Some Comments by Bill

Hi Richard -

Thanks for forwarding the MS note from Laszlo Urban. I've now had a first read of this document and I will make two observations.

1) The MS scattering for e- seems to be "spot-on" (at least at low energy and for very thin radiators). The data comparisons are impressive. But there are two major differences from this "test bench" experimental comparison and us. First the energy is essentially below the principle region of interest in GLAST - but they claim that the energy scaling is good. The second difference and the place I suspect there could be trouble is that the radiators used are all very thin (e.g. the 37 mg/cm^2 Au corresponds to ~ .5% while our GLAST tests used 3% and 18% W radiators). If the stepping through the material is done incorrectly - its well known that this leads to errors. From the wallet card " is incorrect to add individual theta0 contributions in quadrature; the result is systematically too small. It is much more accurate to apply (the Highland formula) once, after finding x and X0 for the combined scatterer." We can test this by changing the step size (can we?) in the examples run off by Johann et al.

2) The results given for muons at the end of the paper are in line with those produced by Johann, Francesco, et al. and they are dreadful. The paper does not quote the sigmas for these distributions so I extracted them for the 1 GeV case from the FWHMs and found sigma-G4 = .0015 and sigma-HL =.0019 hence almost exactly what the GLAST collaborators found. These are the only distributions in the paper that are close to the GLAST case in both energy and material thickness (were the 1 GeV muons in fact the GLAST case and if so why were the 1 GeV e's left out of this document?). The separation of the MS formalism into separate pieces depending on particle type doesn't make much sense to me. MS depends on the p, beta, and z of the incident particle. Hard processes of course differentiate and always MAKES IT WORSE. As such the conclusions in section 5 of the paper are not substantiated by what's come before.

Regards - Bill Atwood


----- Original Message -----

From: "Dubois, Richard" <richard@SLAC.Stanford.EDU>

To: "Usher, Tracy" <usher@SLAC.Stanford.EDU>; <>; <>; <>; "'Johann COHEN-TANUGI'" <>

Cc: "'Francesco Longo'" <>

Sent: Wednesday, June 04, 2003 9:16 AM

Subject: FW: MCS model of version 3.2 back?


> Here is a **draft** document from Laszlo on his understanding of the

> state

of MS. I've just printed off a copy to read, so I have no thoughts on it yet.


> Richard


> -----Original Message-----

> From: Laszlo Urban []

> Sent: Wednesday, June 04, 2003 4:23 AM

> To: Francesco Longo

> Cc: Laszlo Urban; Michel Maire; Vladimir Ivantchenko; Dubois, Richard;

Mount, Richard P.

> Subject: Re: MCS model of version 3.2 back?



> Hi Francesco,


> the parameter facrange is useful only for low energy particles, in

> fact it defines a step limitation for these particles in order to have

> better backscattering simulation.It does not any constraint for high

> energy particles.

> The parameter facxsi can change the value of the boundary

> separating the central (or small angle) part of the distribution from

> the tail, its value is 1 by default, playing with it you can change

> the distribution in some degree.


> I send here a draft of a note about the multiple scattering angle

> distribution in GEANT4, it confronts GEANT4 simulation results with

> data/theory and with GEANT3 results as well. In its present form it

> is a draft, I'll continue to work on it.


> cheers


> Laszlo


> On Wed, 4 Jun 2003, Francesco Longo wrote:


> > Another small question. To reproduce the tails or the distribution

> > could we use the methods SetFacrange() and Setfacxsi()

> >

> > What's the meaning of these two values?

> > The facrange should affect the backscatter I understood from the

> > documentation, what for the other? Shall we use other values for

> > parameters? The default value were adopted fow which reason?

> >

> > Bye, Francesco

> >

> >

> >


> --