Rounding error

[ha]

Hey folks,

Can someone please help me with the next two problems:

I use AIMMS to solve a MIP (Mathematical Interger Program???) and I
see that AIMMS gives some rounding errors (I think because of the
divisions with big denominators but I'm not sure). Is there something
I can do to avoid this?

The goal/ main execution is maximize the variable " G"

And sometimes I see that the best solution (when press in Ctrl+P) is
another value than the variable G itself (when calculate with Excel
or Maple I see that the value of G is correct and the value of best
solution is incorrect). Do you know how can this be?

Thank you very much. With kind regards,

Ha

[Marcel Roelofs]

Subject to solver-specific option settings, MIP solvers may accept and
return solutions that lie within a certain tolerance of a true integer
solution to satisfy feasibility. This can especially be a problem if
your problem is scaled badly. Your message seems to suggest that there
may be a problem with the scaling of your model.

The best solution shown in the progress window is the best integer
solution found so far. At the end of the solve I would expect that its
value be equal to the actual objective value. You're saying that the
real objective value and the best solution reported in the progress
windows differ?

Cheers,
Marcel

[Ha Nguyen]

Thank you very much for your answer :)

yes, thereal objective value and the best solution reported in the progress
windows differ with some parameterset, by the most parameter set they're identical
:(

2011/3/23 Marcel Roelofs <marcel....@gmail.com>

--
You received this message because you are subscribed to the Google Groups "AIMMS - The Modeling System" group.
To post to this group, send email to ai...@googlegroups.com.
To unsubscribe from this group, send email to aimms+un...@googlegroups.com.
For more options, visit this group at http://groups.google.com/group/aimms?hl=en.

[Marcel Hunting]

Hi,

The best solution reported in the progress window should be equal to
the value of the objective variable. So this might be a bug. Which
solver are you using? Would it be possible for you to send your
project to our support e-mail address such that we can investigate
this issue?

Marcel Hunting
AIMMS Software Developer

On Mar 24, 1:27 am, Ha Nguyen <hanguyen0...@gmail.com> wrote:
> Thank you very much for your answer :)
>
> yes, thereal objective value and the best solution reported in the progress
> windows differ with some parameterset, by the most parameter set they're
> identical
> :(
>

> 2011/3/23 Marcel Roelofs <marcel.roel...@gmail.com>

> >http://groups.google.com/group/aimms?hl=en.- Hide quoted text -
>
> - Show quoted text -

[Ha Nguyen]

Wow, that would be very nice if you can take a look at it.
Our goal was to maximize the objective variable "gem_groen".
the best solution reported in the progress window differs a little bit from
the value of the objective variable, we dunno if it is correct.
In the attachment you'll find the project "simpel_model". Furthermore I also include the data set where the values differ a bit.

Thank you very much ^.^
Ha

2011/3/24 Marcel Hunting <Marcel....@aimms.com>

[Ha Nguyen]

Excuses me,

The attachment was not correct.
Here it is!!!

[Marcel Hunting]

Thanks for the project.

I was able to reproduce the behavior in AIMMS 3.10 using CPLEX 12.2 as
the MIP solver. After the MIP solve, AIMMS by default lets CPLEX do a
postsolve step in which the integer variables are fixed to their
nearest integer value; the resulting LP is solved to obtain a real
integer solution. (See the description of the CPLEX option ‘Postsolve
Integer Variables’ in the help for more info.)

The info shown in the progress window for the postsolved model was not
correct. It showed:

Best Solution : 0.4461099213 (Post: 0.4461099213)

while it should have shown

Best Solution : 0.4461099213 (Post: 0.4405944443)

This 0.4405944443 was the value passed to AIMMS and assigned to
objective variable ‘gem_groen’.

This issue will be fixed in a next AIMMS release.

For the LP solved during the postsolve step, CPLEX seems to return a
slightly sub-optimal solution. After setting the CPLEX 12.2 option
‘optimality’ to 1e-9 CPLEX returned a solution with objective value
0.4461099213 for the postsolve LP; that solution then was also passed
to AIMMS.

Marcel Hunting
AIMMS Software Developer

On Mar 24, 4:39 pm, Ha Nguyen <hanguyen0...@gmail.com> wrote:
> Excuses me,
>
>  [star].rar
> 242KViewDownload

>
> The attachment was not correct.
> Here it is!!!
>
>
>
> > Wow, that would be very nice if you can take a look at it.
> > Our goal was to maximize the objective variable "gem_groen".
> > the best solution reported in the progress window differs a little bit from
> > the value of the objective variable, we dunno if it is correct.
> > In the attachment you'll find the project "simpel_model". Furthermore I
> > also include the data set where the values differ a bit.
>
> > Thank you very much ^.^
> > Ha
>

> > 2011/3/24 Marcel Hunting <Marcel.Hunt...@aimms.com>

> >> > >http://groups.google.com/group/aimms?hl=en.-Hide quoted text -

[Ha Nguyen]

Thank you very musch
It's clear for us now :))

Kind regards,
Ha

2011/3/24 Marcel Hunting <Marcel....@aimms.com>

Thanks for the project.