Документ взят из кэша поисковой машины. Адрес оригинального документа : http://www.adass.org/adass/proceedings/adass96/reprints/gaffneyn.pdf Дата изменения: Wed Jan 14 23:26:00 1998 Дата индексирования: Tue Oct 2 14:54:18 2012 Кодировка: Поисковые слова: http astrokuban.info astrokuban

Astronomical Data Analysis Software and Systems VI ASP Conference Series, Vol. 125, 1997 Gareth Hunt and H. E. Payne, eds.

Planning and Scheduling Software for the Hobby· Eberly Telescope
Niall I. Gaffney Hobby · Eberly Telescope, RLM 15.308, University of Texas at Austin, Austin, TX 78712 E-mail: nial l@astro.as.utexas.edu Mark E. Cornell McDonald Observatory, RLM 15.308, University of Texas at Austin, Austin, TX 78712 E-mail: cornel l@puck.as.utexas.edu Abstract. We present the initial design of the planning and scheduling software for the 9.2 meter Hobby· Eb erly Telescop e (HET). These tools are to b e used by the PIs to prepare prop osals for telescop e time and to construct observing plans to b e executed by the HET's observing queue, and by the astronomers running the telescop e to evaluate and schedule the prop osed queued observations. We also outline our model for the op eration mode of the HET in queued observing mode.

1.

Introduction

The HET1 (Sebring & Ramsey 1994) is a 9.2 meter telescop e located at McDonald Observatory near Ft. Davis, Texas. To reduce construction costs, the telescop e is fixed in altitude at 55 . By changing the azimuth of the telescop e, different declinations can b e reached. Ob jects can then b e tracked with an Arecib o-style tracker. Because of this geometry, the HET can access only a limited range of hour angle at any given declination and can track an ob ject for roughly an hour at a time. Thus, efficient use of the telescop e will rely strongly on software tools to know when and how an ob ject can b e observed and how b est to sequence observations over the course of a night. Once completely op erational, 85% of the nights are exp ected to b e used in a queued mode where the night is dynamically scheduled as observing conditions change. In this mode, data are acquired by resident astronomers on b ehalf of, p otentially, many different PIs for many different observational pro jects each night. 2. Op erations

The op erations of the HET (Kelton & Cornell 1994; Kelton, Cornell, & Adams 1996) is four phased: a prop osal phase, a planning phase, an observing phase,
1

The HET is operated by McDonald Observatory on behalf of the University of Texas at Austin, the Pennsylvania State University, Stanford University, Ludwig-Maximilians Universitat Munchen, and Georg-August Universitat GЁttingen. Ё o

379

© Copyright 1997 Astronomical Society of the Pacific. All rights reserved.


380

Gaffney and Cornell

and a verification phase. The first two phases take place three times p er year, when the database driving the queue for the telescop e is constructed, while the last two are executed every night data are taken. The prop osal phase consists of the traditional PI-TAC interaction, where the PI, using our planning tools, creates a telescop e prop osal which is reviewed and granted time by the TAC. In this phase, the PI needs to b e able to predict approximately the ability of the HET to observe a typical ob ject under nominal conditions. In the planning phase, the TAC informs the HET op erations team of the time allocated to the prop osal. The HET op erations team then gives the PI an account on the op erations database server. The PI comp oses a plan, which is submitted to the database server via e-mail. The plan is checked via a procmailstyle e-mail system. In this phase, the PI requires tools to determine how to make observations under b oth nominal and degraded conditions. Further, the PI must b e able to determine what constraints need to b e set for the observations, and what effect these may have. The observing phase b egins with the op erations team compiling a master database of observing pro jects. This database is then shipp ed to the telescop e where resident astronomers compile a plan for each night based on the current and forecasted conditions. They then execute this plan, make real time changes to it as conditions change, and gather data. Data are then shipp ed back to Austin via a KPNO "Save The Bits" style queued ftp transfer protocol (Seaman & Bohannan 1996). For this phase, a manual or automatic scheduling tool is needed to determine what to do tonight, based on the current and forecasted conditions, time allocation shares remaining, TAC rankings, and other constraints. The verification phase is one in which the PI determines whether the data are as intended. The data are transfered to Austin, where they are bundled for each PI. The PIs are informed by e-mail of new data, which they then retrieve via ftp. Alternately, the PI may use a ftp mirroring package to retrieve new data automatically. The PI then examines the data and informs the op erations team of any needed changes in the plan. This phase requires only that PIs b e able to retrieve data via ftp and use their own data analysis tools to examine the data. 3. Software Design

