Home » Archived » AMP » Model/Experiment Management For AMP
| Model/Experiment Management For AMP [message #559580] |
Fri, 17 September 2010 09:22  |
 Tony Kennedy
Messages: 17
Registered: June 2010 |
Junior Member |
|
|
This post is in the context of a series of four discussion between myself and the sponsors / users of the models produced and expresses there concerns as to how we ensure the provenance and integrity of the results,ensure modeling transparency,model reuse,preservation of model and experimentation descriptions,model coupling,work flow discovery,validation manage risk and costs (don't knock its a sign that ABM is entering the world of big business you may soon be able to eat as well as model); this is in the wake recent financial constraints, and the debark of the environmentalist missing emails and "adjusted" data of the last few months and the subsequent growth of blogs such as bad science who will leap at any chance to discredit results from any source not ensuring the data's provenance is impeccable, delivered on time and costs are minimised, and can be understood by any one from the age of 10 upwards. (some of the UK media publish at a reading age of 10).
Two of four discussion have taken place the first introduced AMP its position in the modelling world and technologies used in particular meta models and DSL's; and AMP's capabilities, the second introduces how we satisfied the above requirements i.e. Experiment management, the technologies introduced i.e. Mylyn its capabilities and its extension by AMP's meta modeling techniques, EMF and xText, to give transparency,reuse,preservation,coupling, work flow discovery and validation; The introduction of BIRT and the standards based repository development,its mining and discovery. so satisfying their concerns while maintaining a low diversity project, the discussions are at a low / medium technical level.
The third will discuss the introduction development of specialist domain requirements on the agenda so far are: X-Streams (financial domain), GPU, Grid, Embedding, dynamics, Meta model extensions, AMP functional extensions, cross platform coupling; (general cross domain),Map Algebra's, specialist techniques i.e. Generalized Proximity Matrix etc., development of an ABM extension for Geotools, Degree, Kalipso, Hydra comparative studies against start from scratch (GIS based domains).
The fourth will discuss legacy model porting, refactoring, tooling; on the agenda so far: automatic translation techniques, unit normalization,examples from the principal domains.
If any of these are of interest I will post them in full, or in part, together with examples at present they are in the form of meeting notes but will be developed into project proposal's if approved.
|
|
|
| Re: Model/Experiment Management For AMP [message #559710 is a reply to message #559580] |
Fri, 17 September 2010 16:20  |
Miles Parker
Messages: 1341
Registered: July 2009 |
Senior Member |
|
|
Tony, this is awesome. I assume this means that you find the basic AMP / AMF approach and community fits your groups needs. reL communications, at this point, as we are getting into nitty gritty, let me suggest that we take Werner's suggestion and move this into amp-dev, especially the proposal side. Please let me know if there is anything I can to clarify process on this end, but basically we'll keep it as light-weight as possible. If you'd like to schedule a conf call at some point, then we can do that using Eclipse services.
re: process itself, I've become a firm believer in build first and let everything else fall in place behind that.  I know that's the opposite of how one is supposed to do software, but by slapping the agile label on it we can get away with it.  Seriously, if at any point you have functionality that you want to get into the code base, let's open a bugzilla and just do it. That's one big advantage of being at 0.8.0 / Incubation - there is a lot of flexibility to try things out.
There have been discussions about widening/deepening the scope of the overall project, and I'd love to have feedback from you and others on how we might do that. It's become apparent to me that the positioning of the project itself as "Agent-Based" has meant that people don't see what it can do in a broader context, which is a real issue.
| Tony Kennedy wrote on Fri, 17 September 2010 05:22 | This post is in the context of a series of four discussion between myself and the sponsors / users of the models produced and expresses there concerns as to how we ensure the provenance and integrity of the results,ensure modeling transparency,model reuse,preservation of model and experimentation descriptions,model coupling,work flow discovery,validation manage risk and costs (don't knock its a sign that ABM is entering the world of big business you may soon be able to eat as well as model); this is in the wake recent financial constraints, and the debark of the environmentalist missing emails and "adjusted" data of the last few months and the subsequent growth of blogs such as bad science who will leap at any chance to discredit results from any source not ensuring the data's provenance is impeccable, delivered on time and costs are minimised, and can be understood by any one from the age of 10 upwards. (some of the UK media publish at a reading age of 10).
Two of four discussion have taken place the first introduced AMP its position in the modelling world and technologies used in particular meta models and DSL's; and AMP's capabilities, the second introduces how we satisfied the above requirements i.e. Experiment management, the technologies introduced i.e. Mylyn its capabilities and its extension by AMP's meta modeling techniques, EMF and xText, to give transparency,reuse,preservation,coupling, work flow discovery and validation; The introduction of BIRT and the standards based repository development,its mining and discovery. so satisfying their concerns while maintaining a low diversity project, the discussions are at a low / medium technical level.
The third will discuss the introduction development of specialist domain requirements on the agenda so far are: X-Streams (financial domain), GPU, Grid, Embedding, dynamics, Meta model extensions, AMP functional extensions, cross platform coupling; (general cross domain),Map Algebra's, specialist techniques i.e. Generalized Proximity Matrix etc., development of an ABM extension for Geotools, Degree, Kalipso, Hydra comparative studies against start from scratch (GIS based domains).
The fourth will discuss legacy model porting, refactoring, tooling; on the agenda so far: automatic translation techniques, unit normalization,examples from the principal domains.
If any of these are of interest I will post them in full, or in part, together with examples at present they are in the form of meeting notes but will be developed into project proposal's if approved.
|
|
|
|
Goto Forum:
Current Time: Thu Apr 25 13:37:47 GMT 2024
Powered by FUDForum. Page generated in 0.03046 seconds |
] 
Search
Help
Register
Login
Home
