Hi Martin,
You have a different binning in option1 and option2.
Check limits and bin width.
When using option1, Root computes automatically the
profile limits and optimizes the limits around the
min/max of the distribution.
So, no surprise if you get slightly different ditributions.
Rene Brun
On Sat, 13 Jul 2002, Martin Turner wrote:
> Hi, I noticed a perculiarity with TProfile.
>
> my_tree->Draw("var1:var2",cuts,"prof"); //option1
> gives a slightly different plot to:
>
> my_tree->Draw("var1:var2>>myprof,cuts); //option2
>
> Now, the number of points, mean and rms of both plots are the same, so the
> same data has gone into both. I set the number of bins in myprof to 100
> which I believe is the default number of bins when using the
> "prof" option. Nothing is cut out on the TProfile initialiser, so the
> number of non zero points are the same for both. The only difference is
> that a small percentage of the points are in different places.
>
> I looked into a very zoomed in region of both plots and it appeared that
> the bin boundaries are different for option1 and 2. Fits to option2 give
> better chi squareds than option1, like a 7 or 8% effect, but I haven't
> looked at enough examples to conclude anything about chi squareds.
> I would like to understand what exactly is going on. Has anyone else
> noticed this?
>
> Martin
>
This archive was generated by hypermail 2b29 : Sat Jan 04 2003 - 23:50:59 MET