0.3.0

0.3.0 (2022-04-06)

• Switched from ChaCha20 to Pcg64. This will change the output as compared to earlier versions even with the same seed.
• Changed lots of little implementation details.
  • On the Lee dataset with default settings, as compared version 0.2.1 it is...
  • About 3x faster
  • And uses 60% of the memory
• Version tags are no longer prefixed with v

v0.2.1

v0.2.1 -- 2021-01-22

Changed

• Reduced memory usage and data copying.
  • Use an online mean function for Yi_MH values rather than storing all non-burn iterations. This saves a good bit of memory.
  • Reuse pre-allocated vectors and avoid some needless copying in the dsyrk results. It's saves a bit of runtime and memory.
• Move documentation from the ./docs directory into a new repository. This sets up using mdBook for generating nice online documentation.
• Fixed some Clippy warnings.

v0.2.0

0.2.0-- 2020-01-19

Most important: You can now seed the random number generator for more reproducible runs, and v0.2.0 uses 2/3 of the memory needed by v0.1.1.

Added

• Makefile for testing the Lee phylum dataset during development.
• Documentation for:
  • Properly setting MC iter and MC burn options
  • Setting OpenBLAS threads
  • Getting sample order correct in the sample data file

Changed

• Updated CHANGELOG.md.
• Updated docs to describe seeding.
• Users must now specify a seed for the random number generator in the config files.
• Switched to the ChaCha20 rng rather than thread_rng()
• Changed the way the MC iterations are stored. Now divnet-rs uses 1/3 less RAM than before!
• Updated the rand crate dependencies.

v0.1.1

v0.1.1 -- 2020-01-16

Added

• Better installation instructions
• An full worked example using the Lee dataset from DivNet
• More documentation

Changed

• When compiling, divnet-rs now uses the system installed OpenBLAS library rather that using the one bundled with openblas-src.

Fixed

• The eigenvector test failed on certain machines because the sign of the eigenvectors is arbitrary. This has been fixed.

v0.1.0

This is the first release!