Support for running simulations of the main LSST survey, primarily for prenight or progress evaluations.
There is a pyproject.toml file here, but ts_fbs_utils is not packaged outside of the lsst-ts conda channel, and ts_config_scheduler is only available via git clone.
One option for installation of this package is as follows (note directions for handling $RUBIN_SIM_DATA_DIR can be found in rubin_scheduler and rubin_sim data download directions):
pip install git+https://github.com/lsst-sims/lsst_survey_sim
scheduler_download_data --update
The advantage of the above is that you will have the appropriate versions of ts_fbs_utils and rubin_scheduler that should be in use at the summit and it's very relevant for running command-line simulations.
When running in the RSP, or if you want to access the notebooks in the 'notebook' directory - using git clone works better.
In the RSP or for running with the notebooks:
git clone [email protected]:/lsst-sims/lsst_survey_sim.git
cd lsst_survey_sim
pip install -e . --no-deps --no-build-isolation
And then at the USDF RSP:
os.environ["RUBIN_SIM_DATA_DIR"] = "/sdf/data/rubin/shared/rubin_sim_data"
or follow the instructions to download the relevant rubin_scheduler and rubin_sim data as above.
The only required packages for lsst_survey_sims that are not provided in the RSP are ts_fbs_utils. In general, installing and using develop
of ts-fbs-utils is fairly safe for new simulations, but checking the version in the lsst-survey-sims dependencies can be helpful. Either of
pip install --user git+https://github.com/lsst-ts/ts_fbs_utils
or
git clone [email protected]:lsst-ts/ts_fbs_utils.git
cd ts_fbs_utils
pip install -e . --no-deps --no-build-isolation
are suitable.
It is possible that in the RSP the version of rubin_scheduler could not match the summit. Most of the time this will be fine; adding a pip installed editable version of rubin_scheduler is also possible however.
The configurations for the FBS are kept in ts_config_scheduler which can be independently cloned
git clone [email protected]:lsst-ts/ts_config_scheduler.git
or will be cloned for you when running the demo notebook. In general, ts_config_scheduler can be pointed to the tip of develop; sometimes a temporary run branch is deployed for patches during the night.
An example of running a simulation is shown in notebooks/lsst_eval.ipynb.
A notebook similar to this is available in Times Square.