suitesparseget

suitesparseget makes it easy to get matrices from the SuiteSparse Matrix Collection into Python. It provides a simple interface to download and access matrices from the collection, which is a widely used resource for testing and benchmarking numerical algorithms.

It is a modern version of ssgetpy, which allows downloading of matrix files, but does not provide a method to import them into Python.

suitesparseget uses the pandas library to manage the index of matrices, which allows easier manipulation and filtering of the data than ssgetpy.

An example in examples/ssget_example.py utilizes my C++Sparse library to make spy plots of the matrices.

The repo is a work in progress, so check back soon for more features and examples!

Usage

import suitesparseget as ssg

df = ssg.get_index()  # get the index of matrices
print(df.head())  # print the first few rows of the index

# Get a problem from the index
problem = ssg.get_problem(index=df, name='arc130')
print(problem)

Configuration

You can configure the download directory for matrices by creating a suitesparseget.toml file. On import, suitesparseget will look for this file in:

1. the current working directory
2. the user's home directory
3. the default data directory (~/.suitesparseget). The first file found will be used. If no configuration file is found, data will be downloaded to the default data directory (~/.suitesparseget).

The configuration file can contain the following sections:

[paths]
data_directory = '.suitesparseget'

The path can either be absolute, or relative to the user's home directory.