Re: [ojAlgo-user] QuadraticSolver issue

[Simon G. <si...@ya...>]

Yes, I have to finish the lower bounds, that part I'm not worried about.
But I don't understand, I must be doing something different from how you did it, 
because I get:

X class org.ojalgo.matrix.PrimitiveMatrix
{{1.000000000000013},
{3.2333333333333143},
{-3.233333333333325}} 
This is how I get X:

Result rslt = qdrtc_slvr.solve();
BasicMatrix rslt_mtrx = rslt.getSolution();
PhysicalStore<Double> result = rslt_mtrx.toPrimitiveStore();
System.out.println("\nX " + rslt_mtrx.toString());

I this correct?
Thanks again for all your help by the way!

SG



________________________________
From: Anders Peterson <an...@op...>
To: oja...@li...
Sent: Tue, June 14, 2011 4:34:59 PM
Subject: Re: [ojAlgo-user] QuadraticSolver issue

Here's what I get:

[AE] = class org.ojalgo.matrix.PrimitiveMatrix
{{1.0,    1.0,    1.0}}

[BE] = class org.ojalgo.matrix.PrimitiveMatrix
{{1.0}}

[Q] = class org.ojalgo.matrix.PrimitiveMatrix
{{4.0,    1.0,    2.0},
{1.0,    9.0,    1.0},
{2.0,    1.0,    16.0}}

[C] = class org.ojalgo.matrix.PrimitiveMatrix
{{10.0},
{15.0},
{18.0}}

[AI] = class org.ojalgo.matrix.PrimitiveMatrix
{{1.0,    0.0,    0.0},
{0.0,    1.0,    0.0},
{0.0,    0.0,    1.0}}

[BI] = class org.ojalgo.matrix.PrimitiveMatrix
{{1.0},
{1.0},
{1.0}}

[X] = class org.ojalgo.matrix.PrimitiveMatrix
{{-0.17441860465116266},
{0.6279069767441859},
{0.5465116279069767}}

[LE] = class org.ojalgo.matrix.PrimitiveMatrix
{{8.976744186046512}}

[LI] = class org.ojalgo.matrix.PrimitiveMatrix
{{0.0},
{0.0},
{0.0}}

[SE] = class org.ojalgo.matrix.PrimitiveMatrix
{{0.0}}

[SI] = class org.ojalgo.matrix.PrimitiveMatrix
{{1.1744186046511627},
{0.37209302325581406},
{0.4534883720930233}}

I do not get any weights larger than 1.0, but I do get one weight smaller than 
0.0. There's no constraint to restrict that! Add constraints for that and you 
should get the solutions you expect.

Here's how I would find the efficient frontier:

        final BasicMatrix tmpCovars = ...;
        final BasicMatrix tmpReturs = ...;

        final MarketEquilibrium tmpME = new MarketEquilibrium(tmpCovars);

        final MarkowitzModel tmpMarkowitz = new MarkowitzModel(tmpME, 
tmpReturs);

        for (int i = 1; i < 10; i++) { // Some different risk aversion factors

            tmpMarkowitz.setRiskAversion(new BigDecimal(i));

            final List<BigDecimal> tmpWeights = tmpMarkowitz.getWeights();
        }