Re: [Help-glpk] glpsol, arbitrary precision and large numbers

[Robbie Morrison]

Hello Edd

------------------------------------------------------------
To:           Robbie Morrison <address@hidden>
Subject:      Re: [Help-glpk] glpsol, arbitrary precision and large numbers
Message-ID:  <address@hidden>
From:         Edd Barrett <address@hidden>
Date:         Fri, 29 Jun 2012 18:57:34 +0100
------------------------------------------------------------

> On Sat, Jun 30, 2012 at 04:34:19AM +1200, Robbie Morrison wrote:

[snip]

> Ah, I am failing the KKT.PB check. From the .sol file
> I attached in my earlier mail:
>
> ---8<---
> Integer feasibility conditions:
>
> KKT.PE: max.abs.err = 0.00e+00 on row 0
>         max.rel.err = 0.00e+00 on row 0
>         High quality
>
> KKT.PB: max.abs.err = 5.00e+00 on row 15
>         max.rel.err = 1.00e+00 on row 15
>         SOLUTION IS INFEASIBLE
> ---8<---

Yeh, I saw that earlier too.

> So I guess the result is unreliable. I will have to
> read into this more.

The information in the wikibook I mentioned earlier
might be a good place to start:

  
http://en.wikibooks.org/wiki/GLPK/Solution_information#Karush-Kuhn-Tucker_optimality_conditions

More KKT theory here too:

  
http://en.wikibooks.org/wiki/GLPK/Background_theory#Karush-Kahn-Tucker_.28KKT.29_optimality_conditions

I think richer return codes from GLPSOL could help in
these kind of cases.  That said, I use the API and can
easily get this information programmatically.

> Cheers for everyone's help.

good luck with the bug hunt, Robbie
---
Robbie Morrison
PhD student -- policy-oriented energy system simulation
Technical University of Berlin (TU-Berlin), Germany
University email (redirected) : address@hidden
Webmail (preferred)           : address@hidden
[from Webmail client]