Parent issue for variability support
willu47 opened this issue · 1 comments
Goal
We want to propagate an ensemble of e.g. weather scenarios through a system-of-systems model to assess the resilience of the system-of-systems
Ideally, this could be available to the user at the command-line, with the subcommand smif prepare
and arguments [modelrun_template] [scenario_name] [variant range]
, or perhaps some useful helpers for specifying a range of scenario variant names like --start weather_001 --end weather_100
Related goal
We want to perform a global sensitivity analysis, and have generated a sample of input data we wish to propagate through the system-of-systems to assess sensitivity of one or more model outputs to the inputs.
Core functionality
- Write a script which generates one scenario variant for each replicate in the ensemble #362
- Write a script which generates one model run for each scenario variant in the ensemble #363
- Perform the batch model run #179
- Obtain and collate results (extension of #350)
- Add command line argument (#364)
Alternatively, the steps could be:
- Add the ensemble as a dimension within a smif scenario
- Autogenerate a batch of model runs, one for each ensemble realisation
- perform batch model run
- obtain and collate results
Issues #362 , #363 and #364 have been addressed within branch variability
.
Scripts addressing issues #362 and #363 can be found in
-
src/smif/sample_project/write_scenario_variants.py
(#362) -
src/smif/sample_project/write_variant_model_runs_from_template.py
(#363)
Currently those scripts are merely drafts for two functions with the same names in cli/__init__.py
, which the smif prepare
command actually invoke (see comment in #364)
Note - For the purpose of testing, I slightly modified the energy_central
model run definition, adding the weather_at_home
scenario.