AI Reference Panel Meeting No. 1

[sarahcleary]

The first AI RP meeting is planned for 17/2.
If you have agenda items for this meeting, can you please add them (as you think of them) to this issue?
thanks.

[sarahcleary]

• Discussion on APSIM Special Section / Book

[Keith-Pembleton]

Drafting a paper on the best management/governance of a model?
Promotion and training for next gen?

[sarahcleary]

@APSIMInitiative/reference-panel - any further ideas, agenda items?

[sarahcleary]

@uqghamme

[sarahcleary]

Development plan

• what's new? aspirational? new developments?
• what sprints are required?
• topics?
• who should be involved?
• how to get engagement with the right people

[uqghamme]

How do we optimise engagement/collaboration/communication on design of new elements BEFORE they are implemented

• reinvigorate science-software nexus
• work better across organisations
• involve interested parties at design phase
• have SC fund specific design workshops where needed?

[sarahcleary]

@hut104 - any suggestions for agenda items for the upcoming AI RP meeting (17/2)?

[sarahcleary]

Any further input? any other topics for discussion?

[sarahcleary]

@jwhish

[sarahcleary]

Please find attached a draft agenda for your input.

As you are aware, this is our first RP meeting with a longer format, replacing the monthly 2 hour meetings. If there are any additional topics you wish discussed, please let me know.
Please ensure you have reviewed GitHub recently and provide any feedback on relevant issues.

Time	Topic	Lead
8:50	Welcome	Jason Brider @jbrider
9:00	How do we optimise engagement/collaboration/communication on design of new elements BEFORE they are implemented: reinvigorate science-software nexus work better across organisations involve interested parties at design phase have SC fund specific design workshops where needed?	Graeme Hammer @uqghamme
10:00	Development Plan Update on recent sprints – Autodoc/Stock what's new? aspirational? new developments? what sprints are required? topics? who should be involved? how to get engagement with the right people	Jason Brider @jbrider
10:30	Drafting a paper on the best management/governance of a model	Keith Pembleton @Keith-Pembleton
11:00	Training Update on training sub-committee Promotion and training for next gen	Training ‘sub-committee’ @kchenu @Keith-Pembleton @JulianneLilley @sno036 @peter-devoil
11:30	Discussion on APSIM Special Section / Book	Karine Chenu @kchenu
12:00	Close

We had booked from 8-2, but I don’t think we need all of that time – but can call for additional topics.
As discussed, this meeting is to bring together people beyond the @APSIMInitiative/reference-panel, so please invite others you feel can contribute to these discussions.

I note there are a few workshops which have been discussed but never actioned. There is the possibility of utilising the 12-2pm window for such a discussion. If anyone would like to lead one of these, please let me know.

[sarahcleary]

@erik-van-oosterom; @uqschap7

[sarahcleary]

2021-02 Minutes

Members: Video/Teleconference:

@yashvirchauhan; @MarkLieffering; @LouisAK; @peter-devoil; @sarahcleary; @Keith-Pembleton; @kchenu; @jbrider; @sno036; @HamishBrownPFR; @JulianneLilley; @kchenu; @sarchontoulis; @hol353 @erik-van-oosterom; @hut104; @uqschap7; @uqghamme;

Apologies: @EnliWang

Wednesday 17th February 2021 - 8:50 am AEST

Agenda Item

Welcome

Welcome from @jbrider to first meeting of the year. Noting this is the first meeting with the new longer format

How do we optimise engagement/collaboration/communication on design of new elements BEFORE they are implemented:

• reinvigorate science-software nexus
• work better across organisations
• involve interested parties at design phase
• have AI fund specific design workshops where needed?

@uqghamme worked through the attached presentation
RP meeting discussion Feb21.pptx

Split into 4 breakout rooms for 15 minutes and asked to consider the following questions

1. Do we need to enhance communication on science-software issues related to APSIM development across interested parties? (i.e. beyond RP, organisational groups, and github) If so, how?
2. Is it desirable to have involvement of interested parties in design of new module/process algorithms before they are implemented? If so, how?

Key points from Group 1:

• agree with sentiment
• note there have been success stories for APSIM. An example is the wheat phenology where 20+ people met initially. Meetings still occur bringing the software and science together, with the project meetings occurring weekly
• however, very limited interaction across crops.
• require, improved communication when implementation is occuring that impacts other crops

Key points from Group 2:

• Discussion focussed on GitHub
• How it does or doesn’t work
• Set up with Teams – more about notifications than communication; needs to be about communication.
• Noted sprints work well around a single topic
• Need to use github as a lead up to the event – should use github to contribute to that discussion but does not replace it.

