Message: Re: Geometry Parameterization Not Logged In (login)
 Next-in-Thread Next-in-Thread
 Next-in-Forum Next-in-Forum

None Re: Geometry Parameterization 

Forum: Geometry
Re: Question Geometry Parameterization (Matt Worden)
Date: 06 Oct, 2009
From: John Apostolakis <John Apostolakis>

Dear Matt,

Could you please explain your geometry setup with a bit more detail?   
It is difficult to be sure how you implemented it from your brief  
description.  I will interpret it and offer some feedback and advice  
for what could be your setup - hoping this guess is correct.

My interpretation of your geometry is that the following volumes are  
place in a common parent volume:
- 1 Sphere Hemisphere
- 121 Tubs
And that these volumes are placed so that the 'edge' circle one end of  
the Tubs rests on the surface of the sphere.  This creates a geometry  
which will not behave correctly in Geant4.   ( So long as the volumes  
overlap there is a problem - not just if the overlap is as large as  
this. )

Volumes cannot overlap in Geant4 - like this or in any similar way.    
Some geometries that G4 could accept would be:
- the Tubs and Sphere touch at a point (clearly useless for capturing  
the photons), or
- either the Tubs or the sphere must be cut to create a fit between  
the neighbouring volumes (a sphere and each Tubs in turn).

If this is your setup, you will need to modify your geometry. The  
simplest way to obtain a Geant4-legal geometry would be to create a  
volume starting from the Tubs and subtracting the Sphere.  If you use  
this to replace the Tubs in your setup, I expect that both your  
geometry definition (either using 121 solid Tubs, 121 logical + 121  
physical OR 1 solid, 1 logical and 121 physical) would work and give  
the same results.  ( If you experience a difficulty with such a  
geometry, we will be pleased if you could report it and we will try to  
help. )

I note that overlaps between physical volumes are not allowed or  
supported by the Geant4 geometry modeler.  In case macroscopic  
overlaps exist, the behaviour of the navigation is not well defined.   
( The impact of microscopic overlaps should be negligible - if the  
modeler works as specified. )  If your geometry currently includes  
macroscopic overlaps, it is not surprising if changes in its  
implementation cause different behaviour.

To see whether your geometry has what Geant4 considers overlaps,  
please look in the User Guide for Application Developers - where the  
different tools to detect this are documented.  ( For a geometry of  
around 100 volumes, the newest method which checks each volume when it  
is placed using the optional additional parameter when creating a  
Placement should be very effective. )

In case I have misunderstood your geometry, please excuse my  
assumptions - and let me/us know about your implementation so that we  
can offer more advice.

Best regards,
   John Apostolakis

P.S. As you write from a free email account, could you please also let  
us know you affiliation and your project.

On Oct 6, 2009, at 1:38 PM, Matt Worden wrote:

> *** Discussion title: Geometry
> Hello, I am using a quartz sphere to generate photons. I surrounded  
> the
> sphere with air so that there is a large difference in the dielectric
> constant. On one hemisphere of the sphere are 121 parallel quartz
> channels that overlap into the sphere. The channels were made to guide
> the photons. Originally I had an array of 121 different Tubs  
> (channels)
> and this worked fine.
> Then I changed to use only One Tub Solid, One tub Logical, and an  
> array
> of 121 Tub locations. Now the photons do not go into the channels from
> the overlapping sphere. Does anyone have any idea why?
> -------------------------------------------------------------
> Visit this GEANT4 at message (to reply  
> or unsubscribe) at:

 Add Message Add Message
to: "Re: Geometry Parameterization"

 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 ]