Getting Started Documentation Glish Learn More Programming Contact Us
Version 1.9 Build 1367
News FAQ
Search Home

AIPS++ Quarterly Report: 1999 Q3

T.J. Cornwell, NRAO

1999 November 23


Contents

Introduction

This quarter, we have focused almost exclusively on preparing the first release. This included testing the system, fixing defects, issuing two patches, preparing and testing the actual CDROM release, and distributing the final release. The official release date was October 4, and was made at the ADASS meeting in Hawaii.

We are now attempting to establish a 6 month plan-develop-test-release cycle. To start the next cycle, the development plan for the next release (the next one: 1.3 targeted for April 2000) was revised, and targets assigned.

Developments in 1999 Q3

In Single Dish support, work was slowed this quarter due to absence of Joe McMullin from the group for over two months. The work done by Jim Braatz, Joe McMullin and Bob Garwood continues to be focussed on support of the GBT and development of the DISH environment for general single dish analysis in AIPS++. There continue to be regular visits by Garwood and McMullin to Green Bank and Braatz to Charlottesville.

Jim Braatz spent much of this quarter learning his way around AIPS++ and learning about his Green Bank specific duties from Joe McMullin. In particular, the tipper software was transferred from McMullin to Braatz as was the pointing software. Braatz and McMullin continue to work on transferring responsibility for the commissioning tools.

There was a second test of the GBT holography backend this quarter. Garwood fixed several defects reported in the GBT filler after the first test in preparation for this test. In addition, Braatz developed AIPS++ scripts to verify the integrity of the data during these test. The tests went well as far as AIPS++ was concerned although as of that test, the majority of the post-observation data reduction was done in UniPOPS. Braatz has worked on converting the UniPOPS scripts to AIPS++ and the results of this work (which is nearly complete) was presented at ADASS in October. Garwood continues to work to enhance the GBT filler. In addition to the fixed defects noted above, the filler was turned into a true AIPS++ distributed object in preparation for its use in an on-line setting. The remaining Green Bank work was concentrated on weaning a few engineers off of a very old installation of AIPS++. The interface to several utilities, especially plotting, has drifted somewhat over the past year such that these older scripts did not work with the current installation. This work has been completed and the older installation removed from Green Bank. Finally, the format for the GBT spectrometer backend FITS files was finalized.

The non-GBT focus this quarter was primarily on responding to reported defects and enhancement requests from users of DISH in the AIPS++ pre-release. The current state of DISH was summarized in a poster presented at ADASS in October.

Work continued on single dish imaging tools. NRAO 12m On-the-fly data was successfully converted from UniPOPS to an AIPS++ MeasurementSet. This data set was used by Tim Cornwell in the first successful imaging of single dish OTF data in AIPS++. These results were presented at the October ADASS meeting.

A poster summarizing the single dish FITS binary table convention was prepared this quarter for presentation at ADASS.

In Synthesis support, this quarter is unique in that it has been significantly concerned with preparations for the first release which was issued on October 4. As a result, general synthesis development was tapered off over the quarter in favor of CDROM testing, the identification and correction of code defects, the preparation of high-level documentation and an associated public outreach effort to scientific users. These efforts have been successful and represent a significant transition in the focus of the project, and the operational status of the synthesis code in particular. This is in keeping with previous planning in this area in which this was identified as a key objective. We have reached more scientific users in this quarter, and this focus will continue as a high priority for the foreseeable future. Synthesis development has continued in targeted areas in this quarter, however, as discussed below, and will resume aggressively at the start of the current quarter to meet planned requirements for the second release.

Release preparations were extensive and encompassed careful testing by developers, and an increasing number of outside users. The latter component is expected to increase from this point on, but in keeping with the general project philosophy, internal testing will remain a clear responsibility and priority of the project. All identified defects were filed in the standard ClearDDTS defect tracking system, with associated defect severities, and were addressed with special focus on the severity one and two categories. This system has worked well in synthesis, as in other areas, to track and provide user feedback on defect status. AIPS++ has unique capabilities to provide automated testing due to the powerful command line (Glish) and the high-level connection between Glish and the underlying C++ code. The automated test scripts have been reviewed and their completeness assessed as part of an effort to revitalize and modernize this capability. This will enhance daily feedback on the synthesis code correctness, as well as improve the speed of release and release patch testing.

