Biasing problem: track weight is sometimes reset to 1

Hi everyone!
I am studying the environmental background due to radioactive decays in the walls of the experimental hall. Since there is a lot of suppression, I used a geometrical biasing to increase the particle flux, however I have an anomalously high number of particles that arrive in the detector with weight=1; even thought they must have been split several times before. Do you have any idea how this could happen?

Here more details: the detector is encased in a cylindrical copper shield, 12 cm thick, I divided this shield into 12 concentric shells, and each gamma (main source of background) which goes from one shell to the next is doubled (I also make sure that the particles are split only when moving towards the detector, not away from it). I would expect that all the gamma’s that arrive at the detector have a weight of 1/2^11 (since they are not split when entering the first shell nor when they come out the last one).

The results I find are the following:

a) When NO biasing is used, with 10^5 primary events I find 9 gamma’s in the detector
b) With biasing (and 10^5 primary events), most of the events (around 12,200) have the expected weight of 1/2^11; if we multiply this number by the weight, we get around 6, which IMHO is consistent with the previous result
c) I have few events (order of few hundreds) with lower weight (between 1/2^12 to 1/2^20), my guess that this is due either to multiple splitting (i.e. the cloned track is wrongly seen in the wrong volume: for the biasing I mostly followed example GB03, but for other reasons I had to remove the condition track->GetCurrentStepNumber()!=1 ), or maybe from particles bouncing around. This is not really a big problem, however, since the total weight of this kind of tracks is quite low, moreover all these gamma’s have very low energy
d) What concern me the most is the following: there are few events (of order of 60) with weight larger than 1/2^11, in particular there are 26 events with weight 1. These gamma’s must have been split before arriving to the detector (for one, it’s physically impossible to arrive to the detector without crossing the 12 layers of shield, but most important because these 26 events, if they actually have w=1, would be wildly inconsistent with the unbiased results: I got only 9 gamma’s in the detector without the splitting).

Does anyone has any idea what could be the problem? Let me know if you need more details, I tried to keep this short (or not-so-long) for readability…

Dear Emilio,

I have noticed a similar problem with a different type of primaries, but I haven’t solved it yet. Using GB01 and protons, I see that if that radioactive decay resets the weight to 1. So it might be a more general problem with biasing.