<br><font size=2 face="sans-serif"><br>
Alan King<br>
Mathematical Sciences, IBM Research<br>
http://www.research.ibm.com/<br>
http://www.research.ibm.com/people/k/kingaj/<br>
<br>
Campus Relationship Manager: University of Washington<br>
http://www.washington.edu<br>
http://www.ibm.com/university</font>
<br>
<br><font size=2><tt>coin-discuss-admin@www-124.southbury.usf.ibm.com wrote
on 06/22/2004 03:21:09 PM:<br>
<br>
> I'm currently sitting in on a course on stochastic programming. There<br>
> are about a dozen Ph.D. students in the class, all of whom must do
some<br>
> computational exercises (such as implementing the L-shaped method)
as<br>
> well as a final project. If I act fast I may be able to convince
one or<br>
> more of them to do final projects that also contribute to SMI.<br>
> <br>
> However, I'm having a little trouble understanding what SMI can already<br>
> do and what it aspires to. I've been looking through the documentation<br>
> and code, including the unitTest code.<br>
> <br>
> I'd appreciate it if someone familiar with SMI could answer these
few<br>
> questions:<br>
> <br>
> 1. Current capabilities of SMI: store scenario-based problems,
read<br>
> SMPS format, use OSI to solve deterministic equivalent (extensive<br>
> form). Is that correct? </tt></font>
<br>
<br><font size=2><tt>Yes: with one addition -- also have ability to store
problems with independent random vectors.</tt></font>
<br><font size=2><tt><br>
> <br>
> 2. Would an L-shaped method algorithm be a good addition? Is
one<br>
> already in the works? Are other solution methods particularly
desirable<br>
> at this point?</tt></font>
<br>
<br><font size=2><tt>Yes: an L-shaped would be a good addition. I
am planning to do one but won't </tt></font>
<br><font size=2><tt>get around to it for a while.</tt></font>
<br><font size=2><tt>Also interesting (and perhaps easier to implement)
a progressive hedging method. JJF has code the quadratic already,
so should be straightforward to implement.</tt></font>
<br><font size=2><tt><br>
> <br>
> 3. Is it correct that adding chance constraints would involve<br>
> developing a new class? (For starters, I guess it would store
the<br>
> description of the chance constraints and would also provide a solution<br>
> through deterministic equivalents for the basic cases.) <br>
> </tt></font>
<br>
<br><font size=2><tt>Hmm. I guess I would subclass SmiScnModel for
this (stick to discrete distributions) and then have a method "declareChanceConstraint"?</tt></font>
<br><font size=2><tt><br>
> 4. Any other ideas for projects that might be useful at this
point?</tt></font>
<br>
<br><font size=2><tt>Would be nice to have:</tt></font>
<br><font size=2><tt> declare integer
variables and propagate through deterministic equivalent</tt></font>
<br><font size=2><tt> quadratic objective</tt></font>
<br><font size=2><tt> more classes of
stochastic random variables -- especially arma type</tt></font>
<br>
<br><font size=2><tt>> <br>
> Students are supposed to decide their final project topics by the
end of<br>
> the week, so a response within a day or two would be appreciated.
<br>
> Thanks,<br>
> <br>
> Brady<br>
></tt></font>
<br>
<br><font size=2><tt>Thanks!</tt></font>
<br><font size=2><tt>Alan</tt></font>
<br><font size=2><tt> <br>
> -- <br>
> Brady Hunsaker<br>
> Assistant Professor<br>
> Industrial Engineering<br>
> University of Pittsburgh<br>
> http://www.engr.pitt.edu/hunsaker/<br>
> <br>
> <br>
> _______________________________________________<br>
> Coin-discuss mailing list<br>
> Coin-discuss@www-124.ibm.com<br>
> http://www-124.ibm.com/developerworks/oss/mailman/listinfo/coin-discuss<br>
</tt></font>