Message: Re: Different perfomance on Linux and Windows Not Logged In (login)
 Next-in-Thread Next-in-Thread
 Next-in-Forum Next-in-Forum

None Re: Different perfomance on Linux and Windows 

Forum: Hadronic Processes
Re: None Different perfomance on Linux and Windows (Gorelick Sergey)
Re: None Re: Different perfomance on Linux and Windows (Vladimir IVANTCHENKO )
Re: None Re: Different perfomance on Linux and Windows (Gorelick Sergey)
Re: None Re: Different perfomance on Linux and Windows (Vladimir IVANTCHENKO )
Re: None Re: Different perfomance on Linux and Windows (Gorelick Sergey)
Date: 19 Dec, 2006
From: Vladimir IVANTCHENKO <vnivanch@mail.cern.ch>

On Tue, 19 Dec 2006, Gorelick Sergey wrote:

> *** Discussion title: Hadronic Processes
> Email replies to PublicHyperNews@slac.stanford.edu must include:
>   In-Reply-To: <"/hadronprocess/615/1/1/1/1"@geant4-hn.slac.stanford.edu>
>   Subject: ...change this to be about your reply.
> 
> ------------------------------------ What problem you may have: some
> protons reflected back, so if beam direction is along Z axis from Z=0,
> Z-distribution of end points will have a peak around the range R(E) and
> the tail with -R(Z)< Z < R(Z). If you have a small statistic, the tail
> may not be seen at all, suddenly 1 event has random end point Z in this
> interval and damage the mean value...
> ------------------------------------
> 
> Hello,
> 
> I have tried to eliminate backscattered protons from the calculations,
> but the results are not very much changed. With 4.8.1.p01 (also with
> 4.8.2 that I currently installed) the number of backscattered protons is
> somewhat 1-2 per 1000 incident projectiles. In case of 4.8.0.p01 the
> number is hardly one per 5000 or even higher. But in any case, I notice
> that the stopping of protons in medium is more reliably predicted with
> 4.8.0.p01 rather than with newer releases. There might've been an option
> that I am doing something wrong, but a) since you mentioned that my ways
> of calculations are correct and b) the same source code on different
> releases produces different results for the same process, there's
> nothing much left but to attribute the differences to the release
> itself. 4.8.0.p01 provides more reliable outcome for stopping than the
> newer releases, since the results are pretty similar with those obtained
> in SRIM. Some few % of differences can be blamed on low statistics and
> slight differences in stopping forces. I have tested only Tantalum
> material. The material is defined "brutally" inside the code via:
> 
> G4Material* Ta =
>   new G4Material("Tantalum", z= 73, a= 180.9479*g/mole, density= 16.69*g/cm3);
> 
> Some differences in the density might be present in different data sets.
>

Hello Sergei,

There is two minor moments:

1) We rediscover that for prediction of precise range one has to use more 
fine binning in dedx and range tables. In the release 8.2 in TestEm7 we 
use 480 bins in the main standard PhysicsList instead of 120 default.

2) NIST Ta is little bit different from yours, try to use 
  G4material* myta = 
G4NistManager:Instance()->FindOrBuildMaterial("G4_Ta");

Does Windows and Linux agree if reflected protons are removed?

cheers,

-- 
    Vladimir Ivanchenko  
    ### CERN PH SFT Tel: +41-22-76-78-871  Vladimir.Ivantchenko@cern.ch ###

Inline Depth:
 1 1
 All All
Outline Depth:
 1 1
 2 2
 All All
Add message: (add)

 Add Message Add Message
to: "Re: Different perfomance on Linux and Windows"

 Subscribe Subscribe

This site runs SLAC HyperNews version 1.11-slac-98, derived from the original HyperNews


[ Geant 4 Home | Geant 4 HyperNews | Search | Request New Forum | Feedback ]