stats

Summary:

Nightly summary statistics, recorded by pass, from a surveysim run.

Naming Convention:

stats_surveysim.fits

Regex:

stats_surveysim\.fits

File Type:

FITS, 810 KB

Note: currently this is an output of surveysim, but in the future may become an output of real operations survey QA.

Contents

Number	EXTNAME	Type	Contents
HDU0		IMAGE	Blank.
HDU1	STATS	BINTABLE	Nighly summary statistics by pass.

FITS Header Units

HDU0

EXTNAME = PRIMARY

Empty HDU.

HDU1

EXTNAME = STATS

Survey summary statistics by pass.

Required Header Keywords

KEY	Example Value	Type	Comment
NAXIS1	1826	int	length of dimension 1
NAXIS2	13	int	length of dimension 2
TILES	desi-tiles.fits	str	Name of the tiles file specified in desisurvey config.
START	2020-03-15	str	YEAR-MM-YY of the first night of the survey.
STOP	2020-04-15	str	YEAR-MM-YY of the most recent night with statistics recorded.

Required Data Table Columns

Name	Type	Units	Description
MJD	float64		MJD of local noon before this night.
tsched	float64	d	Total scheduled time for all programs during this night.
topen	float64[3]	d	Actual time dome was open during this night, per program.
tdead	float64[3]	d	Deadtime during this night (dome open, but idling), per program.
tscience	float64[8]	d	Time spent with the shutter open for science exposures, per pass.
tsetup	float64[8]	d	Time spent setting up to observe a new field, per pass.
tsplit	float64[8]	d	Time spent setting up to reobserve the previous field, per pass.
completed	int32[8]	d	Number of tiles completed this night, per pass.
nexp	int32[8]	d	Number of exposures recorded this night, per pass.
nsetup	int32[8]	d	Number of setups to observe a new field, per pass.
nsplit	int32[8]	d	Number of setups to re-observe a previous field, per pass.
nsetup_abort	int32[8]	d	Number of times a new field setup was aborted, per pass.
nsplit_abort	int32[8]	d	Number of times a repeat field setup was aborted, per pass.

The table contains one row per night, with row N corresponding to N nights after the header START date.

All timing statistics are in units of days.

The topen and tdead arrays are indexed by the number of programs, which are defined in the tiles module. There will always be space for all three programs (DARK, GRAY, BRIGHT), even when using a custom tiles file with fewer programs actually used.

The remaining arrays are indexed by the number of passes, which depends on the actual tiles file being used. In general, a tiles file uses arbitrary pass numbering, which might not be dense or consecutive for each program. The Tiles object defines data structures to map the indices used here to the pass numbers and programs used in the tiles file.

Notes and Examples

A SurveyStatistics object tracks survey statistics during simulation:

import surveysim.stats
stats = surveysim.stats.SurveyStatistics()

Its internal state after a simulation (or each night) can be saved using, for example:

stats.save('stats.fits', comment='Baseline (seed=1)')

This state can then later be restored using:

stats = surveysim.stats.SurveyStatistics(restore='stats.fits')