CERN@school: Fast Cluster Analysis

This code can be used to perform a "fast" cluster analysis on data collected with a Timepix detector for use in research conducted as part of the CERN@school programme. An example analysis using the radiation profile methods described in Section 4 of the CERN@school Contemporary Physics paper (Whyntie et al. 2015) is included. The datasets featured in the paper are included with the code, but may also be found on FigShare here.

Disclaimers

• This code dates from 2016. While every attempt has been made to ensure that it is usable, some work may be required to get it running on your own particular system. We recommend using a GridPP CernVM; please refer to this guide for further instructions. Unfortunately CERN@school cannot guarantee further support for this code. Please proceed at your own risk.
• This repository is now deprecated, and remains here for legacy purposes. For future work regarding CERN@school, please refer to the Institute for Research in Schools (IRIS) GitHub repository. Please also feel free to fork and modify this code as required for your own research.
• Please note that the linearity cluster variable currently fits the line of best fit by measuring the pixel distances from the line in y coordinate. This means it is not technically rotationally invariant as a cluster variable. To fix this, the perpendicular distance of each pixel from the line of best fit should be calculated and used in the minimisation function. This is left as an exercise for the reader.

Getting the code

First, clone the fast-cluster-analysis repo into your working directory:

$ git clone https://github.com/CERNatschool/fast-cluster-analysis.git
$ cd fast-cluster-analysis

To prepare for running, run the setup.sh script with the following command:

$ source setup.sh

Note: if you are not using a GridPP CernVM, the setup.sh script will not work as you won't have access to the CERN@school CVMFS repository and will have to source your own version of the Python libraries such as matplotlib via e.g. the Anaconda Python distribution.

Getting the datasets

Two datasets - one real, one simulated - are provided with the code:

• testdata/kcldata/: real data from a KCl (Lo Salt) sample;
• testdata/kclsim/: simulated data with a KCl test source.

You may also find the datasets in the FigShare repository here.

Running the code

First, process and make the plots for the real data:

$ mkdir ../tmpkcldata
$ python process-frames.py testdata/kcldata/ ../tmpkcldata
$ python make-plots.py ../tmpkcldata ../tmpkcldata

Note: this will process all 120 frames, which may take some time. To process fewer frames, use the --maxframe option for the process-frames.py script like so:

$ python process-frames.py --maxframes=5 testdata/kcldata/ ../tmpkcldata

Then process and make the plots for the simulated data:

$ mkdir ../tmpkclsim
$ python process-frames.py testdata/kclsim/ ../tmpkclsim
$ python make-plots.py ../tmpkclsim ../tmpkclsim

Finally, you can make and view the comparison plots:

$ mkdir ../tmpcompare
$ python compare-plots.py ../tmpkcldata/ ../tmpkclsim/ ../tmpcompare/
$ firefox ../tmpcompare/clusterplots/index.html &

These are Figures 6 a), b) and c) from (Whyntie et al. 2015). By using your own data, for example collected from your own KCl source, you can make your own versions of these plots to use in your own research.

Making changes and exploring the data

Feel free to fork the repo, play with the code, and see what you can find in the data. We'd love to hear about what you've done with it - get in touch with IRIS if you do!

Authors/Editors/Contributors

• T. Whyntie - @twhyntie - Queen Mary University of London;
• A. Coupe - @acoupe - Uni. of Southampton;
• R. Fickling - @fickling - Queen Mary University of London.

Acknowledgements

The cluster finder is based on C++ code originally supplied to the CERN@school project by Son Hoang (NASA/Uni. Houston). It has been adapted to include plotting functionality from the NumPy and PyLab libraries. The authors would like to thank the 2016 MoEDAL summer students for constructive feedback and suggestions for improving the code. A. Coupe's 2014 summer placement was kindly supported by SEPnet. R. Fickling's project was supervised by J. Wilson of the QMUL Particle Physics Research Centre (PPRC).

CERN@school was supported by the UK Science and Technology Facilities Council (STFC) via grant numbers ST/J000256/1 and ST/N00101X/1, as well as a Special Award from the Royal Commission for the Exhibition of 1851.

Useful links

• Setting up a GridPP CernVM;
• Whyntie et al. 2015 - the CERN@school Contemporary Physics paper featuring this experiment (Section 4);
• The datasets on FigShare;
• The Institute for Research in Schools (IRIS) homepage;
• The IRIS CERN@school website;
• The Official IRIS GitHub Organization.