[Coin-SMI] About SMI class hierarchy.
Alan King
kingaj at us.ibm.com
Thu Apr 7 14:46:50 EDT 2005
Dear Fulvio
My humble apologies for the documentation. Please continue to study SMI
and C++! Perhaps you can contribute some documentation as you go along...
:-)
The hierarchy is very flat. The SmiScnModel class encapsulates the
relationship between an OsiSolverInterface and the
SmiScenarioTree<SmiNode> class. The SmiScenarioTree is just a tree with
interfaces that allow one to add a "path to leaf" to the tree.
SmiScenario<SmiNode> is a tree with data elements that are SmiNodes.
SmiNodes implement the "core block-stoch block" pattern described in the
SMPS standard. An SmiScenarioTree<SmiNode> describes a stochastic program
in the SMPS "scenarios" format. The SmiScnModel class has methods that
transform the SmiScenarioTree<SmiNode> structure into an OSI compatible
linear program, and allow for "scenario-stage" semantics for querying data
and solutions.
I am thinking of working with SmiScenarioTree a bit to make it a little
easier to post and retrieve data to nodes. Probably this will take the
form of a virtual base class for the post/retrieve functionality. Does
anyone have objections to the fact that SmiScenarioTree is templated?
Should I remove the templating and just use the post/retrieve classes to
manage data?
So many questions, so little time.
Alan
coin-smi-bounces at list.coin-or.org wrote on 04/06/2005 03:45:18 AM:
> Hi
>
> I am beginning to use SMI. Since I am new both to object oriented
> programming and to stochastic programming I have a lot of questions
about
> SMI. Just start with one.
>
> Looking at the automatic documentation, the class hierarchy is quite
weak
> and the SmiScnModel class does not appear at all. I guess that the
> SmiScnModel class should stay on the top of a diagram under which I will
> put SmiScnData and SmiDiscreteDistribution classes.
> Is this correct or am I wrong (and will study more SMI and C++)?
>
> Thanks for your answers.
>
> Before posting this message I give a look at the SMI mailing list
archive.
> I don't know why but I did not receive all the messages, so I read only
> today about the proposal of implementing the first example in Birge and
> Louveaux book. It is exactly what I am doing, so I can post the SMPS
> files and the code to solve the model starting from them in my next
email.
>
> Fulvio
>
> _______________________________________________
> Coin-SMI mailing list
> Coin-SMI at list.coin-or.org
> http://list.coin-or.org/mailman/listinfo/coin-smi
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://list.coin-or.org/pipermail/smi/attachments/20050407/273c7b9a/attachment.html
More information about the Coin-SMI
mailing list