Re: More fit parameters, Implementation details

Tadeusz Pytlos (pytlos@fizwe5.fic.uni.lodz.pl)
Sat, 9 May 1998 11:48:57 +0200 (CEST)


Hi Rene,
Thank you for your answer. According to your suggestion
I changed TFile hfile("prob.root") to TFile *hfile=TFile("prob.root").
It works, but I observed very strange behaviour. After I click on canvas,
I lost control on ROOT! It doesn't react for anything and I have to
kill it. I suppose it's problem of ROOT implementation on RedHat5.
I'm using ROOT 2.00/04 for RH5.
By the way I would like to mention some similar implementation details.
I tried to use vi under ROOT. Just
root [0] .! vi macro.C
It's difficult find a rule in the behaviour, but very often the vi
spoils, produces characters in command mode and it's impossible
to exit. The solution is only to close a terminal window.
Sometimes is possible to exit, but the terminal window is spoilt,
for example is impossible to execute Enter.
Something interferes between ROOT and vi (vim?). Has anobody
observe such behaviour? Now I've given up the idea of using the vi
under ROOT.
The other problem. Sometimes I have to interrupt a run of ROOT.
Simply I've forgetten of something. Just Ctrl+c. Then .q to exit
to the command line. Everything is all right, however, the characters
are black on black background!
The last one. On my RH5 there is seems that .rootrc and rootlogon.C
work only if they are in the working directory. No effect of them
if they are in $ROOT directory or my $HOME directory. But that's is
probably problem of my configuration. Nevertheless, I think that
default settings ( I mean distributed with relase of ROOT kit)
should be more suitable for the A4 paper,portrait, without gray
background - it will be easy for beginners.
That's enough details.
Best wishes,
Tadeusz

On Fri, 8 May 1998, Rene Brun wrote:

> Hi Tadeusz,
> In your macro below, you have the statement:
> TFile hfile("prob.root");
> When you exit from the procedure prob, all objects created
> in prob (except the ones created via new) are automatically deleted.
> This is the normal scoping rule in C++.
> The object hfile is also deleted. When hfile is deleted,
> all objects in the container hfile are deleted.
> If you simply replace the above statement by:
> TFile *hfile = new TFile("prob.root");
> your program will not delete objects at the end of execution.
> I have modified the macro accordingly.
>
> Concerning, Factorial, Binomial, etc, these functions could be added
> in a TUtils class (as static functions). Just waiting for good
> and efficient code. I have no time to implement these functions.
>
> As described in the TF1 constructor, the parameter referencing
> the function must be of type Double_t.
>
> Rene Brun
>
>
>
> Tadeusz Pytlos wrote:
> >
> > Hello Rooters,
> > I have a few problems linking with fiting of histogram.
> > 1) I'm trying to solve such equation
> > myhist=integral(0,infinity) fun1(dN)*fun2() d(dN),
> > where function fun2() I don't know, so I treat it as parameters
> > to estimate par[dN]. Of course, intergral I replace by sum(0,dN).
> > The problem is that there are only maximum 10 parameters to fit!
> > It's too less for my purposes. Is it possible to implement more?
> > Or is there any better possibility to solve my problem?
> > Can I use TMinuit for such problem? I will be grateful
> > for any ideas.
> > I include my macro and root file with histogram. Myhist is red and
> > right side of equation is black.
> >
> > 2) Is possible to implement such functions like Factorial
> > and Binomial in future ROOT?
> > 3) I don't understand why my macro didn't stop, but I have to use
> > getchar(). What clears my canvas?
> > 4) I went about it, but I want to mention. I had problems with
> > Double_t fitf(Int_t *k, Double_t *par). Is fit function only designed
> > for Double_t?
> > Thank you in advance.
> > Best wishes,
> > Tadeusz Pytlos
> >