harvest

harvest

Manually run harvesting.

Usage

grond harvest <rundir> [options]

Options

-h, --help

show this help message and exit

--force

overwrite existing harvest directory

--neach=NEACH

take NEACH best samples from each chain (default: 10)

--weed=WEED

weed out bootstrap samples with bad global performance. 0: no weeding (default), 1: only bootstrap chains where all NEACH best samples global misfit is less than the global average misfit of all NEACH best in all chains plus one standard deviation are included in the harvest ensemble, 2: same as 1 but additionally individual samples are removed if their global misfit is greater than the global average misfit of all NEACH best in all chains, 3: harvesting is done on the global chain only, bootstrap chains are excluded

--export-fits=EXPORT_FITS

additionally export details about the fit of individual targets. “best” - export fits of best model, “mean” - export fits of ensemble mean model, “ensemble” - export fits of all models in harvest ensemble.

--loglevel=LOGLEVEL

set logger level to “critical”, “error”, “warning”, “info”, or “debug”. Default is “info”.

--status=STATUS

status output selection (choices: state, quiet, default: state)

--parallel=NPARALLEL

set number of events to process in parallel, if set to more than one, –status=quiet is implied.

--threads=NTHREADS

set number of threads per process (default: 1). Set to 0 to use all available cores.