reverse-communication-iterative-solver 0.0.1

Python package/template to build iterative solver via reverse communication and classes

This package aims to provide a common template for building iterative solvers, those solver that approximate problem solutions with incremental steps (called “update” in this project). This template will over a minimal base for developing, tuning and using iterative solvers from the earliest protypting phase to the delivery of block box algorithm.

The interface between an (advanced) user and the iterative solver is decomposed into the following 4 main phases

1. set the solver controls and problem inputs before the next update;

2. reset the solver controls reset problem inputs after a update failure;

3. study any you want combaning the information from problem, unknows, and solver;

4. set when the solver has finished its job.

This decomposition is done via a Reverse Communication (RC) approach, an old style of programming used for example in the ARPACK library, combined with Python classes. We adopt this approach because it let the user work “manually” on the structures describing its problem instead of having to define in advance the procedure to handle the steps 1,2,3,4. This is particularly convient in early step of the algorithm development. A nicer and clenear function hiding the RC cycle to an user who wants a block-box solver can be easily build.

• Free software: GNU Lesser General Public License v3 or later (LGPLv3+)

Installation

pip install reverse-communication-iterative-solver

You can also install the in-development version with:

pip install https://github.com/enricofacca/python-reverse_communication_iterative_solver/archive/master.zip

Documentation

https://python-reverse_communication_iterative_solver.readthedocs.io/

Development

To run all the tests run:

tox

Note, to combine the coverage data from all the tox environments run:

Windows	set PYTEST_ADDOPTS=--cov-append tox
Other	PYTEST_ADDOPTS=--cov-append tox

Changelog

0.0.1 (2021-12-30)

• Updating the README to real package aim

0.0.0 (2021-12-29)

• First release on PyPI.