The public outreach campaign to reach new scientific users was started in this quarter at the AOC in Socorro, and constituted six, hour-long tutorials conducted once per week. Two focused exclusively on synthesis capabilities in imaging and calibration. As an adjunct, the first edition of the cookbook was completed, which will be the highest-level documentation for scientific users. It augments the already extensive user reference manual, but does not replicate the lower-level material. It is indexed on scientific data reduction activities, and contains synthesis chapters on general imaging (T. Cornwell), wide-field imaging (K. Golap and T. Cornwell), and synthesis calibration (A. Kemball), in addition to important chapters in other areas. We expect to expand the cookbook significantly in response to user feedback, and to add more material for the second release. The public outreach campaign will be taken to Europe in November (Cornwell and Kemball), as well as other consortium sites in the US and elsewhere in the short term. Our intent is to publicize general capabilities, including synthesis, to potential scientific users as widely as possible and for all consortium instruments, and we expect this to continue to some time as a high priority to introduce new users to the system.

The general synthesis development which has been undertaken during this quarter has been in areas consistent with previous planning priorities, namely: i) the development of "thin-path" capabilities for as complete end-to-end connected-element reduction as possible; and, ii) advanced synthesis development in targeted areas to finalize key interfaces.

Thin-path development was focused in the areas of data fillers and uv-data visualization using the Display Library framework. The VLA data filler was advanced significantly during this period (R. Marson); more advanced capabilities will be added during the current quarter. The ATCA filler has also been revised to add new features. The uv visualization work in the Display Library moved more slowly, but initial code and a design review has been completed. This will move more aggressively in the current quarter. Basic editing capabilities were added to the visplot tool by T. Cornwell, as well as general optimization. Work on the outstanding code required for full adoption of the MeasurementSet v2.0 format moved ahead, beyond the ms tool and into the synthesis MS access code (M. Wieringa). We plan final implementation in the full package early in this development cycle, well in advance of the second release.

A mosaicing techniques meeting arranged by M. Holdaway, T. Cornwell (AIPS++) and M. Rupen (for ALMA) held in July was a significant success, with important contributions by many participants. Work in mosaicing has continued in this quarter (M. Holdaway and T. Cornwell) in the areas of multi-resolution CLEAN and MEM deconvolution for mosaicing. In addition, work has been undertaken in optimized FFT gridding and primary beam weighting for mosaicing. Planning for simulation work in support of ALMA has advanced and will be a new focus for M. Holdaway once some final mosaicing capabilities have been completed. K. Golap has tested the wide-field imaging capabilities extensively in this quarter, using both real low-frequency VLA data and simulated data. This has primarily involved use of the dragon tool for wide-field imaging, but support utilities to measure the positions of simulated image components have also been developed in support of this testing effort. K. Golap and A. Kemball are now parallelizing the wide-field imaging capabilities as part of the parallelization effort with NCSA. This holds significant promise as these are very computationally challenging problems for all realistic cases. A basic on-the-fly (OTF) gridding capability has been added to imager (T. Cornwell and B. Garwood) in the same framework as synthesis gridding, in keeping with the overall philosophy of combined synthesis and single-dish processing in AIPS++. This OTF effort is part of the strategy of finalizing key high-level code interfaces. A related effort concerns planning for image-plane solvers (K. Golap and T. Cornwell), a final interface which will be established by prototyping.

Synthesis calibration has been extended in this quarter through the addition of data selection and display capabilities, the improvement of basic infrastructure, and through the provision of flux bootstrapping utilities (A. Kemball). The cross-calibration capabilities are in place, as planned for the first release, but there is still work to be done expanding capabilities in this area, and it will be undertaken in the current quarter. NFRA has continued work on utilities in support of WSRT/TMS commissioning (J. Noordam and G. van Diepen). Some assistance has been provided with debugging the transfer of data through AIPS++ by A. Kemball.

