Module: run_search#

class kbmod.run_search.SearchRunner(config=None)[source]#

A class to run the KBMOD grid search.

Attributes:

configSearchConfiguration: The configuration parameters.
debugbool: If True, enable debug logging (and additional computation).
phase_timesdict: A dictionary mapping the search phase to the timing information, a list of [starting time, ending time] in seconds.
phase_memorydict: A dictionary mapping the search phase the memory information, a list of [starting memory, ending memory] in bytes.
timeoutfloat or None: The time at which the search should timeout, in seconds since the epoch. This is a soft timeout that will not interrupt during a processing stage. None means no timeout is set.

Methods

`apply_config`(config)	Apply the configuration parameters to the search runner.
`display_phase_stats`()	Output the statistics for each phase.
`do_core_search`(config, stack, trj_generator)	Performs search on the GPU.
`load_and_filter_results`(search, config[, ...])	This function loads results that are output by the grid search.
`run_search`(config, stack[, trj_generator, ...])	This function serves as the highest-level python interface for starting a KBMOD search given an ImageStack and SearchConfiguration.
`run_search_from_work_unit`(work)	Run a KBMOD search from a WorkUnit object.

apply_config(config)[source]#

Apply the configuration parameters to the search runner.

This function is designed to be called at multiple points allow it to be used regardless of which level of the search is being run.

Parameters:

configSearchConfiguration: The configuration parameters

display_phase_stats()[source]#: Output the statistics for each phase.

do_core_search(config, stack, trj_generator)[source]#

Performs search on the GPU.

Parameters:

configSearchConfiguration: The configuration parameters
stackImageStackPy: The stack of image data.
trj_generatorTrajectoryGenerator: The object to generate the candidate trajectories for each pixel.

Returns:

keepResults: The results.

load_and_filter_results(search, config, batch_size=100000)[source]#

This function loads results that are output by the grid search. It can then generate psi + phi curves and perform sigma-G filtering (depending on the parameter settings).

Parameters:

searchkbmod.search: The search function object.
configSearchConfiguration: The configuration parameters
batch_sizeint: The number of results to load at once. This is used to limit the memory usage when loading results. Default is 100000.

Returns:

keepResults: A Results object containing values from trajectories.

run_search(config, stack, trj_generator=None, workunit=None, extra_meta=None)[source]#

This function serves as the highest-level python interface for starting a KBMOD search given an ImageStack and SearchConfiguration.

Parameters:

configSearchConfiguration: The configuration parameters
stackImageStackPy: The stack of image data.
trj_generatorTrajectoryGenerator, optional: The object to generate the candidate trajectories for each pixel. If None uses the default EclipticCenteredSearch
workunitWorkUnit, optional: An optional WorkUnit with additional meta-data, including the per-image WCS.
extra_metadict, optional: Any additional metadata to save as part of the results file.

Returns:

keepResults: The results.

run_search_from_work_unit(work)[source]#

Run a KBMOD search from a WorkUnit object.

Parameters:

workWorkUnit: The input data and configuration.

Returns:

keepResults: The results.

kbmod.run_search.append_positions_to_results(workunit, results)[source]#

Appends predicted RA, Dec positions to the results table.

Parameters:

workunitWorkUnit: The WorkUnit with all the WCS information.
resultsResults: The current table of results including the per-pixel trajectories. This is modified in-place.

kbmod.run_search.check_gpu_memory(config, stack, trj_generator=None)[source]#

Check whether we can run this search on the GPU.

Parameters:

configSearchConfiguration: The configuration parameters
stackImageStackPy: The stack of image data.
trj_generatorTrajectoryGenerator, optional: The object to generate the candidate trajectories for each pixel.

Returns:

validbool: Returns True if the search will fit on GPU and False otherwise.

kbmod.run_search.configure_kb_search_stack(search, config)[source]#

Configure the kbmod SearchStack object from a search configuration.

Parameters:

searchkb.StackSearch: The SearchStack object.
configSearchConfiguration: The configuration parameters