Em Sex, 2005-08-12 às 10:21 +0200, Michal Kaukic escreveu:
> ---Reply to mail from Afonso Celso Medina about Solver with Constraints 
> failure
> 
> > Hi!
> > 
> > When I changed my 1.4.2 gnumeric (linux) to the 1.5.2 version(win32), I
> > have noticed that the Solver tool didn't work never more. Using the
> > Program report, I could discover that it is not a problem in Target
> > Cell, but there is a problem in Constraints parsing. Now, all my
> > correct old models runs to Unbounded Solution.
> > 
> 
> Hi, Afonso,
> 
>   could You please, supply a small representative example, 
> of LP program, where your problem is demonstrated?
> 
> 
>   I can confirm, that in following problem:
> 
> -------------------------------------------
> Maximize:
>             x1 + x2
> 
> Subject to:
> 
>              2 x1+ x2   <= 2
>               x1 + 2 x2 <= 2
> 
> (x1, x2 >=0)
> -------------------------------------------
>  
> the constraints are parsed incorrectly as:
> 
>             -x1 + x2  <= 0
>           -2 x1 +2 x2 <= 0
> 
Same to me... Always negatives numbers in first column and changes in
RHS column.

>   Solver seems to be broken in 1.5.2.
> 
>   I suggested patches for working with 
> affine expressions of type c0 + sum(i=1..n, c_i * x_i),
> they worked in 1.4.3; hope someone of Gnumeric developers
> could quickly identify the cause of solver failures in 1.5 series, 
> altough I am not sure the problem is connected 
> with my suggestions. 
>   
>   I will try to compile myself the sources of 1.5.2 and see,
> if I can do something with the mentioned problem.
> 
>                                Mike
> 
>  
> ------------------------------------------------------------------------
> Michal Kaukic                                  ([EMAIL PROTECTED])         
> Dept. of Math. Methods, Fac. of Manag. & Informatics
> University of Zilina,   Slovak Republic
> 
> 

_______________________________________________
gnumeric-list mailing list
[email protected]
http://mail.gnome.org/mailman/listinfo/gnumeric-list

Reply via email to