Parallelization and high-performance computing, a collaborative effort between NCSA and NRAO (A. Kemball and D. Roberts), has a close connection with synthesis development, in the areas of wide-field imaging, mosaicing and the planned image-plane solvers. This effort is summarized elsewhere in this report.

A new planning cycle has been completed in this quarter to define the synthesis development targets for the second release. The primary focus will lie in expanding the connected-element thin-path in areas in which it is weak at present (uv-display and editing), and in providing a basic VLBI capability (data filler and fringe-fitting). Automated reduction and imaging pipelines will also be a key focus in this development cycle.

In Glish, no substantial changes were made.

In Parallelization, work continued consistent with the previous planning in this area. The broad goals of this effort are to demonstrate, and make scientifically available, a capability to process the most challenging data reduction problems in radio astronomy, and to introduce the parallelization infrastructure in AIPS++ to make these high-level applications possible.

A continuing infrastructure development area which received attention in this quarter has been a research effort into the use of parallel I/O in AIPS++. In the first two quarters of 1999, a graduate student with the computer science department at the UIUC added I/O instrumentation calls to the AIPS++ library. All reads, writes, opens, closes, and seeks have been instrumented. During execution, the counts and times of the I/O calls are written to a trace file in Self-Defining Data Format. The Pablo analysis tools are used to investigate the resulting trace file. D. Roberts proposed and was granted access to an NCSA Origin 2000 before it was introduced into production. The parallel group had access to the machine for 10 days, during which 8 experiments of various sizes were performed. In preparation for this experiment, D. Roberts and W. Young worked on validating correct execution of pimager (the MPI-based, parallel version of imager) on large images. Several problems were identified including a 1 GB limit on memory allocation with 32-bit programs under IRIX. After validating the correct image generation, D. Roberts carried out the experiments and analyzed the resulting I/O trace files. The results were that several files were opened and not closed. More importantly, for large images, I/O to the TempLattice file that was created in the imaging process dominated the total I/O by a large factor. For runs that created 2 GB images, the total I/O to the TempLattice was 500 GB, with frequent small size I/O requests. The parallel group is looking into optimizations to address this performance bottleneck.

After the I/O experiment W. Young visited NCSA and worked on porting AIPS++ to the newest generation of the SGI MIPSpro compilers (7.3). This effort has been completed. The 7.3 compiler has the newly defined OpenMP binding for C++. D. Roberts started a project to parallelize GridFT using OpenMP and the new compiler. This should be complete in the next quarter.

After porting AIPS++ to MIPSpro 7.3, D. Roberts and W. Young investigated what would be necessary to port AIPS++ to a 64-bit compiler in order to address the 1 GB memory limit discussed above. W. Young is summarizing the specific required changes to the code and will submit the details to the project in October for consideration of their broader impact. W. Young has been using a code verification and memory tracking utility, Insure++, in this work.

D. Roberts did some site administration duties and installed AIPS++ on a new Linux system in order to be able to test Linux changes on a local machine. During a visit to NCSA, A. Kemball and D. Roberts worked on a document specifying NCSA computational requirements and revision of the plan for parallelization and optimization for the next 6 months.

An initiative in the parallelization of wide-field imaging within AIPS++ has been started by A. Kemball and K. Golap. The existing parallelization infrastructure has been refined to address some requirements of this initiative as part of this work. The wide-field parallel capability is expected to be available in this quarter. Performance profiling of the existing serial wide-field imaging capability was completed this quarter.

