Future Features
On this page, we maintain a list of planned future features of bayesim. They’re listed under a variety of categories, in loosely descending order of priority within each.
Visualization
- visualizing model uncertainty on parameter grid
- visualizing measured vs. simulated data with experimental and model uncertainty
- option to generate a plot of which particular observed data was used in an inference run
Other New Capabilities
- option for entropy-based thresholding as well as th_pm and th_pv in
run() function
- capability to pass a DataFrame objects to
attach_model() and attach_observations() functions instead of just filepaths
- multiple output variables
- completely “closed-loop” version of
run() function that does subdivision as well (need a way to save a handle to a model function for running new simulations)
- interpolation of simulated data
- alternative error models
- alternative initial sampling states other than grids
Efficiency/Speedup
- speed up
list_model_pts_to_run() function
- do some more code profiling to identify other particularly slow functions/components
- more parallelism generally
Interfaces
- finish developing command-line interface
- build a GUI!