Andy Turner's MoSeS Technical Meeting 2007-01-26 Page

People

• Belinda
• Jie
• Mark
• Paul
• Andy (chair)

Agenda

1. Moses as a Virtual Organisation:
   • Review the slides from Steven Pickles presented at last week's e-infrastructure workshop.
2. Responsibilities and interdependencies:
   • There has been a flow of e-mail traffic regarding the data and model interfaces (see Moses Technical Meeting.msg, Moses Technical Meeting.txt)
     • Review and check that we all understand who is doing what.
3. Use Case:
   • The document from John England from Leeds Social Services provides us with a tangible set of user requirements for the portlet. I'd like to at least start a discussion of how we can meet these requirements.
4. Update of progress on actions from last technical meeting

Notes

• Mark explained that Belinda is being delayed by Profesor Rees.
• As chair I've decided we can wait for 30 minutes and work online...
• Investigating Paul Williamson's work:
  • Small-area synthetic population microdata. This site reports the outcomes from 'The creation of a national set of validated small-area microdata' (ESRC award no. R000237744)
    • Methodology
      • Method: A 'combinatorial optimisation' algorithm is used in the creation of the synthetic population microdata hosted on this website. The aim is to select a combination of household from a sample survey that, when aggregated, best-fit published small-area Census tabulations.
      • It seems Paul et al were going about this population intialisation in a very similar way to what we have been doing in MoSeS.
      • Mark recalls that Paul tried investigating using Genetic Algorithms but didn't get very far/have much success with it.
      • A COMPARISON OF SYNTHETIC RECONSTRUCTION AND COMBINATORIAL OPTIMISATION APPROACHES TO THE CREATION OF SMALL-AREA MICRODATA
• Hurray Belinda has escaped and is here just as we were about to start without her :)
• Let's go through the agenda:
  1. NGS, MoSeS and virtual Organisation (VO)
     • MoSeS is already a VO
     • For MoSeS we want to set up different VO's for each application.
       • Paul is already thinking abot portal changes for this
     • Paul and Jie are expecting a new member of the team that is working specifically on NGS. All NGS work with MoSeS is to be put on hold until we meet this person.
     • Grid Oriented Lifecycle Development (GOLD) collaboration with Newcastle (Rob Smith) is a possibility
       • CS-TR: 940 Service Oriented Middleware for the Formation and Operation of Virtual Organisations: The GOLD Project, Periorellis P., Cook N., Conlin, A., Hiden, H.G., Watson P., Smith R., Wright A. School of Computing Science, Newcastle University, Jan 2006
     • Will ESRC give us brownie points for using NGS?
     • We do not know what resources are needed to use the NGS
     • What about sakai and the NCeSS e-Infrastrucutre project?
  2. Interface between modelling activities
     • Passing data between us.
     • Agreeing one way is best.
     • Probably file and text delimitted...
     • Jie raised a concern about whether the database Belinda is using will be able to cope with the large data volumes once we scale up to modelling the UK rather than Leeds.
       • The short answer is that we're not sure but that maybe the problem can be split spatially into chunks which can be processed concurrently, so this probably not be an issue.
     • Belinda passed round a printed out query of how she finds a marriage candidate. This should be emailed to me to link here.
     • Parallelisation in the dynamic model
       • This is hard. Especially migration.
       • Mark suggested that Georgios from Birmingham might have experience of a similar problem.
     • Using SRB as extended memory
       • I like this idea. It can help with parallelising the dynamic simulation too.
     • Belinda and I discussed about keeping track of individuals and their history as they potentially move in and out of households and Communal Establishments
       • Such histories are interesting for examining social and family networks for the dynamic model and anyway.
       • If the simulation were run for long periods spanning many generations one might expect that the amount of history information becomes stable. (People with the biggest histories die and people are born with none per se)
     • We need to think about unique ID values for each individual and household
       • How are we going to code it?
       • What is the order of magnitude we are dealing with?
  3. Leeds social service requirements
     • John England (Head Honcho) Dennis Holmes (Guru)
     • How to generate a report for them?
       • Develop a mock up
         • Kick start this with a meeting on Feb 8th using flip charts and pens...
       • Aim to meet with them in about a month and present our ideas.
       • Based on feedback generate something and get them to access it from the portal.
     • Paul to configure the Portal to cater for different users of the system.
  4. Update on progress of action points
     • Belinda
       • The dynamic model now pushes people from households into Communal Establishments
       • Ready to recieve data from Andy, but wants to agree a new format so that each record has the same variables with no redundancy.
     • Paul
       • Paul gave a presentation which detailed this on slides :)
       • The plan was for him to email them to me, but they were lost and only an older version could be recovered.
     • Andy
       • Investigation of Paul Williamson's work underway (see above)
       • UK result has been delayed due to logic coding error in optimisation routine. UK result will be generated using easier convergence criteria and a run set going next week once new results for Leeds have been generated.
       • Still need to agree how to transfer the data to Belinda
         • Andy to harmonise the ISAR and HSAR data.
         • Mark will look at this with Andy next week. Hopefully it is just a matter of aggregation for all but Age where we plan to build on Jin's work
         • Paul has made progress with our joint publication
         • With Mark work on the publication is yet to start
• Action point review
  • Andy went through action points he had written against all based on our discussions.
  • Others asked to feedback.
• Meeting close
  • Thanked all and asked for feedback on my charing
    • Action point review should have been at the start of the meeting
    • No major complaints :-)

Actions

• Paul
  • To investigate Gold as a VO tool
  • Email Andy and Belinda about text delimitted file format and metadata
  • Investigate GoogleMapCreator from GeoVUE with Andy
  • Carry on as per the presentation - email slides to Andy to be linked here.
  • Outstanding actions from last time
• Andy
  • Make headway on the general aggregation
  • Investigate parallel algorithms for migration
  • Investigate MPICHG2 and differences with MPJExpress
  • Investigate GoogleMapCreator from GeoVUE with Paul
  • Meet with Mark to start joint publication and harmonise ISAR and HSAR
  • Outstanding actions from last time
• Belinda
  • Investigate a profiler for Eclipse
    • This will allow to analyse the time being spent in different functions
  • Use profiler and maybe organise so that a smaller run can be done to get useful profile output
  • Email round the data query passed in the meeting. Date for next meeting and if you can come to a workshop on Feb 8th
  • Outstanding actions from last time
• Mark
  • Meet with finance and report on how much funding we've got.
  • Organise workshop tentatively scheduled for the 8th of Feb to tie in with the social event.
  • Meet with Andy to start joint publication and harmonise ISAR and HSAR

Next Technical Meeting

• Date TBC, Belinda to email possible dates for around 1 months time
• Belinda to chair

References

• Paul's progress report slides (more or less - a recovered older version).
• Moses Technical Meeting.msg, Moses Technical Meeting.txt
• actions-201206.pdf
• Moses Needs Analysis V3 (accepted changes).doc
• NGS-NCeSS-eInfrastructure-20070115.ppt
• Andy Turner's MoSeS Technical Meeting 2006-12-20 Page
• Andy Turner's MoSeS Home Page