[kepler-dev] Using Kepler for Inductive Process Modeling?

[Christopher Brooks]

Hi Rich,
I'm not familiar with Pat Langley's Inductive Process Modeling
Techniques, but I've addressed your questions below.

On 10/25/13 7:12 PM, Rich Morin wrote:
> I am working with a research system that implements Pat Langley's
> Inductive Process Modeling techniques.  By way of background, it:
>
>    *  allows modelers to specify abstract collections of models,
>       using definitions of constraints, entities, and processes
>
>    *  defines entities in terms of attributes and processes in
>       terms of algebraic and differential equations
>
>    *  constrains process use in instances (eg, ALWAYS-TOGETHER,
>       ATMOST-ONE, EXACTLY-ONE, FORBIDDEN, NECESSARY)
>
>    *  generates and evaluates specific model instances, using
>       randomized subsets of the possible entities and processes
>
>    *  allows inspection of the best-scoring model instances
>
>    *  provides assistance in modifying specifications
>       (eg, adding constraints, extending or simplifying models)
>
> I would like to assess the fit between these capabilities and
> Kepler.  If Kepler already supports any of these (or there is
> an obvious way to add them), I'd love to get caveats, pointers,
> and suggestions.  And, if some issue looks like a show-stopper,
> I'd really like to hear about it...
>
> -r
>
>
> Specific questions
>
> Assume that some program (possibly set up as a Kepler plugin)
> has generated some Ptolemy model descriptions (eg, MoML and
> associated files):
>
>    http://ptolemy.eecs.berkeley.edu/publications/papers/00/moml/moml_erl_memo.pdf
>
> We now want to evaluate the models (eg, make sure they converge,
> fit their parameters to training data, solve sets of equations)
> and present the "best of breed" to the modeler for inspection.
>
> Q:  Is there an API for evaluating a single model?
Yes, see 
https://kepler-project.org/developers/reference/executing-kepler-from-the-command-line

Kepler uses Ptolemy as its execution engine, so you might find 
http://ptolemy.eecs.berkeley.edu/ptolemyII/ptIIfaq.htm#invoking%20Ptolemy%20II? 
of use.
>
> Q:  Is there an API for evaluating a set of models?
See
http://chess.eecs.berkeley.edu/ptexternal/wiki/Main/ParameterSweeps#in_browser
>
> Q:  Can Kepler auto-layout the diagram for the model graph?
The devel version of Kepler has View -> Automatic Layout  and
View -> Configure Layout

This uses the layout tools from the University of Kiel, see
http://www.informatik.uni-kiel.de/en/rtsys/kieler/

There is a test driver at 
ptolemy/vergil/basic/layout/kieler/test/junit/KielerJUnitTest.java
that could be used as the basis of a standalone tool to layout models 
from the command line.
>
> Q:  If not, are there any examples of using (say) GraphViz to
>      generate layout metadata (eg, node positions and sizes)?
Kieler might have some GraphViz code somewhere, but I'm not sure.

The kepler-users mailing list has some traffic:
http://www.mail-archive.com/search?l=kepler-users%40kepler-project.org&q=graphviz&submit.x=0&submit.y=0

See http://ptolemy.eecs.berkeley.edu/ptolemyII/ptIIfaq.htm#MoML
and https://code.kepler-project.org/code/kepler/trunk/modules/moml2graphviz/
>
> Q:  Is there an API for adding models to Kepler's inventory,
>      taking advantage of Kepler's notion of provenance, etc?<
I'm not that familiar with how the Kepler repository works.   Start with
https://kepler-project.org/users/faq#what-is-the-kepler
What is the Kepler repository and how can I use it?
>
> Q:  Is there an authoritative archive of the material on GSRC
>      that was stored on "http://www.gigascale.org/semantics"?
>      This URL is referenced in
>
>        MoML - A Modeling Markup Language in XML - Version 0.4
>        http://ptolemy.eecs.berkeley.edu/publications/papers/00/moml
There is probably an archive somewhere, but digging it up would take time.
Did you have specific questions?

_Christopher
>
>   --
> http://www.cfcl.com/rdm            Rich Morin
> http://www.cfcl.com/rdm/resume     rdm at cfcl.com
> http://www.cfcl.com/rdm/weblog     +1 650-873-7841
>
> Software system design, development, and documentation
>
>
> _______________________________________________
> Kepler-dev mailing list
> Kepler-dev at kepler-project.org
> http://lists.nceas.ucsb.edu/kepler/mailman/listinfo/kepler-dev


-- 
Christopher Brooks, PMP                       University of California
Academic Program Manager & Software Engineer  US Mail: 337 Cory Hall
CHESS/iCyPhy/Ptolemy/TerraSwarm               Berkeley, CA 94720-1774
cxh at eecs.berkeley.edu, 707.332.0670           (Office: 545Q Cory)