The port of a snapshot of AIPS++ to Windows NT has proceeded as part of the MS research project of P. Cortes under supervision of A. Kemball at NRAO, and a reasonable fraction of the aips library has been compiled under Microsoft Visual C++ in this quarter. The C++ syntax differences are being tracked between the NT and UNIX compilers, along with differences in template instantiation and related issues. The current plan calls for completing the NT port for the aips and trial computational libraries, and then running parallel applications on the NT super-cluster at NCSA. Porting of Glish and the Display Library is deferred until later. It should be emphasized that this is not a final port of AIPS++, but instead the trial port of a snapshot. Part of this work will be a report detailing the implications of a full port. The related work by P. Cortes concerns the performance of AIPS++ parallel applications on cluster, or loosely coupled, parallel architectures. We have obtained an allocation of time on the Linux super-cluster of the Albuquerque High Performance Computing Center as part of this work, and have started to install AIPS++ on this system. We expect I/O and memory access to be significantly different in this environment as compared to more tightly-coupled architectures (eg. SGI), and the effect this will have on radio astronomy reduction algorithms in part of this research effort. Clusters are likely to play a strong role in radio astronomy reduction in future years.

Work continues in ensuring build stability between the SGI AIPS++ versions at the AOC and NCSA. We have undertaken system administration work to synchronize operating systems and compiler versions between the two sites in this quarter, as in previous quarters. The SGI build stability has improved significantly as a result of these and related measures.

In Measures, no substantial changes were made. One amusing exception was that ALMA was added to the database of observatory locations.

In AIPS++ Infrastructure, most AIPS++ time has been spent on testing the system and resolving defects. Little new development has been done. Defects were mainly resolved in the tools catalog, os, and logger to make them adhere to the GUI guidelines, to have better interaction between GUI and command line, and to add new options. A few defects in cxx2html have been fixed to make it process header files where summary or class declaration is split over multiple lines.

In Image Analysis, we fixed defects, considerably augmented the test scripts, added some new capabilities (concatenation of images, improved viewing), and improved the documentation.

In Documentation, the first version of the Cookbook was assembled by Athol Kemball, with contributions from Cornwell, Kumar Golap, David Barnes, and Neil Killeen.

In Management, at the end of the quarter, we revised and updated the development plan. We intend to settle down on a 6 month plan-develop-test-release (PDTR) cycle. The time scale is as follows, counting from the date of the previous major release.

+0 weeks
Revise planning documents
+2 weeks
Finalize planning, assign targets
+2 weeks
Start development, testing
+2 weeks
Initiate completion of testing of existing capabilities
+4 months
Freeze new checkins, start dedicated testing
+5 months
Hard freeze, uprev to next versions of release and develop masters.
+5.5 months
Cut final CDs
+6 months
Ship CDs

In Outreach, we conducted a series of tutorials talks by Tim Cornwell and Athol Kemball at the AOC. These were based on various pieces of documentation, including ``Getting Started in AIPS++'' by Anantharamaiah, and the AIPS++ Cookbook assembled by Athol Kemball. These talks were well received with about 20-25 people attending each of the 6 talks. Our first impressions of the response are:

AIPS++ had a considerable presence at ADASS in Hawaii:

In System, a lot of effort has been put into preparing the release, especially by Darrell Schiebel and Tim Cornwell. The system tree has been split into a release and development tree. Patches are checked into the release system, which can only be done by Tim. New developments have been started in the development tree. Most notable are:

It appeared that the Rational products like purify do not support egcs 1.1.2 and it looks like it takes quite some time before it gets supported. The beta versions of purify, etc. support egcs 1.0.3 to a greater degree. Therefore it has been decided that AIPS++ code will add egcs 1.0.3 as a secondary compiler in the interim. Where needed, macros or other constructs have to be used to make it compile. In Socorro it is investigated whether Insure++ is a possible replacement for purify and purecoverage. A good replacement for quantify does not seem to be readily available.

In the Quality Assurance Group, we continued on a review of the compliance with various rules. This resulted in a substantial number of minor changes here and there in the system. We expect this work to continue at a lower rate for the next development cycle.

In Visualization, we concentrated on fixing defects and improving documentation. NRAO was just awarded an NSF grant (from Computing, Info Sci, Engineering) for visualization development in AIPS++. This amounts to two positions for three years. This will augment staffing at NCSA and ATNF to double the total effort. The goal of the proposal was to allow visualization and editing in two different domains while transforming between the two using AIPS++ capabilities. We are currently working to fill these positions.

First release

In the preparation for the release, we issued two patches (binary fixes) to the CDROM pre-release. These patches were in response to the defect reports submitted by our testers.

