Clone the distribution with submodules and compile it using included Makefile
prompt> git clone --recurse-submodules https://github.com/gavalian/driftchambers.git
prompt> cd driftchambers
prompt> make
The code is compiled into one exacutable located in install/bin. The usage of the program can be seen by typing:
prompt> ./install/bin/denoise2.exe -h
Available parameters:
-h --help
This parameter is optional. The default value is ''.
-i --input (required)
input file name
-o --output
output denoised file name
This parameter is optional. The default value is 'output.h5'.
-n --network
neural network file name
This parameter is optional. The default value is 'network/cnn_autoenc_0f_112.json'.
-t --threads
number of threads to run
This parameter is optional. The default value is '8'.
-f --frames
number of events in each frame
This parameter is optional. The default value is '16'.
-l --level
cut off level for background hits
This parameter is optional. The default value is '0.050000'.
-e --events
maximum number of events to process
This parameter is optional. The default value is '-1'.
The program will run multithreaded (default is set to 8 threads), the level option is used for debugging purposes and is not needed for regular running. The requiret parameters are the input file and the network file (usually JSON) used by the neural network. The default networks are included with the package and located in directory denoising/code/network. default network is cnn_autoenc_0f_112.json. Example usage:
prompt> ./install/bin/denoise2.exe -i clas_005977.hipo \
-o clas_005977_denoised.hipo \
-n denoising/code/network/cnn_autoenc_0f_112.json
The input file must contain the bank DC::tdc for the denoiser to do it's job.