[Halld-tracking-hw] CDC dE/dx and truncation

[Sean Dobbs]

Hi Nacer,

My 2 cents on this:

- We should evaluate the desired truncation using the amplitude for dE/dx,
but from Naomi's studies it looks like we are not too sensitive to this.
Hopefully this will make coming to a decision on the correct value easier.

- I think that one truncation value is fine, and higher-order corrections
can be made on a particle-type-dependent basis.  (With luck, changing to
amplitudes will make them smaller!)

- I think the particle-dependent corrections will need to be over a wide
range of phase space, so we should add some more channels to study.  I
don't think we need them for the truncation study.

Cheers,
Sean


On Wed, Aug 16, 2017 at 5:42 AM Abdennacer Hamdi <nacer at jlab.org> wrote:

> Good Morning,
>
> Thank you Naomi for this study.
>
> This confirms that we could keep more hits (80% rather than 50%), this
> may improve the tracking.
>
> It seems that using the Amplitude rather than the the integral in dE/dx
> measurement gives a better result.
>
> As the dE/dx truncation was optimized for the integral, Do we want to
> study the optimal truncation again for the amplitude ?
>
>
> PS: Concerning the Protons & Pions the truncation delivered ~ 20%.
>
> In the case of electrons & koans, since the statistics is very low (2016
> Golden period), and the phase space in (p,theta) is different from the
> protons and Pions (different reaction cover different area), which makes
> it hard to compare Pions to kaons in the same (p, theta) bins.
>
> The plugins used for:
>
> * Electrons/ Positrons: e+,e-,g,p (I may also use the e+, e- conversion
> plugin)
>
> * Kaons: K+,K-,p
>
> Since the truncation value should not depend on the particle type, do
> you think that comparing Protons & Pions is enough at this level ? or it
> is still interesting to look into other channels ?
>
>
> Cheers,
> Nacer
>
>
>
> On 08/15/2017 10:08 PM, Naomi Jarvis wrote:
> > The attached files each have a few plots in comparing runs at different
> rates and changing the dE/dx truncation from 50% to 20% (this is the % of
> hits which are removed, starting with those with largest dE/dx).
> >
> >
> > My observations:
> >
> > The dE/dx from pulse integral broadens at higher rates, quite a bit more
> than the dE/dx from amplitude does.
> >
> > Changing the truncation to 20% is very beneficial to the dE/dx from
> integral, only slightly beneficial to dE/dx from amplitude (the 20% was
> picked for optimal dE/dx from integral).
> >
> > dE/dx from amplitude gives better separation than integral, even with
> 20% truncation.  Maybe this would improve further with a different %.
> >
> >
> > Best,
> >
> > Naomi.
> >
> >
> >
>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://mailman.jlab.org/pipermail/halld-tracking-hw/attachments/20170816/04985d08/attachment.html>