In the previous quarterly report, the number of total defects was still rising linearly with time. In this quarter, the number of defects saturated, indicating either that the package was becoming more robust and stable, or that our testers were exhausted! The common experience amongst our testers (mostly project staff) affirms the former, though the latter is probably also true. The mean time between failure of AIPS++ is typically a few hours.

Week Date # New # Resolved Diff # Unresolved
1 4-25-1999 9 0 9 9
2 5-2-1999 5 2 3 12
3 5-9-1999 3 0 3 15
4 5-16-1999 15 22 -7 8
5 5-23-1999 35 11 24 32
6 5-30-1999 39 18 21 53
7 6-6-1999 41 11 30 83
8 6-13-1999 18 7 11 94
9 6-20-1999 21 11 10 104
10 6-27-1999 37 24 13 117
11 7-4-1999 60 35 25 142
12 7-11-1999 45 43 2 144
13 7-18-1999 18 25 -7 137
14 7-25-1999 53 35 18 155
15 8-1-1999 17 24 -7 148
16 8-8-1999 3 4 -1 147
17 8-15-1999 90 19 71 218
18 8-22-1999 20 11 9 227
19 8-29-1999 63 32 31 258
20 9-5-1999 36 30 6 265
21 9-12-1999 21 32 -11 254
22 9-19-1999 19 17 2 256
23 9-26-1999 28 15 13 269
24 10-3-1999 12 4 8 277
25 10-10-1999 6 10 -4 273
26 10-17-1999 6 12 -6 267
27 10-24-1999 0 1 -1 266

The highest severity defects run about 10% of the total. We expect to act on those as quickly as possible, and the lower severity defects as a background task consuming about 25% of our available time. As the number of defects saturated, we became confident that the package was ready for the first releease. The first release was made on Oct 4 at the ADASS meeting in Hawaii. Approximately 225 CDROMS were distributed at ADASS, 100 being for the most recent version of Linux, and the rest split equally between the previous edition of of Linux, and Solaris. Our mailing list of names gathered at meetings prior to ADASS has over 150 names, and we continue to receive new requests about 1 per day. Interestingly, many requests are coming from non-academic addresses such as aol.com, indicating that we may soon have to consider the pros and cons of restricting distribution. However, for the moment, the attention is welcome.

The release delivered the following capabilities:

The following were missing:

Project overview

After passing a very important milestone, the first release, it is a good time to consider the overall health of the project, and to make any required changes:

The core package
We have delivered a working system that contains much, but not all of what we hoped to. The basic tools and command language (Glish) are acceptably robust and stable for the first release, though improvements must still be made. The graphical user interface is now useable, though improvements will be required as we get user feedback. The Measures and Quanta system functionality is superb. We have a first version of what will grow to be a very capable display package. Image handling is good. Synthesis imaging is excellent in many categories, with major improvements yet to be made only in mosaicing and single dish imaging. The agreed calibration ``thin-path'' is present. Overall, the core package is about a year away from being complete enough that other competitive packages can be decommissioned. Required action: The consortium should maintain a focus on the core package for at least one more year. We note that this is as stated in the first quarterly report (1995 Q2) under the current management:

The long-term goal of the Project has been defined to be the achievement of functional equivalence to AIPS by 2000. At that point, AIPS will be a very small subset of AIPS++ and most applications areas will look quite different from the corresponding areas in AIPS.

External programmability
We have not been able to devote any significant resources to a developer's release. The earliest that this can occur based on current planning is early 2001. This relative priority is in accord with various advice that we have received. We can, of course, give a limited number of groups the required access to develop in the system.

The project staff
We have assembled a very high quality team that works well together even spread across the globe. This investment is substantial. While productivity in Glish is usually good almost immediately, training a programmer to be productive in C++ typically takes 6-12 months. Every effort should be made to keep this team together. One key deficiency is the lack of programmers to work within the synthesis calibration package. To expand the synthesis calibration package beyond the current thin path, we still need all the coordinated effort that we can muster. Given this, it is worrying to see that members of the consortium are planning to use the synthesis calibration package in imaging pipelines. We believe that while such use is close to being possible, first some key areas of the calibration must be finished, and second, coordination is needed to avoid unnecessary duplication of core capabilities. Required action: Explicit coordination by the project center is needed. We request that the various people dedicated to these imaging pipelines first be utilized to finish the synthesis calibration package under the direction of the project center.

