Dear members
have a model in which I have to use ceil function in one of my constraints. but when I use DNLP solvers it does not result in optimal solution. When I tested my constraints I figured out that function ceil does not operate correctly. More precisely I have a positive variable pi(k) in the constraint:
Fk(k)… sum(n $(ord(n)=t), z(k,n)) - z(k,k)-ceil(pi(k))+pi(k)=g=1;
but in .lst file I have:
Fk(n0)… (1)*pi(n0) - z(n0,n0) + z(n0,n11) =G= 1 ; (LHS = 0, INFES = 1 ****)
.
.
.
Fk(n10)… (1)*pi(n10) - z(n10,n10) + z(n10,n11) =G= 1 ;
I can’t understand why infeasibility of the constraint is 1. Can anyone help me?
Maria
–
To unsubscribe from this group and stop receiving emails from it, send an email to gamsworld+unsubscribe@googlegroups.com.
To post to this group, send email to gamsworld@googlegroups.com.
Visit this group at http://groups.google.com/group/gamsworld.
For more options, visit https://groups.google.com/groups/opt_out.
Dear Maria
The output you show is from the model generation BEFORE the model is solved. It just means that the point from which you start is infeasible.
However, you will most likely have problems with a model with Ceil. The DNLP solvers are solving the model as an NLP and is ignoring discontinuities. You have a function that jumps and the solver will not be able to predict the jumps. If you want to create a ‘good’ model I think you need an integer variable, piceil(k), that describes your ceil-value in addition to your continuous pi(k) variable. The relationship is then Pi(k) =L= Piceil(k) and Piceil(k) =L= pi(k)+1. The cost is that you must use a MIP or MINLP solver (depending on whether there are other nonlinearities in your model).
Good luck
Arne
\
Arne Stolbjerg Drud
ARKI Consulting & Development A/S
Bagsvaerdvej 246A, DK-2880 Bagsvaerd, Denmark
Phone: (+45) 44 49 03 23, Fax: (+45) 44 49 03 33, email: adrud@arki.dk
From: gamsworld@googlegroups.com [mailto:gamsworld@googlegroups.com] On Behalf Of Maria Afshari Rad
Sent: Thursday, November 21, 2013 12:02 PM
To: gamsworld@googlegroups.com
Subject: Emergency
Dear members
have a model in which I have to use ceil function in one of my constraints. but when I use DNLP solvers it does not result in optimal solution. When I tested my constraints I figured out that function ceil does not operate correctly. More precisely I have a positive variable pi(k) in the constraint:
Fk(k)… sum(n $(ord(n)=t), z(k,n)) - z(k,k)-ceil(pi(k))+pi(k)=g=1;
but in .lst file I have:
Fk(n0)… (1)*pi(n0) - z(n0,n0) + z(n0,n11) =G= 1 ; (LHS = 0, INFES = 1 ****)
.
.
.
Fk(n10)… (1)*pi(n10) - z(n10,n10) + z(n10,n11) =G= 1 ;
I can’t understand why infeasibility of the constraint is 1. Can anyone help me?
Maria
–
To unsubscribe from this group and stop receiving emails from it, send an email to gamsworld+unsubscribe@googlegroups.com.
To post to this group, send email to gamsworld@googlegroups.com.
Visit this group at http://groups.google.com/group/gamsworld.
For more options, visit https://groups.google.com/groups/opt_out.
–
To unsubscribe from this group and stop receiving emails from it, send an email to gamsworld+unsubscribe@googlegroups.com.
To post to this group, send email to gamsworld@googlegroups.com.
Visit this group at http://groups.google.com/group/gamsworld.
For more options, visit https://groups.google.com/groups/opt_out.