|
[Rivet] Proposals for make-plotsFrank Siegert frank.siegert at cern.chFri Apr 8 11:12:43 BST 2011
Hi Hendrik, >> 1. If bin sizes are increasing along the x axis, then use LogX=1. >> Rationale: I often write a quick MC analysis and don't want to >> bother with setting all the plotting options. That's fine for >> x/y-labels and title, since I can guess them from the histo name. >> But for logarithmically binned x axes the histogram becomes not very >> nice to look at with LogX=0. > > This shouldn't go into make-plots, but into compare-histos, since > make-plots isn't responsible for deciding how the plot looks like. It's > just for plotting. Not the place for magic. Besides that I doubt that > you'd really like this option. Just consider the two attached plots; bin > size increases, but I have the vague feeling that lin.pdf is preferred. Thanks, two good points. So if I come up with a more stringent test telling whether it's logarithmic in x then I'll try to find the right place in compare-histos to put it into. >> 2. The dimensions of the .png files have recently been increased >> quite a bit by changing density=100 -> density=120 (in r2775). I >> don't care about the --png option, but for --pdfpng I would like to >> decrease them at least back to the original value, if not even >> further (e.g. back to the default of 72). > > I'm all in for 100. In fact I didn't like that commit, either. But 72 is > too small in my opinion. I like to look at plots in detail on the > webpage, without having to open all of them, and only open those where I > want to have a really close look. Have you tested it? I actually found it to be working quite nicely. I have prepared the following for testing: http://fsiegert.web.cern.ch/fsiegert/tmp/density/72/ http://fsiegert.web.cern.ch/fsiegert/tmp/density/85/ http://fsiegert.web.cern.ch/fsiegert/tmp/density/100/ If you still think 72 is too small, what about a compromise at 85? That way I could at least have two plots on a half-width window of a 1680 pixel display. ;) Cheers, Frank
More information about the Rivet mailing list |