Re: [Help-glpk] Dual variables as result from a LP solver

[stilak1]

Hi Everyone,

I want to know has anyone tried MIP for 0-1 knapsack problem using GLKP ?
I want to try few approximate techniques since the problem itself is
NP-hard.

Are there any specialized programs with more sophisticated techniques that
I should try as well ? Anyone with experience of GLPK for similar problem,
please let me know.

-- Sameer.



On Wed, 22 Sep 2004, Brady Hunsaker wrote:

> On Wed, 2004-09-22 at 09:03, address@hidden wrote:
> > Hi subscribers,
> >
> > I want to solve several hundred LPs to get the results of the dual
> > variables.
> > Naturally I could do some coding within GLPK, but I don't want to
> > reinvent the wheel. Has someone done this?
> > An alternative would be to convert an original problem to the dual
> > problem. Does someone know a free program, who can handle this?
> >
> > All suggestion are welcomed,
> > Harald.
> >
>
> If using the API, use the routine lpx_get_row_dual (and possibly
> lpx_get_col_dual) after you have solved the LP.  I find that different
> texts and solvers sometimes treat sign (plus/minus) differently, so you
> may want to be sure that the sign convention GLPK uses is what you
> expect.
>
> If you don't want to use the API, glpsol's output with the --output
> option includes a column "Marginal", which contains the dual values.
> For several hundred LPs this probably doesn't make sense unless you
> write a parser to extract the dual values from the output.  But using
> the API is probably easier.
>
> Brady
>
> p.s. Harald, I have not forgotten your other example, but this month has
> been too busy for me to do much.  I do hope to look at it within a few
> weeks.
>
>