XMILE Draft Spec version 0.1

This forum is for discussion of the System Dynamics model representation standard XMILE. Discussion here will be monitored by the Society's technical standards committee with ideas and concerns conveyed to the OASIS Technical Committee responsible for defining the standard.
Forum rules
Please note: By posting here you grant permission to the Society Technical Committee members to repost with attribution on the OASIS discussion forum. If you have material for which you wish to maintain copyright control please post a link to the copyrighted work.
Post Reply
Robert Eberlein
Site Admin
Posts: 179
Joined: Sat Dec 27, 2008 8:09 pm

XMILE Draft Spec version 0.1

Post by Robert Eberlein » Fri Jan 31, 2014 9:02 pm

Hi Everyone,

The XMILE Technical Committee has published a draft specification version 0.1 (as opposed to 1.0). We would very much like to get feedback on that specification and I encourage you too look it over. It is a very early draft, so there are is a lot to be done still, but the sooner we get comments the easier they are to incorporate.

The spec can be downloaded from the OASIS site: https://www.oasis-open.org/committees/d ... %200.1.doc

Bob Eberlein

Travis Franck
Posts: 34
Joined: Sun Jan 11, 2009 7:48 pm

Re: XMILE Draft Spec version 0.1

Post by Travis Franck » Mon Feb 03, 2014 2:25 pm

Bob,

What is the best way to provide feedback?

Question, that might be feedback: Is there anything in the standard that can help ensure the modeling results are the same?

The problem I see: We reverse engineered Vensim solver recently for a mdl to Ruby translator. We read the Vensim manual and spec about equation ordering and tried it. Got different results. Tried it a second way -- got difference results. Tried a a third way... We finally worked out the subtleties of equation ordering to make sure that the results were the same.

But what will happened when two groups "run" a XMILE described model through their own solvers? What is "right"?

We would argue that that the first several ways we tried were 1) same as Vensim documentation and 2) mathematically correct.

Travis

Travis Franck
Posts: 34
Joined: Sun Jan 11, 2009 7:48 pm

Re: XMILE Draft Spec version 0.1

Post by Travis Franck » Mon Feb 03, 2014 4:19 pm

I mentioned this post to a colleague that did most of the translator work. He commented that the equation ordering was one thing, but some of the less-basic variable types have specific initialization orders and means that aren't mathematically necessary (or well documented).

Does the XMILE standard "standardize" this?

Robert Eberlein
Site Admin
Posts: 179
Joined: Sat Dec 27, 2008 8:09 pm

Re: XMILE Draft Spec version 0.1

Post by Robert Eberlein » Tue Feb 04, 2014 6:31 am

Hi Travis,

The good news is that the functions and their operation are pretty clearly laid out. The bad news is that the focus in the standard is on the more-basic functions and variable types.

As far as equation ordering, that is useful feedback. We can probably expand the writeup in this area a little bit. Once again the spec is focusing on the core functionality which is largely unambiguous (some things like the order of computation of stochastic functions are not completely determinant and there is little to do other that note that). The really interesting (hair-pulling) cases here are the less standard one (eg managing rates in order to keep stocks positive) where the solutions do involve a bit of art.

All that said, the intent is to have a pretty good suite of models with their reference behavior so that compliant implementations have a basis for testing and - yes - getting the same results.

Post Reply

Who is online

Users browsing this forum: No registered users and 1 guest