Allocating Time on the Hubble Space Telescope

1. Allocating Time on the Hubble Space Telescope Olivia Gillham

2. Contents • What is Hubble Space Telescope? • Current scheduling framework • Formalization as a scheduling problem • Optimization of basic problem • Additional issues • Recommendations

3. What is Hubble Space Telescope? • 2.4 meter mirror collects light (and some forms of radiation), directs it to scientific instruments • Orbits 569 km above the Earth • Solar-powered • Location above Earth’s atmosphere provides unusual clarity - no distortion or absorbtion • Completes an orbit every 97 minutes

4. Who gets to use Hubble? • Anyone can apply for time on Hubble • Proposals are submitted and reviewed annually • Proposals are accepted based on their value to the scientific community • Mostly professional astronomers, but amateurs are occasionally approved

5. How are proposals selected? • Time Allocation Committee(TAC) composed of international experts selects proposals based on their value to the scientific community • TAC tries to select even proportions of long, medium, and short projects • Director of Space Telescope Science Institute (STScI) approves selection

6. How are projects scheduled? • STScI scheduling experts form a tentative long-range schedule for the year • Detailed schedules are created ~ 3 weeks ahead of time and transmitted to Hubble weekly

7. Contents of a Phase I Proposal • A summary of the observations requested • The scientific justifications for the project. • A list of targets with their celestial coordinates. • The instrument(s) to be used with desired modes, filters, and dispersers. • An estimate of the number of spacecraft orbits needed to accomplish the observing program

8. Basic Formalization • Single machine • pj = # orbits required • wj = scientific value • All jobs must be completed within 1 year, so dj = D = 1 year = 5422 orbits • Maximize value => minimize weight of tardy jobs = sum of wjUj. • 1/ dj=D /sum wjUj = knapsack! • Dynamic algorithm runs in O(nD)~O(5.5mil)

9. Complications • “Snapshot” proposals requiring 45 minutes or less (<1/2 orbit) are prioritized • Targets of Opportunity(TOO) occur at unknown times • Unforeseen celestial events • Set-up times (to lock on to new coordinates) vary • STScI Director allots 10% of total time (usually to TOO and projects too large to be approved by the committee)

10. Snapshots can be treated as normal jobs, pj ≤ .5 • Targets of Opportunity have unknown rj, dj, but wj and pj are known, so they can be included in the selection process • Unforeseen events have unknown parameters - online scheduling scenario • Weekly programming means we can address TOO and unknown jobs dynamically, as they come up • Create original schedule for 90% of total time, or 4880 orbits

11. Further Complications • Some jobs can only be completed at certain times of the year (differing rj and dj) • Set-up time between jobs varies (locking onto guidance stars) • Both can be incorporated into the formal model as stochastic elements, but complicate the solution enormously

12. Blue box = set-up White box = observation Pink bars = target blocked Red bars = guide star blocked

13. Preliminary Recommendations • Re-allocate responsibility so that TAC expert committee assigns weights, STScI scheduling staff selects and schedules • Include snapshots, TOOs, and “too long” projects in weighting and selection • Dynamic knapsack algorithm on 4880 orbits provides initial “optimal” solution • Use Director’s time primarily for TOOs and unexpected events, possibly reducing from 10%

14. Q&A

15. References • Villard, Ray “The Hubble Space Telescope Observing Program.” http://www.nasa.gov/mission_pages/hubble/servicing/series/How_science_is_done.html • “Hubble Essentials.” Hubble Site http://hubblesite.org/the_telescope/hubble_essentials/ • Space Telescope Science Institute, Hubble Space Telescope. http://www.stsci.edu/hst/