RE: [AMPL 12359] Computational cost between variable boundary constraints and inequalities constraints

[Robert Fourer]

This post was already answered in a different thread; you can see the answer at

https://groups.google.com/d/msg/ampl/6Y3DBFe4Dco/HQfNWqddCgAJ

Please add to that thread if you have more questions about this topic.

Bob Fourer
am...@googlegroups.com

=======

From: am...@googlegroups.com [mailto:am...@googlegroups.com] On Behalf Of eku...@gmail.com
Sent: Tuesday, July 26, 2016 6:11 AM
To: AMPL Modeling Language
Subject: [AMPL 12359] Computational cost between variable boundary constraints and inequalities constraints

I set my mod using AMPL. And I found there are some computation cost differences between variable boundary constraints and inequalities constraints when I solved it.

For example,

I defined variables with upper and lower bound like below:

var aij{i in I,j in J} >= -amax, <= amax;
or

I defined variables and inequalities constraints

var aij{i in I,j in J};
s.t. Bound_upper_aij {i in I,k in J}:
aij[i,j]<= amax;
s.t. Bound_lower_aij {i in I,k in J}:
aij[i,j]>= -amax;

When the two mods are solved there are different from a computational elapsed time on CPU. (The solutions are derived as same)
I wonder why it is different and I want to know some logical reasons.

Please, help me to know why there are differences.
Or please, let me know some references to read.

[문재영]

Thank you very much!

I deleted my post to avoid overlapping. But It still remain.

I have no idea how to delete this thread...

Anyway, Thank you very much!

2016년 7월 27일 수요일 오후 3시 42분 41초 UTC+9, Robert Fourer 님의 말: