[Scip] How can SCIP solve division?

[Ambros Gleixner]

And here is the attachement.


Am 21.04.2014 13:54, schrieb Ambros Gleixner:
> Dear Wei,
>
> SCIP can handle powers of negative exponents, but the current Zimpl
> version 3.3.2 does not support this.  It will be contained in the next
> Zimpl release.
>
> In order to work around this, I attached you a Zimpl file with three
> possible reformulations of x^-p that work in Zimpl.  The resulting
> models, however, have additional artificial variables and probably
> perform worse in SCIP.  For a proof-of-concept model it may work, though.
>
> We will announce as soon as the new Zimpl release is available.
>
> Kind regards,
>
> ambros
>
>
> Am 21.04.2014 09:26, schrieb weikaimas:
>> Hi,
>>      I want to get the value of sum(1/x[i]), which x[i] are variables,
>> using SCIP. I have used the format ZIMPL to build my model, however, it
>> can not solve sum(1/x[i]). Could you give me some suggestions about the
>> my problem.
>>    Thanks.
>> 2014-04-21
>> ------------------------------------------------------------------------
>>    Wei kai
>> School of Economics & Management
>> Tongji University, Shanghai 200092, China
>>
>>
>> _______________________________________________
>> Scip mailing list
>> Scip at zib.de
>> http://listserv.zib.de/mailman/listinfo/scip
>>
>

-- 
____________________________________________________________
Ambros M. Gleixner
Zuse Institute Berlin - Matheon - Berlin Mathematical School
http://www.zib.de/gleixner
-------------- next part --------------
param n := 10;
param p := 3;
set N := { 1 .. n };

var x[N] >= 1;
var y[N];

minimize obj: x[1];

# version 1
subto inv1: forall <i> in N: y[i] * x[i]^p == 1;
subto cons1: sum <i> in N: y[i] == n;

# version 2
subto inv2: forall <i> in N: y[i] == exp(-p * ln(x[i]));
subto cons2: sum <i> in N: y[i] == n;

# version 3
subto cons3: sum <i> in N: exp(-p * ln(x[i])) == n;