Re: [ojAlgo-user] Possible bug in ExpressionBasedModel: setMinimisation/setMaximisation

[Anders P. <an...@op...>]

I haven't tried your code (yet) but I see something strange:

>  tmpModel.setMinimisation(false);
>  tmpModel.setMaximisation(true);
>  Result result = tmpModel.minimise();      


Essentially you should just do either

Result result = tmpModel.minimise();

or

Result result = tmpModel.maximise();


If you look inside those methods they do:

{
this.setMinimisation(true);
final Result tmpSolverResult = this.solve();
return this.handleResult(tmpSolverResult);
}

and

{
this.setMaximisation(true);
final Result tmpSolverResult = this.solve();
return this.handleResult(tmpSolverResult);
}


Is this something that would explain/solve your problem?


/Anders




> On 20 nov 2014, at 13:18, Nico Potyka <Nic...@gm...> wrote:
> 
> Hi,
>  
> I use the newest version of ojAlgo (v37.0) and just tried the ExpressionBasedModel for linear optimization.
> The solver seems to minimize independently of the configuration via setMinimisation/setMaximisation.
> Here's a simple example:
>  
> 
>         int n = 4;
>         Variable[] tmpVariables = new Variable[n];
>         for(int i=0; i<n; i++) {
>             tmpVariables[i] = new Variable("x"+i).lower(BigMath.ZERO);
>         }
>         
>         
>         ExpressionsBasedModel tmpModel = new ExpressionsBasedModel(tmpVariables);
>         
>         
>         Expression tmpExpr = tmpModel.addExpression("Objective");
>         tmpExpr.setLinearFactor(0, 10);    
>         tmpExpr.setLinearFactor(1, 1);    
>         tmpExpr.setLinearFactor(2, 1);        
>         tmpExpr.setLinearFactor(3, 1);
>         tmpExpr.weight(BigMath.ONE);
>         
>         
>         tmpExpr = tmpModel.addExpression("Normalization").level(BigMath.ONE);
>         for(int i=0; i<n; i++) {
>             tmpExpr.setLinearFactor(i, BigMath.ONE);
>         }
>         
>         tmpModel.setMinimisation(false);
>         tmpModel.setMaximisation(true);
>         Result result = tmpModel.minimise();
>         System.out.println("optimal value: "+result.getValue());
>         for(int i=0; i<n; i++) {
>             System.out.println("x"+i+": "+result.get(i));
>         }
>  
>  
>  
> We maximize 10*x0 + x1 + x2 +x3 subject to non-negativity and normalization to 1. The output is
>  
>    optimal value: 1.0
>    x0: 0
>    x1: 1.00000000000000
>    x2: 0
>    x3: 0
>  
>  
> but in fact the optimal value is 10 and x0 should be 1 instead of x1. It seems that the solver always minimizes
> the function. Do I use the solver in a wrong way or is there a bug?
>  
> Best,
> Nico
> ------------------------------------------------------------------------------
> Download BIRT iHub F-Type - The Free Enterprise-Grade BIRT Server
> from Actuate! Instantly Supercharge Your Business Reports and Dashboards
> with Interactivity, Sharing, Native Excel Exports, App Integration & more
> Get technology previously reserved for billion-dollar corporations, FREE
> http://pubads.g.doubleclick.net/gampad/clk?id=157005751&iu=/4140/ostg.clktrk_______________________________________________
> ojAlgo-user mailing list
> ojA...@li...
> https://lists.sourceforge.net/lists/listinfo/ojalgo-user