Project process
For the most part, the Project has established well-documented processes for many activities. For example, substantial changes are proposed by a standardized change proposal process. Defects are handled using a standard package and protocols. The package health is monitored by documented tests run on a regular basis. Coding and documentation standards are enforced on review of the code. The latter slipped in priority during the preparations for the first release, but is now getting more attention. An informal comparison of our processes with those employed by ESO show them to be comparable, with different areas of strength and weakness but with overall parity.
Internal communications
The project is effectively run by consensus amongst the various managers, with ties and day-to-day decisions being made by the project management in Socorro. The unclear demarcation between these categories leads to in some cases excessive communication over trivial matters, and in others, missed consultation over areas that merit more discussion. Required action: To address these points, we plan to institute more open notification and record keeping of decisions. In addition, a planned get-together in April will help the entire group communicate.
Planning, Scheduling, etc.
The actual release date was about 6 months later than we anticipated a year ago: about a 50% underestimate. There are three main contributing factors of roughly equal importance:
Outreach
Now that we have a working package, outreach will get easier and easier. For the moment, it is clear that we are beginning to erase some negative perceptions of AIPS++. The case for AIPS++ has yet to be made convincingly in a number of contexts, but we are confident that with the core package in hand, such a case can be made. Our existing plan for outreach will be followed.
Advice
Due to the vagaries of the release process, the STAG has not met since February 1998. To ensure that the next advisory committee operates more from direct knowledge of the package, we plan to reconstitute this group drawing from early adopters of the package. We also plan to split off the technical advice into a separate committee. We plan that both groups will meet around the time of the next release.

Developments planned for 1999 Q4

The prime goals for the next quarter are to support the release, and to work towards the new developments needed for the April 2000 release (1.3). Details of the latter are covered in the the development plan. We anticipate that the work allocated to support and to development will be roughly equal.

Appendix: ATNF contribution Neil Killeen

General

This quarter the ATNF had 3 people working in AIPS++. These are Neil Killeen (75%, also local manager), Mark Wieringa (30%) and David Barnes (75%). Wim Brouw's time has ramped back up and he spent 60% of his time in aips++ this quarter. He will spend about 20% of his time next quarter to conclude his work in ATOMS. In addition, Frederic Badia at Narrabri started to use the system (see below) for some prototyping of ATCA pipeline software.

This quarter was heavily focussed on testing, and preparing code and documentation for the release in early October.

System

This quarter we jettisoned the Solaris/Gnu system in accordance with project policy and moved to purely egcs systems. The Linux file locking problems appear to have gone away after we finally got our Linux systems upgraded to Redhat 6.0

At Narrabri, Digital Unix is the main platform. Since there is no binary release for that, the release will need to be built. There may also be a Solaris or Linux platform provided at Narrabri too. Parkes operates Solaris only.

In addition, we hope that use of the egcs 1.0.3a system at Socorro (for use with Purify, PureCoverage and Quantify) will be satifactory for our needs so that we don't have to generate that extra system here (depends upon link speed).

Talks

Neil gave a 1-hour talk at Mt. Stromlo on Glish. He also gave two short aips++ talks, one to general ATNF and CTIP staff, and another to the visiting Gemini technical committee.

Visits

David spent a few days at Dwingeloo discussing the Display Library with Tom Oosterloo (one of its designers). Neil spent a few days in Dwingeloo talking mainly with Ger van Diepen and Jan Noordam.

Individuals

David Barnes

David's time (75%) was spent on

Wim Brouw

Wim's time (60%) was spent on

Neil Killeen

Neil's time (75%) went on:

Mark Wieringa

Mark's time (40%) went on

Frederic Badia

Frederic works in the Narrabri computing group.

