v0.27.1 is released!

Fixes

• Some requests are now wrapped with exception handling so that when a connection resets, the original request is retried.
• Ingredient Fractions were added to the Predictor modules
• Defaults for the Molecular Structure Descriptor are now handled correctly

v0.26.0 is released!

What’s New

• We now have a formulation descriptor and two formulations-related predictors (#324)
• LabelFractionsPredictor will compute total fraction of all inputs with a particular label (#332)
• SimpleMixturePredictor will provide features based upon fraction-weighted averaging of mixture components (#333)
• Data Sources now have the concept of an identifier, or a column that contains values that reference another row of the table
• Supporting Molecular Structure value types, we now support Ara Columns for SMILES or InChI representations of those structures
• Retrieving the responses of a predictors from the Citrine platform (#331)

Improvements

• More documentation. Specifically, we've added descriptions of how to build data sources and predictors for mixtures
• And, of course, it all get unit tests.

v0.20.0 is released!

What’s New

• Added a molecular structure featurizer (#321)
• Update requirements so that GEMD libraries will support molecular structures

Improvements

• Documentation was added around ProductDesignSpace objects (#313)
• TableCollection.read was updated to allow retrieval through ids instead of just Table objects (#317)
• Improved pagination behavior (#323)
• Add structure to the predictor report by creating a PredictorReport object (#319)

Fixes

• Gracefully handle multiple descriptors in a predictor report that have the same key(#325, #326, #329)
• Various additional bug fixes (#322)

v0.17.4 is released!

What’s New

• Several list-by and get-by methods have been added to support traversals through networks of GEMD objects (#306)
• A .name alias for FileLink.filename for consistency with other resources (#311)

Improvements

• This package is now tested against python 3.5, 3.6, 3.7 and 3.8 (#314)
• Documentation for the restricted set of expression operators (#312)
• Documentation of the GEMD validation methods (#308)

v0.16.8 is released!

What’s New

• The Citrine Python module has been migrated to take advantage of the GEMD libraries
  • https://github.com/CitrineInformatics/gemd-python/
  • https://pypi.org/project/gemd/

Improvements

• Now, when the register_all method is used on a set of objects, the method will add the Citrine ID of the newly registered object to the objects that were passed. This means that if objects are linked directly (as opposed to link_by_UID), the next object will correctly include its new identifier.

Fixes

• The normal corrections to documentation and augmentation of testing are here as well

Deprecated

• Table.read from citrine.resources.table is deprecated in favor of TableCollection.read from the same module

Citrine v0.15.7 is released!

Citrine v0.15.7 is released!

Whats new?

• Ability to list all versions of a table. (#289)

Improvements

• The InorganicDescriptor is deprecated in favor of the more aptly named ChemicalFormulaDescriptor. (#288)
• More customizable pagination interface. (#289)

Fixes

• The scores "LI" and "EI" are now always referred to by their correct names in variable names and in docstrings. (#281)
• filter_by_template and filter_by_spec are documented more fully. (#292)

Citrine v0.15.1 is released!

Citrine v0.15.1 is released! This release patches a deserialization issue in the most recent release and introduces a new seeding utility.

Whats new?

• A seeding utility that separates templates from other data objects (#285)

Fixes

• Include EVScore in the dictionary of known Score types, allowing them to be deserialized (#287)

Citrine v0.14.0 is released!

Citrine v0.14.0 is released! This release provides access to the new citrine.seeding package which includes various utilities useful for seeding a new deployment.

What's New?

Convenience utilities for seeding a new deployment are now available (#283)

Citrine v0.13.2 is released!

Citrine v0.13.2 is released! This release includes a ton of documentation and usability improvements in getting data onto the platform. In addition, the AI Engine now supports an important transfer learning use case (models can now accept inputs that were generated by other models in the graph without support in the training data).

Whats new?

• AI Engine: GraphPredictor now supports chaining inputs and outputs for common transfer learning use cases (#252)
• Scores have been renamed and a new score (ExpectedValue) was added (#233, #280)

Improvements

• AI Engine: Processor is now optional in DesignWorkflows (#240, #241)
• Various bug fixes to increase client stability (#224, #253, #255, #257, #258, #259, #260, #261, #270, #271)
• Improved documentation and quality of life features for users getting data into the system (#229, #235, #238, #244, #248, #249, #251, #274, #278, #279)

Coming Soon

• AI Engine functionality is now explicitly marked ALPHA. Watch for these tags to be removed in upcoming releases, signaling a general availability release. (#232)
• GEMD objects now have a dataset field (currently None) that will soon be populated with the ID of the dataset that contains the object.
• AI Engine: Introducing DataSource configurations to add support for a greater range of data sources for training data, design spaces, etc. in the future. (#216)
• Upcoming features related to building tables from GEMD data registered on the platform (#226, #228, #234, #236, #237, #202, #245, #247, #250, #256, #263, #264)
• Contributor documentation and experience improvements (#227, #239, #243, #254, #262, #265, #269, #273, #275)

Citrine v0.8.0 is released!

Citrine v0.8.0 is released! This release includes significant improvements to the reconstruction of material histories and the serialization and deserialization of the client version of taurus objects.

Improvements

• Material histories will be reconstructed completely, including the ingredients of processes and measurements of materials (#223)
• Taurus v0.6.0 is now included, providing enhanced flatten and json support (#223)
• Sphinx is now configured to inherit docs from parent classes (#197)
• Added documentation of the unit testing system this package uses (#201)

Fixes

• Expression predictor documentation had reversed the order of the alias name and the full name (#185)
• The session is no longer added to data platform resources that might be json serialized (#207)
• The workflow execution controller extracts the right fields from the API response (#221)

Changes

• citrine.attributes has been deprecated and will be removed in v0.9.0. Attributes should be imported from taurus.entity.attribute instead (#223)
• Client support for the resources and components of the AI Engine have been marked as [ALPHA] to note that their APIs are unstable. They will be omitted from future release notes until they reach beta.