bomat

Cultural learnings of Bohemia for make benefit glorious library of LAPACK

A set of Fortran functions for visualizing Bohemian matrices

See the original project: https://github.com/BohemianMatrices/BHIME-Project

Download

git clone --recursive https://github.com/JeffIrwin/bomat
cd bomat

Build

Use the provided CMake wrapper script. If using gfortran and reference LAPACK, build that first one time too:

./build-lapack.sh
./build.sh

If using ifort, there is no need to build LAPACK. Just export the environment variable to tell bomat's CMake list to use the Intel Fortran compiler:

export BOMAT_INTEL=true
./build.sh

Run

The bomat program loads its input from a JSON file:

./build/bomat examples/bomat-7c--2022-02-01.json

where examples/bomat-7c--2022-02-01.json configures a few input options:

{
	# This is a comment (non-standard JSON extension)
    
	# Complex numbers
	"Population":
	[
		# Re       , Im
		 0         ,  1,
		-0.8660254 , -0.5,
		 0.8660254 , -0.5,
		 0.1       ,  0.1,
		 0.1       , -0.1,
		-0.1       ,  0.1,
		-0.1       , -0.1
	],

	# Integers-only.  Zeros will remain 0, non-zeros will be sampled randomly
	# from population.  This JSON array is rank-1, but it is reshaped in
	# a row-major sense into a rank-2 matrix
	"Template matrix":
	[
		1, 1, 0, 0, 0, 0, 0, 0,
		1, 1, 1, 0, 0, 0, 0, 0,
		0, 1, 1, 1, 0, 0, 0, 0,
		0, 0, 1, 1, 1, 0, 0, 0,
		0, 0, 0, 1, 1, 1, 0, 0,
		0, 0, 0, 0, 1, 1, 1, 0,
		0, 0, 0, 0, 0, 1, 1, 1,
		0, 0, 0, 0, 0, 0, 1, 1
	],

	"Samples": 100000000,
	"Image size": 3840,

	"Colormap file": "submodules/colormapper/submodules/colormaps/ColorMaps5.6.0.json",
	"Colormap name": "Magma (matplotlib)"
}

The path to the Colormap file, if not absolute, must be relative to the runtime directory.

Command-line arguments

Use -h for help:

 Starting bomat

Usage: bomat [-h] [-p] [-e] FILE.JSON

Calculate Bohemian matrix eigenvalues and export a plot to a PNG file

Positional arguments:
FILE.JSON   Configuration filename for setting inputs

Optional arguments:
-h, --help  Show this help message and exit
-e          Calculate and export eigenvalues without plotting
-p          Plot eigenvalues from previous job

Sample FILE.JSON contents are like this:
{
[omitted from README]
}

I recommend initially running with a low number of Samples, around 1 million, while playing with the Population and Template matrix structure. Once you find something interesting, use a higher number of Samples for a higher-quality image.

Finally, you can use -p to re-plot with a different colormap without recalculating eigenvalues from scratch.