This bash script creates convolutions (e.g. Gaussian or Cauchy) of impulse spectra
with gnuplot.
Download repository as a Zip-archive or use:
git clone https://github.com/nspiller/gnuplot-convolution
Add the folder to your path variable and make gnuplot_convolution
executable with chmod u+x gnuplot_convolution.
The input file and the desired distribution with the full width at half medium (FWHM) has to be supplied, e.g.:
gnuplot_convolution inputfile.dat -c 0.2
In the input file, the data needs to be stored in whitespace-separated columns, since awk is used to process the files.
The columns headers are used to generate the name in the key of the plot.
By default column pairs are plotted for each input file,
so column 1 is plotted against column 2, 3 against 4 etc.
The column pair can be specified by giving the --column-pair argument directly
after the input file, e.g:
gnuplot_convolution inputfile.dat -p 3 1 inputfile.dat -p 2 7
Only the convolutions are plotted. The impuls spectrum can be plotted
with --plot-impulses
The suffix .dat is automatically stripped from the input file
name and the basename is used to write inputfile.gpi and inputfile.pdf.
The basename can be changed with --output-base.
Some default values can be set directly in gnuplot_convolution
in the "Default variable declaration" section.
To assure a uniform layout of the plots, gnuplot settings can be
specified on two levels of hierarchy. General settings (e.g. appearance) can be
stored in gnuplot_convolution.global.gpi.
Settings specific to each plot (e.g. minimum and maximum of the axis)
can be stored for each inputfile.dat in that folder
in an individual file named inputfile.local.gpi.
Multiple input files are supported:
gnuplot_convolution inputfile_1.dat -p 1 2 inputfile_2.dat -p 4 7 inputfile_3.dat
The execution of gnuplot can be suppressed with --plot-nothing. This can be
useful if the code block is to be used in another programme.
Look in examples for some use cases.