Which event biasing is reasonable?


tbh, so far I read only a bit of Event Biasing Techniques — Book For Application Developers 11.1 documentation while I’m getting an overview. When I saw it correctly, there are no recommendations or so which technique should be used when.
Are there any experiences which method makes sense to utilize?

My application is quite simple: There are several particle sources which emit photons which impinge on multiple detectors. I want to measure the hits of each detector separately. I noticed from my geometry that the detectors are hit by photons from a specific particle only within a very tiny angle, thus, making 90 % of the emission angle of that particle source rather senseless. I could reduce the emission angle manually but this would be against the real lab setup.
Which biasing technique is useful here (and in general)? Would Reverse Monte Carlo apply?

Is it also possible to neglect certain physical processes and/or all processes below a certain energy threshold? I’m using the penelope model.

Are you saying you have a detector that is only covered by a small solid angle of your source?

This really depends on what you intend to learn from your simulation. For instance, if you were looking at an inclosed gamma detector you may want to include your entire environment to see how it effects your source attenuation, backscatter peaks, etc.

If you only care about energy deposition directly from your source, I would absolutely limit your source to the solid angle that hits your detector.

Thanks, yes, exactly. One of the two particles sources covers the detectors in a reasonable angle region but the other particle sources emits photons which hit the detector only within 3-4 ° from a 80 ° emission angle.

I’m simulating a dosimeter so your assumption of an inclosed detector applies somehow. But I’m not much interested in the environment or let’s say the influence of the entire environment matters to me only at a much later stage.

But when I limit the emission angle, the results would provide much more photons from that specific generator compared to the other one (which is/should be more dominant). So I guess when doing so I’d also need to reduce the number of photons?
What about importance scaling? From what I’ve read it sounds quite useful but it also looks a bit advanced with potentially being prone to cause severe issues/errors. If I could simply reduce the emission angle while being align with the reality I’d probably prefer this option.

If you are concerned about count rate, just reduce the ratio of photons from the smaller angle source by the ratio of solid angle/4pi.

But also, have you tried running your simulation without limiting the source? I mean if you are able to get reasonable results in a short amount of time without reducing the emission range, then there is no need to try to improve performance.

The only reason I would try to reduce the emission angle is if you are looking at hours for a single run and you could shorten it to minutes

1 Like

Thanks for your input!
The simulation time itself is fair, I guess. I’m just more concerned that only around 0.4 % of the generated photons actually hit any detector. This matches definitely reality but it also makes the simulation not so efficient in general? I believe in total I’ll run the simulation in total for some days for all studies. Maybe even more.

Reducing the emission angle to 10 % and enabling the second particle source now for only each 10th photon is reasonable, or?

I guess I wouldn’t be worried about simulation efficiency unless it affects me, I don’t feel bad that the computer had to generate and throw out photons :grin:

You will have to calculate the flux out of the limited solid angle. It may not be as easy as 10% / 10%

1 Like

That’s a valid point!

As the distribution of the photons emitted over the angles is uniform I should be able to simply cut it down to 10 %, or?

If your reduced solid angle covers is 1/10 * 4pi

True, I have to add that I forgot to mention that I’m having a 2D-scenario?

It should be the same, just use angle/360

1 Like

This topic was automatically closed 7 days after the last reply. New replies are no longer allowed.