Wishlist: Tracking Issue

[lena-kashtelyan]

Feature requests marked as 'wishlist' will be gathered here going forward, in order to:

• improve discoverability of issues that report bugs or ask questions,
• help us easily assess these requests when roadmapping.

Please still feel free to open new issues for feature requests (or comment them here if they are short/clear), and we will take care of adding them to this post.

Status: will likely be addressed in the short-term

• ~[DONE] Obtaining best parameters (parameters on the Pareto frontier) for multi-objective optimization in Service API (https://github.com/facebook/Ax/issues/656)~
• Support for Ax viz in Colab (https://github.com/facebook/Ax/issues/306)
• Adding a conda distribution (https://github.com/facebook/Ax/issues/608, https://github.com/facebook/Ax/issues/614)
• ~[DONE] Hierarchical search spaces (https://github.com/facebook/Ax/issues/140)~
• Support specifying fixed features in Service API (https://github.com/facebook/Ax/issues/746)
• Support preference learning in Ax (https://github.com/facebook/Ax/issues/754)
• Early-stopping tutorial in Ax (https://github.com/facebook/Ax/issues/851)

Status: will likely be addressed in the long-term

• TensorBoard support (https://github.com/facebook/Ax/issues/248)
• [IN PROGRESS] Optimization algorithm, suitable for combinatorial search spaces in Ax (https://github.com/facebook/Ax/issues/477)
• Include performance benchmarks in Ax documentation beyond the BoTorch benchmarks published in https://arxiv.org/abs/1910.06403 (https://github.com/facebook/Ax/issues/496)
• Exact equality parameter constraints (https://github.com/facebook/Ax/issues/510)
• Automate selection of appropriate parameters for BoTorch components in Ax based on experiment and data size (https://github.com/facebook/Ax/issues/674)
• "random_seed" setting in Loop and Service APIs guaranteeing full reproducibility (https://github.com/facebook/Ax/issues/605)
• Constraints on ordered choice parameters (https://github.com/facebook/Ax/issues/710)
• Make more common functions available with just import ax: https://github.com/facebook/Ax/issues/774
• TRBO in Ax (https://github.com/facebook/Ax/issues/474)
• Add getters/setters for Experiment._properties and Trial._properties (https://github.com/facebook/Ax/issues/566#issue-866378005)
• Support for MOO for hierarchical search spaces without flattening the search space (#1042)
• Gaussian priors for BoTorch models in Ax (https://github.com/facebook/Ax/issues/1647)
• Slice and contour plots for choice parameters (https://github.com/facebook/Ax/issues/1577, https://github.com/facebook/Ax/issues/1624)
• Storage of Data for trials without Dataframes (associating potentially larger data, e.g. image data, with trials, and passing that data to the model directly; https://github.com/facebook/Ax/issues/880)
• Official release of DeterministicModel and documentation for it (https://github.com/facebook/Ax/issues/1192, https://github.com/facebook/Ax/issues/935)
• Multi-task GP tutorial (https://github.com/facebook/Ax/issues/2086)
• Plotting compatibility with Hierarchical Search Spaces (https://github.com/facebook/Ax/issues/2481)
• SQLAlchemy 2.0 support (https://github.com/facebook/Ax/issues/1697)
• Fix RayTune Tutorial (https://github.com/facebook/Ax/issues/2090)
• Add properties to more easily expose acquisition function details re: optimization (https://github.com/facebook/Ax/issues/2411)

Status: uncertain

• Nonlinear parameter constraints (https://github.com/facebook/Ax/issues/153)
• Adding "projects" to group Ax experiments (https://github.com/facebook/Ax/issues/189)
• Integration with Nevergrad (https://github.com/facebook/Ax/issues/163)
• Returning optimal ranges rather than single values for best parameters (https://github.com/facebook/Ax/issues/320)
• Ax on ARM / Raspberry pie (https://github.com/facebook/Ax/issues/412, blocked on official PyTorch support for ARM)
• ALEBO support for parameter constraints (https://github.com/facebook/Ax/issues/424)
• Selective noise inference for some observations in a field experiment (https://github.com/facebook/Ax/issues/491)
• Off-the-shelf models that do not use Gaussian noise (https://github.com/facebook/Ax/issues/491)
• Support sampling individual parameters from specified distributions (https://github.com/facebook/Ax/issues/702)
• Ability to specify input uncertainties (https://github.com/facebook/Ax/issues/751)
• PAUSED trial status (https://github.com/facebook/Ax/issues/862)
• Discrete fidelity parameter support (https://github.com/facebook/Ax/issues/979)
• Accept numpy values as well as Ax primitives (https://github.com/facebook/Ax/issues/996)
• Exposing more settings for ALEBO (https://github.com/facebook/Ax/issues/953)
• Enable intersphinx mapping (https://github.com/facebook/Ax/issues/1227)
• PyCharm support for viz (https://github.com/facebook/Ax/issues/1729)
• Add annealing parameter to Models.THOMPSON (https://github.com/facebook/Ax/issues/2334)

Suggestions for setup changes:

• Migration of website to Docusaurus 2 (https://github.com/facebook/Ax/issues/466, status: blocked)

[rpanai]

@lena-kashtelyan I already built a conda recipe see PR whenever we fix the windows compatibility I'll let you know.

[rpanai]

@lena-kashtelyan it is now available on conda-forge https://anaconda.org/conda-forge/ax-platform

[lena-kashtelyan]

Wow, thank you so much, @rpanai!