Our system contains b oth planning and scheduling tools. Planning tools are used to determine the feasibility of observations to b e made with the HET. Scheduling tools are used to sequence observations during a night and over the course of an observing tri-semester. These tools are b eing used by b oth PIs and the op erations team to schedule observations on the telescop e during all four phases of op erations. We have develop ed tools under SunOS and Solaris op erating systems, using freely available packages, to allow simple p orting to other UNIX systems. We have embraced two fundamental concepts in our software development: (i) the tools that will b e used by the telescop e op erators/schedulers are the same as those used by the PIs, and (ii) that the tools should b e easily modifiable as the optimal method of op erating this new system b ecomes apparent. Thus, we have written simple tools which can b e linked together to predict and schedule the telescop e for any observing plan. Further, we have written Tcl/Tk (Welch


Planning and Scheduling for the Hobby· Eb erly Telescop e

381

Figure 1. Two HET tools, the mirror tracking tool, which shows a p otential track of the tracker across the primary mirror (left) and a Signal-to-Noise estimator (right). The tracking tools shows the TkSteal extension capturing a PGPLOT window and incorp orating it into a single window environment while the Signal-to-Noise tool exemplifies the functionality of a TCL/TK GUI running over our base tools.

1994) graphical overlays to assist the user initially in creating plans. The main b enefit of this system is that a user, once familiar with the system, can then create scripts in any scripting language which has access to system calls, to make plans for an entire observing program, thus reducing the time required to process a great numb er of ob jects in a GUI environment. We used the TkSteal extension (Delmas 1996) to Tk 4.0 to minimize the numb er of windows used by each tool. This extension allows the Tcl/Tk script to "steal" any X-window spawned external to the Tcl/Tk script, and emb ed it in the Tk window as if it were an indep endent widget. By "stealing" windows spawned by PGPLOT (Pearson 1996), rather than creating a Tcl/Tk widget that interfaces with the PGPLOT or other graphics library, we can create a tool that is not tied to the X-windowing environment, minimize the numb er of windows, and allow the program that spawned the window to retain interactions with that window without any additional Tcl/Tk code. This allows us to present the user with a complete GUI in a single window, with graphics that are created by a program external to the Tcl/Tk script.


382 4.

Gaffney and Cornell Pro ject Progress Monitoring

We have also implemented a scheme to allow PIs to retrieve and examine data, and to prop ose future observations, in a secure environment. Each PI is given a password-protected WWW/ftp account from which data and status information can b e accessed. Thus, PIs can actively monitor their pro jects via the WWW. Further, they can access public statistics for previous nights, such as partner share, which pro jects got data, weather/seeing conditions, and instrument availability. However, the PI's password is required to access all sensitive information, such as ob ject names, p ositions, and setups for observations. A more primitive method of monitoring involves a modified GNUfinger Perl script, which checks to see if the PI has new data, much as the standard finger script checks for new mail. PIs may also passively monitor the progress of their pro ject, since an e-mail notification will b e sent to the PI of each pro ject that acquired data in the previous night. Thus, PIs need not worry ab out their pro ject until data are acquired. Finally, PIs will b e allowed to set up a ftp mirroring (McLoughlin 1994) script that will automatically retrieve any data that had not b een previously retrieved. This script may b e run as frequently as daily, or as infrequently as once a week. Acknowledgments. We are thankful to the rest of the HET op erations team memb ers, Mark Adams, Tom Barnes, Ed Duthcover, Earl Green, George Grubb, and Phil Kelton. Many thanks go out to Frank Ray for providing us with the initial work on tracking ob jects with the HET. Further, we would like to thank the Pro ject Scientist, Larry Ramsey, and the myriad of other astronomers who have tested our tools and provided us with useful suggestions. References Delmas, S. 1996, TkSteal2 Kelton, P. W., & Cornell, M. E. 1994, in A. S. P. Conf. Ser., Vol. 79, Rob otic Telescop es, ed. G. W. Henry & M. Drummon (San Francisco: ASP), 136 Kelton, P. W., Cornell, M. E., & Adams, M. T. 1996, in A. S. P. Conf. Ser., Vol. 87, New Observing Modes for the Next Century, ed. T. Boroson, J. Davies, & I. Robson (San Francisco: ASP), 33 McLoughlin, L. 1994, ftp mirroring software (mirror.pl), details and code are available at ftp://src.doc.ic.ac.uk/computing/archiving/mirror/ Pearson, T. J. 1996, PGPLOT User Manual3 Seaman, R., & Bohannan, B. 1996, in ASP Conf. Ser., Vol. 101, Astronomical Data Analysis Software and Systems V, ed. G. H. Jacoby & J. Barnes (San Francisco: ASP), 432 Sebring, T. A., & Ramsey, L. W. 1994, in Advanced Technology Optical Telescop es V, SPIE Tech. Conf. 2199 Welch, B. 1994, Practical Programming in Tcl and Tk (New York: Prentice Hall)
2 3

http://panther.cimetrix.com/sven/tksteal.html http://astro.caltech.edu/tjp/pgplot/