Appendix: BIMA/NCSA contribution Doug Roberts

Harold Ravlin worked to fix some bugs in the PostScript driver for the Display Library. Additionally, Ravlin has worked a bit on Aipsview; this includes porting Aipsview to compile and work correctly under the 64-bit IRIX compiler. Ravlin added True Color support to Aipsview, which now displays on either pseudocolor or True Color visuals. When displaying in True Color manipulations of the color map are applied to all windows after the mouse is released and provides good performance on the target consortium machines. Internally, Ravlin changes Aipsview to use a modified version of the PostScript Driver from the Display Library. The integration of the PostScript Driver into Aipsview resulted in identifying to bugs which were fixed in the DL. Ravlin finished up some system administration duties including recovering the NCSA IRIX server after a disk crash. Doug Roberts was involved in testing Aipsview, checking the affect of the Aipsview changes. Roberts also updated the Aipsview User's Guide to reflect the changes to the program that we introduced this quarter.

Percent of time on AIPS++: Doug Roberts 70%, Harold Ravlin 70%

Appendix: NFRA contribution Ger van Diepen

Local project members: Ger van Diepen (GVD), Jan Noordam (JEN, local manager).

General

After a prolonged period of enforced introspection due to the pressures of the WSRT upgrade, NFRA is now ramping up again its contribution to the global AIPS++ package. Oleg Smirnov will arrive any day now and will work in the area of uv-data calibration, with the emphasis on the ionosphere. In addition, NFRA is taking part in a collaboration with ATCA to create a modular 'calibration pipeline'. Its modules should be directly usable in the global package.

In this quarter, the only substantial contribution to the global package was made by Ger van Diepen in his role as Technical Leader, and as the man responsible for a number of essential infrastructure modules (tables, lattices, ms2uvfits, etc).

In the meantime, the upgraded WSRT continues to rely heavily on AIPS++. The online system (TMS) uses AIPS++ Tables, Measures and Glish, and MS2uvfits for data export. The local WSRT 'mini-package', written in Glish by Jan Noordam, is operational and widely used for inspection of uv-data and determination of instrumental parameters prior to observations. It is available in the global package too.

This quarter, the NFRA contribution in FTE's was as follows:

  Global AIPS++ Local AIPS++ Other
Ger van Diepen 25% 45% 30%
Jan Noordam: 10% 60% 30%

(NB: The 'other' category includes vacations)

The program ms2scn has been upgraded to deal with the new values of the WSRT in the ARRAY table.

The program j2convert has been extended considerably to make it possible to fix errors in older WSRT measurement sets. Also the way UVW and directions are calculated has been improved.

The program ms2uvfits has been improved a lot to be able to export AIPS++ data to other packages. It took quite some effort to get a correct file in the poorly documented UVFITS format.

A framework (MS2IWOS) has been developed to make AIPS++ measurement sets accessible for the existing IWOS package in Westerbork. Using this framework a program like MAKCAL could rather easily be converted.

Appendix: NRAO contribution Tim Cornwell

The core NRAO AIPS++ group is now Barnes (100%), Braatz (100%), Cornwell (100%), Garwood (100%), Golap (100%), Holdaway(100%), Kemball(100%), Marson (100%), McMullin (100%), Schiebel (100%), Weatherall (50%), and Young (100%, Golap and Young are funded by NCSA-NRAO collaboration). We have two open positions funded by a three year NSF grant for visualization.

Peter Barnes

This quarter was marked by the conclusion of my trip to Sydney, extensive testing and system work, and completion of targets (about 2 weeks spent on science):

Jim Braatz

Jim Braatz was hired on to the AIPS++ project in April 1999. His primary responsibilities are to manage the Green Bank installation of AIPS++, to support GBT operations, and to work on single-dish applications. Specifically, Jim worked on the following tasks during this quarter:

Tim Cornwell

Tim worked on:

Bob Garwood

Bob Garwood's primary responsibility is to oversee and contribute towards the single dish work in AIPS++. This work remains focussed on the DISH environment and the support of the GBT. His contribution in support of the GBT is primarily through the GBT fillers which convert the GBT FITS data files to an AIPS++ MeasurementSet. In addition to these duties, he is responsible for the maintenance and enhancement of the FITS classes and, as author and maintainer of the GUI guidelines, he spends some amount of time reminding other AIPS++ workers on the importance of following those guidelines.

