v0.1.1

Fix conda-forge installation by adding missing __init__.py to the subsolvers package. The sdist previously excluded the subsolvers directory, causing ImportError when installing from conda-forge.

v0.1.0

Caution

The conda-forge package for this version is broken due to a missing __init__.py in the subsolvers subpackage. Please use v0.1.1 instead.

This is a minor release that makes tranquilo a fully standalone package, removing all
external optimizer dependencies.

• #17: Replace estimagic dependency with optimagic; update CI to Python 3.10–3.12
  (@timmens)
• #19: Remove optimagic dependency, making tranquilo fully standalone; migrate to
  pyproject.toml with hatch; make plotly optional; raise minimum Python to 3.10 (@timmens)
• #7: Fix typos (@segsell)

v0.0.5

This version contains several changes that improve the algorithm

• The parallel acceptance step now takes line-step points into account when creating a speculative sample and uses a smaller sampling radius (before we used the search radius for sampling by accident)
• The default target sample size for the noise-free and serial least-squares optimizer has been updated to p + 2 instead of p + 1. While this slows down tranquilo by a small factor it increases robustness.

v0.0.4

• Remove direct dependencies on estimagic to avoid circular imports.

v0.0.3

First version with adaptive noise handling and parallelization

v0.0.2

In this release we

• #3 Update packaging information (@janosg)

v0.0.1

First release of tranquilo in a designated repository.

• #2 Removes all estimagic related files from the estimagic fork (@timmens)