Next: Connection Between the SUBARU Data Analysis System and the Observation Control System
Previous: The VLT Real Time Display
Table of Contents --- Search --- PS reprint


Astronomical Data Analysis Software and Systems V
ASP Conference Series, Vol. 101, 1996
George H. Jacoby and Jeannette Barnes, eds.

Automatic Sequencing for Experimental Protocols

Paul F. Hsieh, Ivan Stern

Smithsonian Astrophysical Observatory, Cambridge, MA 02138

Abstract:

We present a paradigm and implementation of a system for the specification of the experimental protocols to be used for the calibration of AXAF mirrors. For the mirror calibration, several thousand individual measurements need to be defined. For each measurement, over one hundred parameters need to be tabulated for the facility test conductor and several hundred instrument parameters need to be set. We provide a high level protocol language which allows for a tractable representation of the measurement protocol. We present a procedure dispatcher which automatically sequences a protocol more accurately and more rapidly than is possible by an unassisted human operator. We also present back-end tools to generate printed procedure manuals and database tables required for review by the AXAF program. This paradigm has been tested and refined in the calibration of detectors to be used in mirror calibration.

1. Introduction

The detailed mirror calibration is a critical component of the development of the Advanced X-ray Astrophysical Facility (AXAF). To this end, the AXAF program has scheduled a six month period of x-ray measurements at the X-ray Calibration Facility (XRCF) at Marshall Space Flight Center (MSFC). The calibration test conductor, TRW, will coordinate the activities of eight teams, each controlling separate instruments.

As one of the the participating teams, we were presented with a number of challenges in the generation and sequencing (or execution) of measurement procedures:

  1. For each measurement, we are required to provide the test conductor with a list of over 100 instrument parameters in order to coordinate activities with the other instrument teams.

  2. For each measurement, we are required to generate a complete, step-by-step procedure for Program review and approval. The procedure consists entirely of commands to computers and discrete operator actions.

  3. We need mechanisms to modify up-coming measurements to accommodate experimental findings, and ways of red-lining procedures during the performance of a measurement.

  4. We need a mechanism to insure that our procedures are followed accurately and efficiently.

We identified that (1) and (2) were two distinct representations of the measurement procedures that needed to be produced and kept in synch. We decided to define a third representation, a high level description of each measurement, from which both the step-by-step measurement procedure and the measurement parameter lists can be generated.

2. High Level Protocol Language (CCL)

The use of a high level procedure language allows numerous benefits. Most notably, it allows structural decomposition of the measurement process. Subtasks may be compartmentalized into subroutines; iterative and conditional language constructs may be used to exploit symmetries of the measurement process; derived quantities and experimentally measured constants may be encapsulated and used consistently.

For our ``Calibration Control Language'' (CCL) we chose to take the Free Software Foundation's GNU awk ( gawk) language and extend it with the functions listed in Table 1.

Two command extensions and two annotation extensions are sufficient to produce the detailed measurement procedures required by the AXAF program. These four commands generate entries in an intermediate tabulated list (CCI file) which contains all the steps and annotations of a measurement.

Four database extensions are sufficient to produce the Calibration Task List (CTL file), or measurement parameter list. An initial subroutine call defines all the fields of the CTL. Each measurement then sets the appropriate fields and writes a full parameter record at the end of the measurement.

3. Back End Programs

 
Figure 1: Example of paper procedure generated by cci2tex.

 
Figure 2: X-windows display generated by automatic sequencer cciexec.
Figure 2: PS 1.0 Mb

From the CCI file generated by the CCL program, we are then able to generate a detailed measurement procedure document. We are also able to use the CCI file to automatically sequence the measurement.

We have two back-end programs, cci2tex and cci2wp, which produce the step-by-step procedures in a format that meets MSFC requirements. cci2tex generates procedure documents using LaTeX as a mark-up language while cci2wp uses WordPerfect 5.1. Figure 1 shows a sample layout produced by cci2tex.

We are able to automatically sequence the CCI file using the cciexec program. cciexec executes every step of the measurement procedure in turn. It issues computer commands directly, thereby avoiding transcription error, and it prompts the operator to perform prescribed actions. cciexec allows steps to by skipped and inserted and it produces an as-run CCI file. Figure 2 shows the X-display presented by cciexec.

The CTL file generated by the CCL program is in a tab-delimited ASCII format that can readily be merged into the overall Calibration Task List.

4. Discussion

We have presented a paradigm for generating and sequencing experimental procedures. This paradigm consists of embedding simple intrinsic functions to a high level language and writing a handful of back-end programs.

Our choice of a high level language was influenced strongly by the UNIX environment under which the data acquisition environment is implemented. Our choice of the gawk language was one of expediency: gawk is readily available and familiar to several members of the AXAF team at SAO. We also considered perl, Tcl/Tk, C and Fortran. Each of these languages would bring its own strength, and the paradigm would work equally well.

We are currently using our paradigm in the calibration of detectors to be used in the mirror calibration. We have found that using a single high level description of a measurement to be valuable in the design and evaluation of experimental procedures.

5. Acknowledgements

This work was supported by NASA Contract NAS8-40224


Next: Connection Between the SUBARU Data Analysis System and the Observation Control System
Previous: The VLT Real Time Display
Table of Contents --- Search --- PS reprint
Wed Jul 3 07:46:51 MST 1996