[Gate-users] doseActor showing strange behaviour in certain cases

[Hermann Fuchs]

Dear Yann,

as far as I understood, the random approach (stepHitType random) works
because the step size of the simulation is compensated.

However, setting the 
stepHitType to middle
and limiting the maximum Step size in this region by
/gate/physics/SetMaxStepSizeInRegion waterphantom 0.01 mm 
results in the same (correct) output than setting the stepHitType to
random.

The important question seems to be, which of the two configurations has
the fewest side effects or - if it can be put that way - is more
correct.
Does limiting the MaxStepSize has downsides? Such as unexpected behavior
or much longer calculation times?

It seems to me that from a numerical point of view, changing the step
size is the more correct approach.

I would greatly appreciate your help and comments.

Cheers,
Hermann


On Tue, 2011-03-22 at 09:58 +0100, Yann PERROT wrote:
> Dear Hermann,
> 
> The hit type refers to the dose deposition along a step, not inside a voxel.
> 
> For each interaction, a particle is moved along a step and the energy is 
> deposited along this step.
> The hit type "Middle" tells to GATE to deposit the energy at a point at 
> the middle of the step.
> Your case is well known and it is recommended to deposit the energy 
> randomly along the step which is consistent since processes are discrete 
> or continuous.
> 
> You can find more explanations in the Gate wiki documentation.
> 
> Bests,
> 
> Yann
> 
> Le 22/03/2011 09:06, Hermann Fuchs a écrit :
> > Dear colleagues
> >
> > It seems that the doseActor exhibits strange unexpected behavior in
> > certain situations.
> >
> > In order to clearly define the situation, I am using the standard
> > Radiation therapy benchmark included with Gate v6.1.
> > Starting the proton.mac file (the proton simulation part of the
> > benchmark) in Gate, returns the energy deposited by penetration depth in
> > the file Config-proton-Edep.txt.
> >
> > Using the unchanged benchmark file, the expected curve can be observed.
> >
> > However, if you change the line
> > /gate/actor/doseDistribution/stepHitType           random
> > to
> > /gate/actor/doseDistribution/stepHitType           middle
> > the resulting curve changes drastically.
> > At the left side of the curve, certain bins register an energy deposited
> > which is by a factor of 20 higher than the expected number, while all
> > other bins show slightly lower numbers than expected.
> >
> > This effect is even easier to see if you increase the resolution of the
> > doseActor by changing the line
> > /gate/actor/doseDistribution/setResolution         1 1 400
> > to
> > /gate/actor/doseDistribution/setResolution         1 1 1000
> >
> > I am unable to explain this effect. As far as I understood, the
> > stepHitType should only change the position within the bin where the
> > energy is measured.
> > This effect can be observed when simulating a few (~500) and as well a
> > bigger number (~1000000) of particles, so it seems that it is not an
> > effect caused by to few simulated particles.
> >
> > Is it possible that there is some unexpected interference caused by
> > mismatched calculation steps?
> > How could I avoid this effect?
> > Simple using random without understanding the underlying effect doesn't
> > seem like a good idea to me.
> >
> > I would greatly appreciate your help.
> >
> > Yours,
> > Hermann Fuchs
> >
> > hermann.fuchs at student.tuwien.ac.at
> >
> > _______________________________________________
> > Gate-users mailing list
> > Gate-users at lists.opengatecollaboration.org
> > http://lists.opengatecollaboration.org/mailman/listinfo/gate-users