From joost at dsj.nl  Fri Feb  3 05:58:27 2012
From: joost at dsj.nl (Joost)
Date: Fri, 3 Feb 2012 11:58:27 +0100
Subject: [Symphony] solution export and var names
In-Reply-To: <CA+GYyctYKX72ZCmEqu8OTAv2YKaCdaChbeFxuootpoa=DvB0ag@mail.gmail.com>
Message-ID: <FFE80205F88B490998884211C76CC73B@joost>

Thanks for looking into this.

Another useful link on Kappa:

http://publib.boulder.ibm.com/infocenter/cosinfoc/v12r2/index.jsp?topic=
%2Filog.odms.cplex.help%2FContent%2FOptimization%2FDocumentation%2FCPLEX
%2F_pubskel%2FCPLEX429.html

In my experience a 'higher' condition number makes a problem (much)
harder to solve, and is a signal to check problem data and formulation.

I will look into CglGomory.cpp and try to compute the condition number
first for the root node.

Joost

-----Oorspronkelijk bericht-----
Van: Ted Ralphs [mailto:ted at lehigh.edu] 
Verzonden: zaterdag 28 januari 2012 16:46
Aan: Joost
CC: symphony at list.coin-or.org
Onderwerp: Re: [Symphony] solution export and var names

2011/12/18 Joost <joost at dsj.nl>:

> I ran into the same problem as described here:
> http://list.coin-or.org/pipermail/symphony/2011-March/000592.html
(using
> Symphony 5.4.3). The same model solved within 30min to optimality with
> Symphony 5.1.2


From joost at dsj.nl  Fri Feb  3 07:45:24 2012
From: joost at dsj.nl (Joost)
Date: Fri, 3 Feb 2012 13:45:24 +0100
Subject: [Symphony] symphony
In-Reply-To: <CA+GYycua9T9o+5qQY5XFuXpgHN2Mr4k69JR1tdSpoxxDs70phw@mail.gmail.com>
Message-ID: <27B67073B0BC424194532488F970848F@joost>

First of all thanks for looking into this.

Re-directing the output to a file in a black box use case and then
parsing it, is indeed a good option.

In addition to the significant digits, the variable names are also
truncated. If you use a math program generation language with
verbose/descriptive variable names, then you loose critical information.
E.g. the columns "assignCustomerToFactory[i,j]" will print like
"assignC", making it impossible to reconstruct the solution. Although
superfluous to state, but note that in real-life problems indices are
typically strings with descriptive names as well. It would be convenient
to allow for example column names upto 128 chars.

Best,
Joost 

Illustration:

+++++++++++++++++++++++++++++++++++++++++++++++++++
Column names and values of nonzeros in the solution
+++++++++++++++++++++++++++++++++++++++++++++++++++
assignC 	       0 	    1.00000
assignC 	       1 	    1.00000
assignC 	       2 	    1.00000
assignC 	       3 	    1.00000
assignC 	       5 	    1.00000
assignC 	       6 	    1.00000
assignC 	       7 	    1.00000
assignC 	      16 	    1.00000

-----Oorspronkelijk bericht-----
Van: symphony-bounces at list.coin-or.org
[mailto:symphony-bounces at list.coin-or.org] Namens Ted Ralphs
Verzonden: zaterdag 28 januari 2012 17:59
Aan: Paulo Gouveia
CC: symphony at list.coin-or.org
Onderwerp: Re: [Symphony] symphony

The easiest way to get the solution into a text file with the black
box solver would be to redirect the output to a file and then parse it
to extract the solution. I suppose it would not be hard to provide an
option for storing the output in a file if that's needed. As for
changing the number of significant digits displayed in the solution,
the only way to do that at the moment would be to change the source
code to add more digits to the print statements. Of course, all this
would be easy if you just wrote a small driver and called the SYMPHONY
library through the API. Then you could grab the solution yourself and
output in any format and to any location you like. If that's not an
option for you, I can try to add these options.

Cheers,

Ted

On Fri, Nov 18, 2011 at 3:52 PM, Paulo Gouveia <paulossg at gmail.com>
wrote:
> Dear,
>
> I am working with the symphony.exe in Windows and I would like to
export the
> solution to a .txt file.
> Is there how to do this?
> Is there how to increasing the amount of characters of the variables
that
> appear in the solution, because it is truncated.
>
> Best Regards
>
> Paulo Gouveia
> _______________________________________________
> Symphony mailing list
> Symphony at list.coin-or.org
> http://list.coin-or.org/mailman/listinfo/symphony
>



-- 
Dr. Ted Ralphs
Associate Professor, Lehigh University
(610) 628-1280
ted 'at' lehigh 'dot' edu
coral.ie.lehigh.edu/~ted
_______________________________________________
Symphony mailing list
Symphony at list.coin-or.org
http://list.coin-or.org/mailman/listinfo/symphony


From hennebry at web.cs.ndsu.nodak.edu  Fri Feb  3 12:50:39 2012
From: hennebry at web.cs.ndsu.nodak.edu (Michael Hennebry)
Date: Fri, 3 Feb 2012 11:50:39 -0600 (CST)
Subject: [Symphony] solution export and var names
In-Reply-To: <FFE80205F88B490998884211C76CC73B@joost>
References: <FFE80205F88B490998884211C76CC73B@joost>
Message-ID: <alpine.DEB.1.00.1202031145380.16646@mail.cs.ndsu.nodak.edu>

On Fri, 3 Feb 2012, Joost wrote:

> In my experience a 'higher' condition number makes a problem (much)
> harder to solve, and is a signal to check problem data and formulation.
>
> I will look into CglGomory.cpp and try to compute the condition number
> first for the root node.

Note that a Gomory fractional cut is pretty much
guaranteed to increase the condition number.
Starting with an all-integer basis,
one gets another all-integer basis with a smaller determinant.

-- 
Michael   hennebry at web.cs.ndsu.NoDak.edu
"On Monday, I'm gonna have to tell my kindergarten class,
whom I teach not to run with scissors,
that my fiance ran me through with a broadsword."  --  Lily