Over the past 3 months (July through September) he has done the following:

Mark Holdaway

Mark Holdaway organized the AIPS++ sponcored Mosaicing techniques Meeting in Socorro, NM on June 14-17. The meeting was attended by about 35 people. The presentations were informal and stimulating, and the discussions, led by Tim Cornwell and Ron Ekers, were insightful.

Mark has revived Mark Wieringa's simulation software. He ran several suites of simulations comparing Multi-Scale Clean with MEM and the Clark Clean. These simulations document MEM's positivity bias due to noisy data, and demonstrate that the Multi-Scale Clean suffers no bias. Yet, the Multi-Scale Clean is much better than MEM and Clark Clean at accurately reproducing extended structure. This work was included in posters on Multi-Scale Clean at the URSI and ADASS meetings.

Mark has also worked on getting MEM and Multi-Scale Clean into imager. Many bugs were fixed.

Athol Kemball

I have continued to coordinate synthesis development during this quarter, in addition to code contributions in the area of calibration, and related defect identification and correction. This is described in the synthesis report in more detail. Other direct involvement in coding or infrastructure includes participation in design of uv-data visualization code within the Display Library, the production of the first edition of the cookbook, automated imaging design, wide-field parallelization, intercomparisons of the AIPS and AIPS++ VLA data fillers and preparations for MS v2.0. System involvement included monitoring of overall memory use and development of a megaserver solution to reduce this use, work in support of the adoption of egcs 1.1.2 as the project compiler, and evaluation of different memory checking tools and related compilers. I have continued also to coordinate the parallelization and high-performance computing initiative, which is a collaborative venture with NCSA. Work in this area has included research in parallel I/O, contributions to build stability on the SGI, wide-field parallelization as described above and the current NT port and cluster research with P. Cortes. The parallelization effort is described separately in this report. Release preparations have required a considerable fraction of time this quarter, both in testing and defect correction, and in verification of installation procedures under various architectures. I have also participated in the public outreach program at the AOC to provide a tutorial for new scientific users of AIPS++, in addition to general administration.

Joe McMullin

Joseph McMullin's main responsibilities are work on single dish applications within AIPS++; he was away on US Army Reserve training from 20 July - 30 Sept 1999.

Kate Weatherall

Wes Young

Wes and Lisa had a baby.

Appendix: Summary of AIPS++ Personnel Changes

In this section, we give the names of people in the various AIPS++ groups and the nominal fraction of time allocated to AIPS++.

The ATNF group is: Neil Killeen (75%, also local manager), Wim Brouw (60%), Mark Wieringa (30%) and David Barnes (75%).

The BIMA/NCSA group is: Harold Ravlin (70%), Doug Roberts (70%).

The NFRA group is:

  Global AIPS++ Local AIPS++ Other
Ger van Diepen 25% 45% 30%
Jan Noordam: 10% 60% 30%

The core NRAO AIPS++ group is now Barnes (100%), Braatz (100%), Cornwell (100%), Garwood (100%), Golap (100%), Holdaway(100%), Kemball(100%), Marson (100%), McMullin (100%), Schiebel (100%), Weatherall (50%), and Young (100%), Golap and Young are funded by NCSA-NRAO collaboration). We have two open positions funded by a three year NSF grant for visualization.

Thus, in aggregate, we have 20 people contributing about 16.65 FTEs to the AIPS++ Project. Of these, 12 are employed by NRAO (one funded by NCSA), and contribute 11.5 FTEs. The numbers for the other partners are: ATNF 4 and 2.4, BIMA/NCSA 2 and 1.4, NFRA 2 and 0.35. We have two open positions in NRAO.


next up previous
Please send questions or comments about AIPS++ to aips2-request@nrao.edu.
Copyright © 1995-2000 Associated Universities Inc., Washington, D.C.

Return to AIPS++ Home Page
2006-03-28