Skip to content

Releases: rappdw/redblackgraph

Minor fixes to log statements and argument handling of rbgcf

27 Feb 04:12
ceae0bf
Compare
Choose a tag to compare
Correctly account for various classes when ingesting csv file

* Fix outdir argument handling

* Add additional default filter to confirm to latest fs-crawler version

* Correctly account for skipped vertices

Enhanced cannonical representation

21 Feb 04:03
Compare
Choose a tag to compare

A large number of misc changes

  • travis build
  • remove support for < python 3.6
  • enhance cannonical sorting
  • enhance processing of relationship file for ingestion

Initial unsigned and sparse matrix implementation

23 Dec 15:04
329dc74
Compare
Choose a tag to compare
Major refactoring of code (#9)

* Create branch to better explore vector space implications

* Add compute_sign to avos
Remove nz_min and use functools.reduce in it's place
Ensure tests are working

* add compute_sign to reference avos implementation

* rename redblackgraph.simple to redblackgraph.reference to reflect this is a reference implementation

* Type hinting for reference implementations of vector avos product and Warshall algorithm

* Type hinting for calc relationship

* Type hinting for calc relationship

* Further type hinting
Refinements on triangularization
Identify Identity bug in optimized implementation

* Add .Trash* to gitignore

* Refactor c implementation to avos_sum and avos_product

* A couple of minor changes in c implementation
Export c implementations of avos_sum and avos_product

* fix error and warning in c impl

* Comment out avos_sum and avos_product test cases for now... Somehow the c extension wasn't rebuilt properly... deleting all *.so and invalidating the pycharm cache seemed to fix things

* Fix dockerfile for notebook
Update some of the vector algebra in notebooks

* Update requirements
Switch to CPU-only notebook

* Fix script that generate latex and pdf
Update a few calls in the main notebook to the new return signature for warshall

* add pyproject.toml to capture numpy build dependency

* clean up cruft

* Fix setup.py to pip up "setup_requires=numpy"
update versions for numba
update notebook dockerfile based on changes to base

* update requirements.txt based on python version

* remove stale comments

* Working on the basic math. What is avos product for x, y < -1?

* Use absolute rather than relative imports for rb_multiarry
Support building wheels
Add gcc to notebook docker container

* Fix typo

* Add c extension functions to expose underlying avos sum/product implementations
Add test cases

* Tune the math just a bit (range of Avos number, Avos even/odd notation)

* Update docker config
Start working on sparse impl

* Re-integrate sparse implementation
standardize on extension module naming
add color class
Introduce two failing test cases to call out further investigation

* Latest updates to sparse code

* Fix bug in sparse avos_product impl
Fix test_rbm test cases (adjust to red black diagonal vs. old style 1/0 diagonal)

* Inline avos operators

* fix bug in avos_product

* Add copy parameter to reference warshall
Move generate_sparsetools.py to scripts directory

* Address all (almost) c/c++ warnings
Add reference impl of topological sort
Add research.MD (precursor to bibliography)

* Remove debug.py

* Editoral changes (and fallout) to define red as 2^w - 1 (essentailly -1)
and some of the language and notation cleanup in main document.

* Major Changes:

Editorial changes (replace pedigree number with relationship function)
Simplification resulting from editorial changes
Expand multiarray implementation to account for all types defined on platform, both signed and unsigned.
Make base RBG math unsigned to allow for more relationship representation

* remove @maxval@ from computations (parallelism between c and c++ impl)

* Add unsigned test cases for sparse implementation

* Two major changes:
* Rework setup to be more parallel to scipy and prepare to utilize cython
* Begin to change triangularization from my algorithm to just use topological_sort (not canonical, but we'll get there)

* Remove dependency on reference in core submodule
Rename references from warshall to transitive_closure

* Collect types in types submodule

* Update current work

* Editorial changes for clarity and consistency

* check notebooks against current code base

* Update readme

* Change to AGPL while this module is under development. If you need a more commercial friendly license, please message me.