AG-SAG - Full Meeting #5

University of Glasgow, 7 Feb 2005, 10.00-16.00

Supporting Paperwork:

Agenda:

The AGSAG has been asked to produce a document for the February meeting of the AstroGrid OversightCommittee. I think the topic will overlap closely with the discussion at this meeting:

• Discussion on Science Use with current for AG Release 1
• Science Analysis - priority needs into AG2

Some more specific points to consider and discuss:

• How has the AGSAG contributed to AstroGrid (1)?
• Overview of AGSAG interest in and contribution to the AstroGridTen science projects.

-- AndrewJaffe - 31 Jan 2005

Location: Room 604, Kelvin Building, Physics and Astronomy, University of Glasgow.

A local map can be found at http://www.physics.gla.ac.uk/location.html

Information on how to get to the University is at http://www.gla.ac.uk/General/maps/index.html

A list of accommodation can be found at http://www.astro.gla.ac.uk/users/eduard/rhessi/workshop2/hotels.html

Please contact Lyndsay Fletcher (lyndsay@astroNOSPAM.gla.ac.uk) if you require help with accomodation.

Meeting minutes prepared by SilviaDalla

Status: Draft 8 Feb 2005

Present:

• Carlton Baugh (CB)
• Bill Chaplin (BC)
• SilviaDalla (SD)
• Lyndsay Fletcher (LF)
• EduardoGonzalez (EG)
• AndrewJaffe (AJ)
• Danny Lennon (DL)
• RichardMcMahon (RM)
• AnitaRichards (AR)
• Nial Tanvir (NT)
• NicholasWalton (NW)
• JeremyYates (JY)
• TonyLinde (TL)

Apologies

• Tom Marsh
• SteveMaddox

1. Actions from Previous Meeting

ACTION FM4-3A TL: solve problem with mySpace file names.

Closed.

ACTION FM4-3B TL: solve problems with having to clic Submit many times when building a workflow

Closed.

ACTION FM4-3C TL: We need to have a window telling the user that she has been logged out. Extend logout time.

Closed / postponed : Logout time has been extended. Window telling user has been postponed - quite difficult to implement.

ACTION FM4-4A TL: solve problem with cursor having to go round to remove a step.

Ongoing. Not a priority at present since there will be development of client workflow builder.

ACTION FM4-4B LF and SD: Contact people at Lockeed-Martin about IDL/Perl issue.

Closed. Co-SEC uses Remote Procedure Calls to allow users to call selected rotines part of solarsoft.

ACTION FM4-7A NW: Organise a Telecon of people interested in developing science cases.

Closed. Telecon did not happen but work on developing science cases took place. Work on science case with JY will take place end of Feb.

ACTION FM4-7B NW: email Wiki page on Itn07 facelift to SAG members.

Closed. Discussion taking place at current meeting.

ACTION FM4-7C TL and NW: check out ways to publish material (iTunes example!)

Ongoing. It requires my space knowing about access permissions to a dataset that the person wants to publish.

ACTION FM4-7D NW: Contact John Alley at Durham

Closed. He was contacted and came to the consortium meeting in Edinburgh.

2. Astrogrid 2 and VOTECH Scope and Targets

presentation by Tony Linde - Scope of AG2 and VOTECH

Rather than with iterations, AG will now work in 6-month cycles. There will be a release at the end of each cycle.

End of February: Release Candidate 1 (RC1)

End of March: V1.0 released (pre NAM)

End of June: V1.1 full deployment

Discussion about VOTech - AstroGrid SAG is one of most active advisory groups within the euro-VO and will contribute towards science direction of the project.

AJ - Will each University have to install AG software, for astronomers there to use AG?

TL - No, astronomers will just need to have access to an AG portal - typically the portal tailored to the research field they work in. In the longer term, Myspace access might have 'local' options, so that depending on where you are you will have different Myspace disk space, access restrictions, etc.

3. Discussion on Science Use with AG Release 1

AJ - Who gets to use V1.0 ?

NW - In principle anyone will be able to use V1.0 - however there is the issue of how many people the system will be able to support. We would like 50-100 users by end of 2005. We plan to package a few workflows into AG services so that people will start using the system via these simple services. Takes away the 'scary' element of working a workflow.

BC - How many portals will be in V1.1?

NW - Most likely 3 portals. They will be hitting the real datasets.

JY - Is there enough hardware for these portals to work ok? AJ - Do we know what's required to support more users?

NW - At present we are probably a factor 2 lower than required to support many users.

DL - How are more machines added to the system? TL - This is via the registry at present - it defines where an application runs. However it would be desirable for CEA to decide which machine to run an application on, depending on demand and load.

AJ - It would be better if it were possible to run an application on the user's machine, rather than on a grid machine. A lot of the jobs that we will be running are not particularly CPU intensive, and do not require a grid per se. Is there a way that a user can register one's machine?

NW - It would make sense also to be able to dump jobs on departmental machine.

DL - This is an important issue because it will put poeple off if running a job on AG is a lot slower than on their own machine.

JY - Will network connections be good enough between the 3 portals of V1.1?