Key points from Group 3:

• Q2 – yes
• Potential solution – might work well, if the AI had someone akin to a chief scientist.
• This role could be mandated to consult widely in the APSIM community and the advise which way integration goes
• Counters the current panel approach

Key points from Group 4:

• Discussed specific issues - examples, communication
• Noting there will legacy issues
• Multiple ways of doing the same thing – large infrastructure things
• Yes, different philosophy; some are fundamentally different
• Agree if do both – get complex like APSIM Classic.

Group Discussion

Further discussion on "chief scientist concept"

• “Module convenors” existed 20 yrs ago which was akin to a chief scientist for areas.

Further discussion on GitHub

• need relevant notifications; but managable notifications
• Agreed to utilise Teams within GitHub allowing 'tagging' to receive relevant notifications.
• Look to see if administrator can inform people of use of teams in GitHub.
• Review what teams are required and what the purpose of teams are - is it for communication which crosses multiple issues/areas/

ACTION: Develop a step by step document and related training video on how to use GitHub for AI purposes.

Noted that GitHub has its use - but does not replace the whiteboard discussions.

Development Plan

Update on recent sprints – Autodoc/Stock

• what's new? aspirational? new developments?
• what sprints are required?
• topics?
• who should be involved?
• how to get engagement with the right people

Update on Sprints

• Autodoc enhancement - Neil, Hamish, Sotirios, Dean
  -- worked well; acheived goals

• Stock Sprint - 3 days – Nev, Val, Dean, Mark – resulting Stock now in release
  -- Noted that Reference Panel preapproval greatly assisted to get Stock into release
  -- Made it possible to plan what to do in the sprint
  -- Utilised projects feature in GitHub

Role of Sprints – work well; but need to ensure people are aware these are happening in case they wish to be involved or wish to be involved in pre-planning.

Reviewed the dev plan

Improved Soil Water balance – it's an educational exercise at present, but might require further discussion/input if promising

@jbrider to add issues to the Dev plan - follow up @sme016 re:eucalyptus development
and hourly temp data development

Drafting a paper on the best management/governance of a model

Looking outside how other models do it? Is what we doing best practice??
Benefit as part of a paper?
Increase prestige of APSIM, beyond cropping/farming systems
Broaden to open source projects in general – not just modelling
Look at how does “R” work?
Scan the journal papers

ACTION:
@Keith-Pembleton to lead
@sno036 to provide initial comment

Review to see if publishable.
Review other models, and/or other platforms
Value for the AI as well – guide how we improve our processes.
Potential to conduct a workshop at ModSIM conference on Agri sytems processes
as part of the

Training

• Update on training sub-committee
• Promotion and training for next gen

@Keith-Pembleton summarised the sub-committee meeting
AI wants to push people along the pipeline – so can contribute to APSIM
Training plan for 2021 seeks to do this.

Open call for additional videos.
Welcome videos which demonstrate applications of research (such as UQ crop physiology lectures)
Simulations will be developed

Discussion on academic working group

Noted that there would be synergies between academic teams.
Discussed potential of a Community of practice for the education setting
Offer AI support if requested.

Discussion on APSIM Special Section / Book

Karine led discussion on reviewing options around special section / book

Difficulties include:

• Cost money
• No. of Quality papers
• Open invite - too many invitations; poor quality?

Possible - choose theme and by invitation only
Possible - In silico - should be indexed next year

• Need 3 guest editors for a special edition
• Possible to 'match' researchers for a review paper

To propose a special edition, require the following

• Short description
• List of guest editors – need 3 – Matt H, David S, and possible earlier career e.g. Ando and/or Jony - depending on capacity etc..; Karine to help. Prefer geographical differences for editors
• 5 possible papers – couple of lines, and authors
• Summary of editorial

@kchenu to contact Matt Harrison as potential lead Guest editior. If positive, ask Matt to contact David etc..

If theme is right – Val happy to do a pre assessment for Agricultural Systems .

Strategy Meeting

Include response to Graeme’s challenges as agenda item.

[sarahcleary]

@Keith-Pembleton - any update on "Drafting a paper on the best management/governance of a model"?
Shall I open a separate issue on this?
Include as an agenda item for the next meeting?