Optimization improvements and more robust input handling

This release provides new utilities that make it easier to specify and combine optimization problems. See the following release notes for details.

What's Changed

• Bumping new mira version by @sabinala in #553
• 556 pattern matching on var finds multiple options by @djinnome in #557
• Myo map heatmap by @marjoleinpnnl in #519
• Add template for max value of QoI within simulated time by @anirban-chaudhuri in #561
• Simplify interchange dictionary processing by @SamWitty in #564
• optimize with fixed set of interventions along with optimizing over specific interventions by @anirban-chaudhuri in #563
• Set the minimum value for the NormalNoiseModel noise variance` by @SamWitty in #572
• Adding support for ensemble output format by @anirban-chaudhuri in #570
• Update setup.cfg to include correct version by @SamWitty in #575
• Add parameters to observables by @SamWitty in #580
• adding template for start time and parameter value intervention for optimize by @anirban-chaudhuri in #585
• Add support for many more MIRA probonto distributions by @SamWitty in #593
• Updated optimize to handle combination of multiple intervention templates by @anirban-chaudhuri in #591
• Allow for interventions on constant parameters by @SamWitty in #597
• Updating optimize to handle multiple risk constraints by @anirban-chaudhuri in #596
• changing interfaces line 535 to include observables by @sabinala in #602
• 477 add progress hook to optimize by @sabinala in #592
• Bump Mira dependency to 0.7.0 by @SamWitty in #603
• added schema.vg.json and heatmap/ to gitignore by @SamWitty in #605
• Coerce AMR inputs to float by @SamWitty in #604

Full Changelog: v0.1.5...v0.1.6

Add informative error messages

This release adds informative error messages for common errors resulting from numerically unstable parameters/models, missing components of the model, etc.

What's Changed

• Add check if parameters have distributions by @SamWitty in #543
• Throw an error in calibrate if no parameters have distributions by @SamWitty in #546
• Updated optimize documentation for static_parameter_intervention definition by @anirban-chaudhuri in #544
• Added warnings for optimize by @anirban-chaudhuri in #542
• Add more informative error if adaptive time step underflows by @SamWitty in #549
• adding INSTALL.md by @sabinala in #547
• Check whether compilation or evaluation of Mira model is succcessful by @SamWitty in #545
• add more in-depth README by @SamWitty in #551

Full Changelog: v0.1.4...v0.1.5

Performance and Expressiveness Improvements to `optimize`

This release introduces the ability to intervene on multiple parameters at the same point in time using the optimize interface, as well as substantially improves its computational performance.

What's Changed

• 506 add trajectory visualization+timepoints tests by @marjoleinpnnl in #520
• Update optimize to be more computationally efficient and add separate notebooks for optimize and sample by @anirban-chaudhuri in #529

Full Changelog: v0.1.3...v0.1.4

Calibrate Ensembles

This release includes ensemble_calibrate, as well as some improvements to error messages, CI, and visualizations.

What's Changed

• Fix calibrate to work on observables as well as state by @SamWitty in #517
• adding testing check_data function to load_data by @sabinala in #458
• Bump mira to 0.5.0 by @SamWitty in #523
• 438 add various interventions to interfacesipynb by @sabinala in #493
• revert removal of notebook test from CI by @SamWitty in #531
• Ensemble calibrate by @SamWitty in #522
• Add an error if euler solver is used without specifying the timesteps. by @SamWitty in #533
• 507 add reference lines to trajectory plot by @marjoleinpnnl in #521
• Robustify notebook test CI to allow notebook outputs by @SamWitty in #536
• adding ensemble calibrate example by @sabinala in #537

Full Changelog: v0.1.2...v0.1.3

Minor fixes for Feb 2024 ASKEM Hackathon

This minor release adds functionality necessary to complete the Feb 2024 ASKEM Hackathon scenario problems.

What's Changed

• Bump mira to 0.4.0 by @SamWitty in #510
• Add parallel sampling back to PyCIEMSS by @SamWitty in #509
• Jensen-Shannon score of 0.15 observed for some hardware choices. by @djinnome in #482
• Updating optimize to add intervention builder that can handle optimizing the start time by @anirban-chaudhuri in #512
• Add parameter interventions to the trace for downstream reporting by @SamWitty in #511

Full Changelog: v0.1.1...v0.1.2

Use ChiRho Dynamics Backend (w/fixes)

What's Changed

• Add progress hook (and example usage in test) for calibrate by @SamWitty in #445
• Add parallel sampling when there are no interventions by @SamWitty in #446
• Adding labels and calibrate interface example by @sabinala in #463
• Visualizations post ChiRho by @JosephCottam in #469
• Optimize interface refactor by @anirban-chaudhuri in #442
• writing test for calibrate with data mapped to observables by @sabinala in #450
• Added risk estimates as an output to results generated from sample interface by @anirban-chaudhuri in #504

Also various bug fixes.

Full Changelog: v0.1.0...v0.1.1

Use ChiRho Dynamics Backend

This large refactoring PR replaces built-in functionality for simulating, calibrating, and intervening in dynamical systems and replaces it with ports to the more robust and feature-rich ChiRho dynamical systems module. Now, PyCIEMSS is only an interface between the ASKEM program and the ChiRho dynamical systems repo.

What's Changed

• Refactor PyCIEMSS to use ChiRho and remove stale functionality by @SamWitty in #396

Full Changelog: v0.0.3...v0.1.0

v0.0.3

• Additional progress hooks by @anirban-chaudhuri (#365)
• Parameter density estimates on demand by @SamWitty (#358)
• Visualization can export SVG or PNG by @marjoleinpnnl (#364)
• pip install improvements by @JosephCottam (#372)

v0.0.2

What's Changed

• Create GitHub actions TA4 integration workflow by @SamWitty in #346
• Visuals rearrange by @JosephCottam in #320
• Added unit test for convert_to_output that checks if observables detach tensor by @djinnome in #351
• bumped mira to use the tagged version by @SamWitty in #354
• Extracted make_quantiles from convert_to_output by @djinnome in #353
• Vega png export issue by @JosephCottam in #364
• Add deterministic_learnable_parameters to docstring by @djinnome in #363
• Extract posterior densities from inferred parameters at user-specified points by @SamWitty in #358
• Updated TA4 interfaces to return inferred parameters for OUU by @anirban-chaudhuri in #367
• Adding progress hook to OUU optimizer by @anirban-chaudhuri in #365
• Bump pyciemss version by @SamWitty in #370

Full Changelog: v0.0.1...v0.0.2

Initial Evaluation Release

This release includes all of the PyCIEMSS functionality that was used for the Phase 1 July Evaluation.