Cluster size analysis doubts

Dear developers,

I’m working on a monolithic detector made of a 3x3 pixel matrix, with a pixel pitch of 25 um and thickness 100 um. I am simulating 2k events with 200 MeV Muons, using Allpix version 1.6.
I want to investigate the cluster size and so I have a look at the DetectorHistogrammer module, but I have some doubts about the output that I obtained.

  • First of all, I’ve not completely understood the two histograms on the left side: what is the difference between the event size and the cluster size distribution?
    I know that the cluster is the group of pixel which shows a signal above a certain threshold (for my case I didn’t change the default value of 600 electrons in the DefaultDigitizer module), but it’s quite strange that have no events with cluster size > 2 (also because if I look at the distribution on x and y I can see something with clusiz = 3).

This is the hitmap that I obtained

in which I see that if the particle impinges at the corner of the pixel, I can obtain cluster size 4 or 5, right? Is this a correct interpretation of this plot? But this doesn’t seem to be confirmed in the cluster size distribution from the 1st picture…

Maybe you can help me to clarify my doubts,
Thank you very much!

1 Like

Hi @cferrero,

The fact that on cluster_size_map you get cluster sizes of 4 and even 5 while you get 2000 events distributed in bins 1 and 2 on the cluster_size histogram is peculiar to me too. The number of events make sense too (they roughly add up to 2000). From my understanding,

  • cluster size is the number of pixels in a single cluster.
  • cluster size x is the number of pixels along the x direction of the cluster.
  • cluster size y is the number of pixels along the y direction of the cluster.
  • event size is the number of pixel hits in a single event.

As for the definition of a cluster; in the detector histogrammer module, a cluster constitutes a set of adjacent pixels that differ by a unit of 1 in index coordinates, in both directions (I have attached the relevant definition inside the DetectorHistogrammerModule). Could you please share a repository with the relevant configuration/root files?

Thank you,


Hi @cferrero , hi @Andrew ,

I’d like to add to @Andrew s answer. The issue of the cluster size graph showing sizes up to 2 is simply a bug in the histogram definition. In the source code this is done here. You can see here that the extent of the horizontal axis depends on the number of total pixels. For a small matrix like this, the calculation seems inappropriate. This means: you certainly get clusters larger than two, they’re just not presented here.

As a very quick solution, you could change the source code at the histogram definition, in case you’re using a self-compiled version. It is certainly an issue that we will have a look at for future releases!


1 Like

Oh, and maybe two short comments on your last plot and question:

  • this is the correct interpretation, yes.
  • I find it a bit easier to look at this plot in color mode, which you can get via the draw option “colz”
1 Like

…and a tiny comment from my side: Version 1.6 is quite old, I would suggest to move to v2.2.2 which is the latest stable release. :slight_smile: (even though your particular problem is not solved there…)

1 Like

Thank you all for your very quick reply!
I agree, it’s simply a problem of histogram ranges, but I will try to overcome it. I’m using Allpix on a system relying on containers/images on which the software has been installed using Docker files, but I think it should be possible apply some changes to the code and then use the image of the modified version?

For what concerns the difference between the cluster size and the event size histos, if I’ve understood correctly, the event size basically comprehends all the pixels that have a signal over threshold, without the constraint for the pixels to be adjacent, which is instead necessary for the cluster size.

Thank you again and kind regards,

Hi @cferrero ,

to your second statement: yes, indeed, that’s the difference between the two.

To the histogram: we could fix this for a newer version of Allpix Squared, so e.g. in 2.2.3. Would you be willing to move your simulation to a more recent version? As @simonspa wrote, we would anyway recommend this, as older versions are not supported anymore in terms of bugfixes.


1 Like

Dear @cferrero ,

we have made a fix for this issue in this Merge Request, and included it in our hot off the press v2.3.0 release.

As mentioned above, we would encourage you to move to the newest version. A few configuration parameter names have changed since then, but it should be very transparent in the documentation. Please let us know if you encounter any issues.


Thank you very much for your support,

I will try to move to the latest version as soon as possible.