical scrolling.

What about one table per plot (probably needed for the links below the
plots), and then the browser can arrange those tables as they fit?

> >- Can you put the Rivet histogram ID in the plot, not just the analysis?
> >   That would make identifying plots you are talking about in your bug
> >   reports sooooo much easier.
> 
> You mean like the "d01-x01-y01" in
> # BEGIN HISTOGRAM /REF/CMS_2011_S8957746/d01-x01-y01
> AidaPath=/REF/CMS_2011_S8957746/d01-x01-y01

Exactly. I have now found it after randomly clicking on the links below
the plots ("more" -> "ATLAS"). Now that I know where to find it, I'm
happy. That's all I need.

> >- Many plots have lousy statistics, like the 1960 GeV UE plots or ATLAS
> >   minbias pT spectra. One doesn't really learn something if the error
> >   bars are larger than the vertical size of the plot. Any chance you can
> >   run those with better stats?
> 
> Lol, you really do like to point out the obvious :) Obviously we are
> working extremely hard on this.

I've looked at this for a while, and since the problem still exists, I
thought I'd better mention it. What is the technical problem you
experience? Maybe we can help you solving it?

Cheers,

    Hendrik

-- 
The minimum number of planes one should own is one. The correct number
is n+1, where n is the number of planes currently owned. This equation
may also be re-written as s-1, where s is the number of planes owned
that would result in separation from your partner.       -- Veluminati