NW - Often data are at places with bad network connections, such as telescopes. Maybe data should be moved to locations with better network connections.

4. Run through of Astrogrid Demos from December 2004 Consortium Meeting

DEMO1 - EXTRAGALACTIC (Eduardo Gonzales)

EGZ - There is a problem with the way in which data are presented to the user by the data centers. For example a cone search query from the INT Wide Field Survey returns all sorts of data, of different exposures resolution etc. However often the only info returned by the data center are filename and position. When querying INT WFS several files are returned for 1 position and size, then it is within worklfow that 1 image per position/size per band is chosen.

NW - AG worked on extending the SIAP service at Cambridge so that the image headers are read and this information is available.

RM - The data center should not have to change their datasets, this is a software limitation.

AMSR - The data centers have to provide the recipes and pipelines so that the data can be used properly. They know the data best.

NW: Trying to solve this should be a priority.

EG - Shows Topcat myspace browser plug-in. EG - This workflow takes about 4 hours to run - it's 28000 images. The step that takes the longest time is s-extractor. Also there's images being copied from Cambridge to Myspace in Leicester.

AJ - it would be useful to see more than just 'set', 'step' - something that tells you what each thing does. Have step name to display next to 'Step'.

DEMO 2 - GALACTIC (Anita Richards)

AR - This workflow does not require to download the images to Myspace, unlike Eduardo's workflow.

NW - The STIL library (used by Topcat) can be used directly from the workflow.

LF - Problem with using java webstart to install Myspace Helper Applications. How can a user be reassured that she installing trusted software?

ACTION FM5-4 TL : Look into security issues for installation with java webstart.

AR - Shows VOSpec tool. Candidate AGB stars to be selected by looking at the whole em spectrum. VOSpec tool reads data from multiple sources and produces spectra. You can select a catalogue of eg radio data.

DEMO 3 - SOLAR (Silvia Dalla)

5. Report of SAG activities to Oversight Commettee

AJ - The Oversight commettee has asked the SAG to produce a document on science which has been done with AG Release 1, and on priorities for the next stage. Also what the SAG input to AG has been so far.

AJ - Has the SAG been useful to AG so far and what are its main contributions?

NW - Useful to have a discussion with scientists involved in science in a variety of fields. Also SAG gives the end user point of view, and pushes the project towards improving the system for the user. Useful input from SAG members to the project.

NW - The system has not got as far as we would have wished in terms of science output. Tehnical problems were encountered in the grid technology, which were not initially expected, and delayed the release of a stable system. This is now there and can be used to do science.

DL - What exactly will be released? It is not clear who are the 'target' 50-100 users.

NW - Initially have users from SAG and beta testers working with science team to do science. As an initial core group from which usage will then spread. Something like ~10 projects to get going in the next few months. What is being released in March is what we have now. Science services will be part of these. They will be a way for people to get interested in the project.

BC - There is a science case related to helioseismology that I would be interested in developing.

LF - Solar project to be developed together with a summer student. Involving a student to try things out helps SAG members to overcome lack of time to dedicate to this project.

NW - Encourages SAG members to work with science team in the next few months. There is funding for science team to visit departments to work on AG projects and promote AG.

AR - How soon can we manage to avoid all the scripting work. Are there resources within the project to work on this?

TL - Probably not until June.

JY - We need a list of things that need to be wrapped into workflow components to avoid scripting.

6. Science with Astrogrid and VOTech

Presentation by Nic Walton on Science Services to be released at NAM:

- RedshiftMaker - ColourCutter - MovieMaker

AG Hands-On workshop Cambridge April 20-22. SAG members very welcome.

Presentation by Nic Walton - Science input to VOTech

There will be a Euro-VO Science Working Group.

Presentation by Richard McMahon: Science driven application & tools: Astrogrid and VOTech

RM - The AG2 proposal says that astronomers will be invited to suggest capabilities for the VO. Projects would be selected by the SAG and Astrogrid management for development by the project.

JY - Experience of Starlink in this respect: a board was established to judge proposals from the community for tools to be developed by Starlink.

RM - can we extend the Astrogrid 10 ?

TL - We need to engage with the big projects, because they need to get their data into the VO. As far as development of tools is concerned, AG should get involved with individuals and dedicate some developer time to supporting scientists. This cannot be done before June though. If SAG thinks that this should be a priority, they can ask the AG management to devote resources to supporting single users.

NW - Should AG have a help desk? This question needs to be addressed.

AR - In the AVO there was no time to go into the Science Reference Mission and tidy up the science cases. Some of the AVO science cases can be generalised and used within the Euro-VO.

NW - Time for running S-Extractor via AG is still taking too long. It's necessary to work out where the overheads are.

ACTION FM5-6 EG + NW: investigate why S-Extractor is taking this long to run.

TL: Applications take time initialising. If an application were in big demand then it would be possible to having it running all the time - avoid the initialising.

7. Next SAG Meeting

Next SAG Meeting - 25 May 2005 at UCL. To be confirmed beginning of March.

-- NicholasWalton - 06 Jan 2005