Xopt v1.0.0

We are pleased to announce the first major version of Xopt, which has been almost completely rewritten from earlier versions to be more modular and general. Some highlights are:

• The core Xopt object now consists of Generator and Evaluator objects, strongly separating out running of a complex evaluation function from the optimization algorithm.

• Generator objects encapsulate the optimization algorithms.

• Evaluator objects encapsulate the user-defined evaluation function, and handle their parallel asyncronous execution using Executors that conform to PEP-3148

• Xopt objects manage these and the data produced in the form of pandas DataFrames.

• Pydantic tooling is now used to handle parsing of all configurations and options, including the YAML formats.

Users of previous versions will need to modify their YAML inputs

Thank you to @roussel-ryan for initiating this rewrite and developing all of the Bayesian agorithms, and also to @jacquelinegarrahan for teaching us the ways of Pydandic and developing novel methods to handle complex objects using it. Finally, thank you to the users of Xopt who pose interesting and novel use cases for this software.

New collaborators and helpful comments are welcome!

Xopt v0.5.2

Bug fix with VOCS constants not being propagated into inputs.

Full Changelog: v0.5.1...v0.5.2

Xopt v0.5.1

What's Changed

• Added restart example by @roussel-ryan in #29
• Bugfixes for specifying initial x and exception handling by @roussel-ryan in #31
• New VOCS Pydantic object by @ChristopherMayes in #32

Full Changelog: v0.5.0...v0.5.1

Xopt v0.5.0

Enhancements

• added the following Bayesian optimization algorithms (thanks @roussel-ryan)
  • Bayesian optimization
  • Bayesian exploration
  • multi-objective Bayesian optimization
  • multi-fidelity Bayesian optimization #11
• improved config setting and error handling (thanks @roussel-ryan)
• added standard python logging

Deprecations

• verbose options and print functions are being deprecated in favor of logging
• vocs['templates'] are deprecated. Items there will be moved to simulation['options']

Bug fixes and cleanup

• fixed experimental compatibility with the latest pymoo #22 (thanks @hhslepicka)
• handle both None and empty dicts in configuration files #20

Docs

• New documentation using mkdocs
• GitHub action publishes docs automatically to: https://christophermayes.github.io/Xopt/
• reorganized example notebooks

Testing

• added pytest testing #10 (thanks @roussel-ryan)
• added notebook testing #19 (thanks @roussel-ryan)

Xopt version 0.4.3

Added a Numpy encoder when dumping to JSON, so that arrays and Numpy types can be cast without error.

Xopt Version 0.4.2

Minor bug fix to allow empty constraints, constants.

Version 0.4.0

v0.4.0

Added version, checked examples.

Generalized algorithm and simulation specification.

v0.2.0

added developer install

test 2, without examples

v0.1.1-alpha

ignore examples

First release for Conda-forge

v0.1.0

ignore examples