[Scip] Counters for Priced Variables (Reduced Costs Question)

[Vivek Periaraj]

Hi Gerald, 


Got it clarified. Thanks. 


Vivek. 

----- Original Message ----- 

From: "Vivek Periaraj" <vivek.periaraj at gmail.com> 
To: "Gerald Gamrath" <gamrath at zib.de> 
Cc: scip at zib.de 
Sent: Thursday, March 21, 2013 4:18:08 PM 
Subject: Re: [Scip] Counters for Priced Variables (Reduced Costs Question) 

Hi Gerald, 

Had a question regarding reduced costs. 

In my model, I have >=, = and <= type constraints. 

I tried following options to compute the reduced costs: 

1) Compute rc as c - Ap 
2) Compute rc as c - A*fabs(p) 
3) Use rc from SCIPgetRedcost() 

1 and 3 generates relatively less columns but it never converges. 
2 generated many columns and gave optimal solution fast. 

Regards, 
Vivek. 


----- Original Message ----- 

From: "Vivek Periaraj" <vivek.periaraj at gmail.com> 
To: "Gerald Gamrath" <gamrath at zib.de> 
Cc: scip at zib.de 
Sent: Wednesday, March 20, 2013 11:12:06 AM 
Subject: Re: [Scip] Counters for Priced Variables 

Great. Thanks. 


Vivek. 

----- Original Message ----- 

From: "Gerald Gamrath" <gamrath at zib.de> 
To: "Vivek Periaraj" <vivek.periaraj at gmail.com> 
Cc: scip at zib.de 
Sent: Tuesday, March 19, 2013 3:52:59 AM 
Subject: Re: [Scip] Counters for Priced Variables 

Hi Vivek, 

yes, there is also the variable pricer which looks for variables with 
negative reduced cost that aged out of the LP. See 
http://scip.zib.de/doc/html/FAQ.shtml#Q5.2 

Best, 
Gerald 

Am 16.03.2013 08:16, schrieb Vivek Periaraj: 
> Hello, 
> 
> When there are two consecutive iterations where there are no variables added for pricing, then the counters returned by the following APIs turn out to be incorrect. They indicate some variables have been added. Does SCIP add variables apart from the pricing callback? 
> 
> SCIPgetNPricevarsFound(scip) 
> SCIPgetNPricevarsApplied(scip); 
> 
> Regards, 
> Vivek 
> _______________________________________________ 
> Scip mailing list 
> Scip at zib.de 
> http://listserv.zib.de/mailman/listinfo/scip