-
redirection
I'm not posting in this journal anymore - Any posts of mine, work-related or otherwise, are now going into my regular blog, joecarnahan. -
Plan for proposal presentation
(Once again, I am finding myself in need of somewhere to type out plans and thoughts. So, the research journal lives on...)
I am trying to work out a plan for my proposal presentation - Here is the tentative outline:
- Problem Description and Motivation
- The need for simulation adaptation
- Simulation reuse under changing requirements
- ( Read more... )
- The need for simulation adaptation
- Problem Description and Motivation
-
Assumptions, assumptions, assumptions!
They say that when you assume, you make an "ass" out of "u" and "me".
It's a cute play on words, but not necessarily true. If it were, you'd hear nothing but endless braying from the simulation community, since we make countless assumptions every single day.
Spiegel is presenting a paper at this year's Winter Simulation Conference that discusses ( Read more... ) -
Password-protected web sites and the hazards of blogging
I haven't been writing in this journal much lately, mostly because I've started a proposal web site. So, I've been taking my notes there instead of here. The thing is, that site is currently password-protected, and right now nobody except Paul or I can read it. (Well, OK, anybody in the CS department can read it with cd /home/jcc5t/public_html/ ( Read more... ) -
relaxing compositional modeling assumptions
Here are some examples, brainstormed with Paul last week, for ways compositional modeling assumptions could be partially relaxed:- Same variable can be causally determined by more than one assumption class, but no transitive chains of causal influences
- Alternately, only linear (no branching) chains of transitive influences
- Contradictions between ( Read more... )
-
Clarification of compositional modeling stuff
I've continued to be thinking about compositional modeling, and I've gotten some good feedback from both Paul and Rob. The short version is that we've decided that the problem of simulation adaptation reduces to the problem of fragment selection if you accept the compositional modelers' assumptions ( Read more... ) -
Simulation adaptation as a compositional modeling problem
It is possible to describe the problem of simulation adaptation as a variation of compositional modeling. Consider Nayak's statement of the compositional modeling problem (paraphrased):Given- a query consisting of
- a list of output variables whose behavior needs to be understood
- a (possibly empty) list of driving variables that need to be included in ( Read more... )
- a query consisting of
-
Qualitative process theory
Just finished a couple readings on qualitative process theory by Forbus (as in the Falkenheimer-and-Forbus paper that introduced compositional modeling) and Kuipers. The bottom line is that I already learned most of what I needed to know about QP from the papers on compositional modeling, though I'm glad that I've now seen the fundamental papers ( Read more... ) -
Analysis of compositional modeling
The compositional modeling work that I have been reviewing over the past two weeks has included work by- Brian Falkenheimer and Kenneth Forbus
- Alon Levy and Yumi Iwasaki
- P. Pandurang Nayak
- Jeff Rickel and Bruce Porter
-
Update, part 2: Describing behavior
Even with a flawless system for applying the code changes associated with various flexible point bindings, simulation coercion is still not useful unless we can leverage expert insight to semi-automatically determine which changes to apply. In other words, we need to know what each change does and how those changes relate to a stated requirement ( Read more... )