[Coin-discuss] Using Osi/Clp

[Matthew Saltzman]

On Thu, 6 Jan 2005, Edgardo Fuchs wrote:

> Dear Coin list members,
>
> I wish to use Osi/Clp within a stochastic optimization model to solve a
> long-term hydro-thermal power generation dispatch problem.
> The model was developed using OSL, but I want to migrate it to Osi/Clp.
>
> >From the example "build.cpp" it is clear how to construct a model from
> scratch and solve it. But I have not found the way to perform additional
> operations required by the dispatch model.

Yes, documentation is kind of sparse.  But if you have a reading knowledge
of C++, you can find most of what you need looking at the OSI code itself.
See the Doxygen docs at http://www.coin-or.org/Doxygen/Osi/index.html.
There are also some other useful examples in the Examples directory of the
source tree and at http://coral.ie.lehigh.edu/~coin/.

> Could you please give some hints or examples on how to perform the
> following operations using Osi/Clp?

I'd suggest looking at the above locations first.  If you still have
questions, please come back and ask them here.

>
> 1. Select primal or dual simplex algorithm

Look for [get/set]HintParam() (as well as the other [get/set]*Param
methods.

> 2. Add a column to the LP matrix
> 3. Add a row to the LP matrix
> 4. Delete and replace a row of the LP matrix
> 5. Replace elements of the LP matrix

All the above methods are in the OsiSolverInterface class.

> 6. Post solve after modifications

The first solve is initialSolve().  After mods, resolve with resolve().
Not all solvers do this more efficiently than resolving from scratch, but
most do.

> 6. Scale LP problem

Hints.

> 7. Save and load an optimal basis

[get/set]WarmStart()

> 8. Direct Osi/Clp messages

See the CoinMessageHandler class.

>
> Thank you in advance.
> Sincerely,

HTH.

>
> Edgardo Fuchs
> KAS Ingenieria
>
>  --
>
> Edgardo Fuchs
>
> Santiago, Chile
>
> Tel: +56 (2) 435 0090
> Cel: +56 9837 8433
>
>

-- 
		Matthew Saltzman

Clemson University Math Sciences
mjs AT clemson DOT edu
http://www.math.clemson.edu/~mjs