Mailing-List: contact issues-help@commons.apache.org; run by ezmlm
Precedence: bulk
Reply-To: issues@commons.apache.org
Message-ID: <1409356097.1250089574772.JavaMail.jira@brutus>
Date: Wed, 12 Aug 2009 08:06:14 -0700 (PDT)
From: "Michael Nischt (JIRA)" <jira@apache.org>
To: issues@commons.apache.org
Subject: [jira] Updated: (MATH-283) MultiDirectional optimzation loops
 forver if started at the correct solution
In-Reply-To: <28815735.1250088734790.JavaMail.jira@brutus>
MIME-Version: 1.0
Content-Type: text/plain; charset=utf-8
Content-Transfer-Encoding: 7bit


     [ https://issues.apache.org/jira/browse/MATH-283?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ]

Michael Nischt updated MATH-283:
--------------------------------

    Description: 
MultiDirectional.iterateSimplex loops forever if the starting point is the correct solution.

see the attached test case (testMultiDirectionalCorrectStart) as an example.

  was:
MultiDirectional.iterateSimplex loops forever if the starting point is the correct solution.

see the following test case (testMultiDirectionalCorrectStart) as an example

{quote}
public class MultiDirectionalCorrectStartTest
{
    @Test
    public void testMultiDirectionalIncorrectStart()
        throws FunctionEvaluationException, OptimizationException
    {
        // works fine
        try
        {
            testMultiDirectionalWithStart(0.1, 0.1);
        }
        catch(Exception e) { fail(e.getMessage()); }
    }

    @Test
    public void testMultiDirectionalCorrectStart()
        throws FunctionEvaluationException, OptimizationException
    {
        // fails because MultiDirectional.iterateSimplex is looping forever
        // the while(true) should be replaced with a convergence check
        try
        {
            testMultiDirectionalWithStart(0.0, 0.0);
        }
        catch(Exception e) { fail(e.getMessage()); }

    }


    private void testMultiDirectionalWithStart(double xStart, double yStart)
        throws FunctionEvaluationException, OptimizationException
    {
        MultiDirectional multiDirectional = new MultiDirectional();
        multiDirectional.setMaxIterations(100);
        multiDirectional.setMaxEvaluations(1000);

        final Gaussian2D function = new Gaussian2D();

        RealPointValuePair estimate = multiDirectional.optimize(function,
                                      GoalType.MAXIMIZE, new double[] {xStart, yStart} );

        final double EPSILON = 1e-5;

        final double expectedMaximum = function.getMaximum();
        final double actualMaximum = estimate.getValue();
        assertEquals(expectedMaximum, actualMaximum, EPSILON);

        final double[] expectedPosition = function.getMaximumPosition();
        final double[] actualPosition = estimate.getPoint();
        assertEquals(expectedPosition[0], actualPosition[0], EPSILON );
        assertEquals(expectedPosition[1], actualPosition[1], EPSILON );
        
    }

    static class Gaussian2D implements MultivariateRealFunction
    {
        private static final double[] maximumPosition = { 0, 0 };

        private final double std = 1;
       
        public double getMaximum()
        {
            return value(maximumPosition);
        }

        public double[] getMaximumPosition()
        {
            return maximumPosition.clone();
        }

        public double value(double[] point)
        {
            final double x = point[0], y = point[1];

            return 1/(2*Math.PI*std*std)*Math.exp(-(x*x + y*y) / (2*std*std));
        }
    }
}
{quote}


> MultiDirectional optimzation loops forver if started at the correct solution
> ----------------------------------------------------------------------------
>
>                 Key: MATH-283
>                 URL: https://issues.apache.org/jira/browse/MATH-283
>             Project: Commons Math
>          Issue Type: Bug
>    Affects Versions: 2.0
>         Environment: none specific for the issue (it's a programming bug)
>            Reporter: Michael Nischt
>            Priority: Blocker
>         Attachments: MultiDirectionalCorrectStartTest.java
>
>   Original Estimate: 24h
>  Remaining Estimate: 24h
>
> MultiDirectional.iterateSimplex loops forever if the starting point is the correct solution.
> see the attached test case (testMultiDirectionalCorrectStart) as an example.

-- 
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.