gbt dynamic scheduling system (dss)

GBT Dynamic Scheduling System (DSS)

Dana Balser, Jim Braatz, Mark Clark, Jim Condon, Ray Creager, Mike McCarty, Ron Maddalena, Paul Marganian, Karen O’Neil, Eric Sessoms, Amy Shelton

Nomenclature

Butler

GBTOpen SessionsWindowed SessionsFixed Sessions

Observing

Scheduling Algorithm24-48 Hours in Advance

Scheduling Probabilities

Post-ObservationReports

Data CollectionAdvance of Semester

Reports of schedule,logs, time lost, etc

Phase IIdata

collection

Sensitivitycalculator

Timeavailabilityprediction

ProposalSubmission

Tool

Sciencegrades

How does it work?

Historicalprobabilities

Weatherforecasts

Monitor schedule

BackupProject run

Observingscripts

AutoschedulerRun

Scheduler modifiesand approves

Notificationsent

Provided outside the DSS

Provided by the DSS

Requirements• Scheduling observers, not

scripts•Observers retain control •Minimum of 24 hours advance notice for observers• Wide array of hardware•Cannot increase workload of staff or observers

Atmospheric Effects

Condon & Balser (2011)

Maddalena

Weather Forecasts

Atmospheric Stability

Maddalena; Balser (2011)Pyrgeometer: non-imaging device sensitive to 4.5-40 micron over 150 deg fov.

Wind Effects

Condon (2003)

21 v

21 v

Weather Forecasts: wind

Obs

Win

d S

peed

(m/s

)

Forecast Wind Speed (m/s)

Day

Night

Balser (2010); Maddalena

Solar Heating

rms330 micron

rms220 micron

Nikolic et al. (2007)

SurfaceWave front errors from OOF maps.

Grayscale: +/- 2 radContour: ½ rad intervals

Primary SurfaceDay: 300 micronNight: 250 micron

Offset PointingDay: 3.3 arcsecNight: 2.7 arcsec

R = (ηS Pαβ Pν

γ) (leff lHA lz ltr lst) (foos fcom fsg ftp)(tttntlettb)

Scoring Equation

Weather:

Observing EfficiencyStringency

Pressure Factors:

Right Ascension Frequency

Performance Limits:

Observing Efficiency Hour AngleZenith Angle

Tracking ErrorAtmospheric Stability

Other Factors:

Observers on SiteCompletion of Projects

Science GradesThesis Projects

Temporal Constraints:

Transit Nighttime

LST ExclusionTime Between

R = (ηS Pαβ Pν

γ) (leff lHA lz ltr lst) (foos fcom fsg ftp)(tttntlettb)

Scoring Equation

Weather:

Observing EfficiencyStringency

Observing Efficiency

s m 15.3

|v| arcsec

and ),/( f where;f)2ln(41

4ln(2)f 1

and ,])exp[-(4 where;) ( 32-exp

exp T)exp( T

4

1-2o

2tr

tr

2

2

2

tr

2asur

2a

2n

2d2

2

sur

2

sys

sysatm

trsuratm

Observing Efficiency

Stringency

R = (ηS Pαβ Pν

γ) (leff lHA lz ltr lst) (foos fcom fsg ftp)(tttntlettb)

Scoring Equation

Pressure Factors:

Right Ascension Frequency

Pressure Factor

d

nln 1 P

R = (ηS Pαβ Pν

γ) (leff lHA lz ltr lst) (foos fcom fsg ftp)(tttntlettb)

Scoring Equation

Performance Limits:

Observing Efficiency Hour AngleZenith Angle

Tracking ErrorAtmospheric Stability

Observing Efficiency Limit

Hour Angle Limit

Condon & Balser (2011)

Tracking Error Limit

errors)flux (10% 0.20 / f tr

Balser (2011); Mason & Perera (2010)

Atmospheric Stability Limit

Packing (Open Sessions)

Problem: a thief with a bag of capacity N, faced with a number (M) of possible goodies each having a different weight (cost) and value, how do you pack your bag to maximize your take?

Brute Force: order (M!)Knapsack Algorithm: order (M*N)

N = number of quarter hours to scheduleM = number of potential sessionsOverhead = 15 min.

Sessoms

Scheduler Page

Schedule

User Home Page

User Project Page

Fini