Message: Re: problems with accuracy Not Logged In (login)
 Next-in-Thread Next-in-Thread
 Next-in-Forum Next-in-Forum

None Re: problems with accuracy  

Forum: Event and Track Management
Re: None problems with accuracy (Nathan)
Date: 01 Sep, 2010
From: Tom Roberts <Tom Roberts>

On Wed, 01 Sep 2010 15:30:54 GMT, Nathan wrote:
> I am trying to simulate a lattice for an interaction region composed of
> dipole and quadrupole magnets. I have been analyzing the optics of a
> gaussian electron beam as it traverses the magnet lattice. I am
> comparing the optics geant produces to optics produced by MAD-X however
> I get results that deviate on the micron scale in rms beam size. 

The default tracking parameters in Geant4 yield tracking accuracies on the order of microns. To accurately simulate beams with sizes of a few microns you will need to improve the accuracy required of the tracking. Some parameters to look at: deltaChord, deltaIntersection, deltaOneStep, epsMax, epsMin.

> Since I
> am using 10,000 initial particles I do not have a perfect gaussian.
> Could this possibly be causing these deviations?

Possibly, but 10k events usually gives pretty good estimates of means and sigmas for Gaussian distributions. To find out for sure, apply the usual statistical formulas to obtain estimates of the errorbars on the mean and sigma of the distribution you get.

> What are the limits of
> tracking in geant, i.e. what accuracy is possible after ~45 m and
> traversing multiple volume boundaries and magnetic fields? 

This depends on the values of the tracking parameters you use. For the default values and beamlines consisting of a few quads and bends with apertures of ~10 cm, accuracies of a few microns should be possible. Sub-micron accuracy should be possible with tighter limits on the tracking, I don't know what is possible, and of course CPU time will increase.

Also important here: how accurately does the problem you're simulating in Geant4 reproduce the problem you're simulating in MAD-X? This is not a simple question.... In particular, be sure you use the same fringe field models in both.

> Is there any
> way to imput a perfectly gaussian beam distribution without increasing
> the number of particles?

Of course not -- particle tracking is inherently limited by the statistics of the particles you track. But there's nothing magical about 10,000 particles -- for a 45-meter beamline with a half dozen quads and bends, my program (G4beamline) would probably track 50-100 events per second, so an hour or two of running should give you excellent statistics.

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

1 Ok: Re: problems with accuracy   (Nathan - 02 Sep, 2010)
 Add Message Add Message
to: "Re: problems with accuracy "

 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 ]