ABI Meeting Minutes 2008-08-06

Present: Tommy Yu, Randall Britten, Dougal Cowan, David Cumin, Justin Marsh, Peter Schmiedeskamp, Catherine Lloyd, Andrew Miller, Geoff Nunns, Sarala Dissanayake, James Lawson, Edmund Crampin, Michael Backhaus

Apologies: Mike Cooling, Peter Hunter, Poul Nielsen

Last week's action items 

1.)  Randall will speak to Peter about the possibility of obtaining this funding for Alan [for conference fee and accommodation for ICBME 2008 in Singapore]
  • Peter is happy to provide funding to help Alan attend ICBME2008

2.)  Everyone who is interested should read Andrew's document [on the lambda calculus based typing system for CellML 1.2] and prepare some feedback
  • A number of people have read the document and have begun to provide feedback to Andrew

3.)  Randall will add an item to next week's agenda re: our implementation of the current Simulation Metadata specifications
  • Item moved forward to next week's meeting
Action item 1.)  Randall will add an item to next week's agenda re: our implementation of the current Simulation Metadata specifications

This week's agenda

Summary of the meeting that we had on CellML best practice, Sarala's tool etc.
  • Mike was not able to attend the meeting today, so we will carry this item forward to the next meeting
Action item 2.)  Add this item [Summary of the meeting that we had on CellML best practice, Sarala's tool etc.] to next week's agenda

Update on repository contributions
  • James has been coding up a number of PCEnv SVG-embedded sessions for Peter Hunter's demonstrations at the Montreal meeting on model sharing he is currently attending.
  • PCEnv does not currently create sessions for 1.1 models with imports properly - some tweaking of the URL references within the session file is necessary
  • Catherine has made some changes to a private model which is currently being reviewed for a publication
  • Catherine has has a meeting with Paul Harrington and Kumar Miithraratne regarding the utility of CellML for their modelling work
  • She has already been working on coding up one of Paul's models, but it is giving NaN errors in PCEnv. It needs to use the DAE solver, so she has been using the recent PCEnv snapshots
  • Some of their future work will require CellML models to be imported into openCMISS
  • Andre has decided to get involved in the openCMISS project; specifically with respect to importation of CellML models
  • Catherine has also coded up a new version of the Luo-Rudy dynamic model which no longer uses the iterative IF function in the calcium handling model in 1994 version. This new version is solvable in PCEnv

Progress report on PCEnv / CellML API
  • Justin has been working on a user interface to the integrator's fixed time step / guaranteed tabulation points function that he has implemented in the API
  • Catherine asked whether Justin had managed to find the answer to her problem with differences she has observed between the Linux and Windows builds of PCEnv; specifically, the Linux build of PCEnv is able to run Paul Harrington's model, but the Windows build is not.
  • Justin has not been able to solve the problem, but Andrew suggested it may be something to do with the treatment of floating point numbers being different between platforms
Action item 3.)  Justin to add tracker item pertaining to this issue, Catherine to add example model as test case
  • Andrew has been working on a user interface to the validation service
  • The UI allows users to click on an error message to be taken to the defective part of the CellML model
  • If an equation is invalid, the UI is able to highlight the offending part of the equation
  • He just needs to make a few tweaks and fix a few bugs before it is ready
  • Randall has run some test models with just one DAE in the PCEnv 0.4 release and found that 0.4 runs them without hitch; the latest snapshots are not able to run these test models so smoothly and accurately
  • James noted that some of Jonna's models load and run without a problem in PCEnv 0.4 but give inconsistent errors in the later snapshots. Sometimes they give an integration error message, sometimes they give an overconstrained error and sometimes they load without a problem
Action item 4.)  James will create a tracker item that raises this issue and links to the models he referred to

Progress report on CellML 1.2 specification
  • Randall has read Andrew's proposal for a lambda calculus based typing system for CellML 1.2 and is preparing some feedback
  • Alan Garny has read the document and has given Andrew some feedback. Andrew will be responding to this on the cellml-discussion mailing list
  • Abhishek Tiwari has written a proposal for including support for delayed differential equations in the CellML specification; Andrew has given some feedback on this proposal to Abhishek
  • There was some discussion about what version of CellML Abishek's proposed solutions were targeting. The consensus was that the document should be taken into account for CellML 1.2, but it may also make sense to implement part of it before then.

PMR2 development update
  • Tommy has been continuing his work on providing support for curation flags in the prototype
  • He has been in communication with Alan Garny and Penny Noble regarding their feedback on the PMR2 prototype - see Tracker Item 248
  • He has been working on migrating the CMISS site from Plone 2 to Plone 3.1
  • The CMISS site is considerably smaller than cellml.org, so this can be considered something of a test run for cellml.org
  • It is not possible to migrate PMR1 over to Plone 3.1 (or Plone 3.0 for that matter,) so PMR2 is essentially a blocker to updating cellml.org unless PMR1 is separated from the rest of the site (see further related discussion below.)

Update from cellml.org working group
  • The cellml.org working group have carried out and discussed a second card sort to define the major content sections and subsections of the redesigned site
  • Proposed are:
    •  A 'software' section, which encompasses tools, the API and development roadmaps etc.
    • A 'getting started' elevator pitch
    • A dedicated 'examples / tutorials section,' as Google Analytics indicates that people are particularly interested in the currently out-of-date content that we currently have
  • We have been considering splitting the repository out into a separate subdomain "models.cellml.org"
  • Poul is reluctant to change the address of the repository without solid reasoning
  • www.cellml.org/models could be easily redirected to models.cellml.org
  • Peter Schmiedeskamp supports splitting the repository out from the main cellml.org site, citing advantages with respect to dependencies
  • The disadvantage of requirement for  multiple sign-on / logon could be mitigated by using a federated identity system such as OpenID