commons-issues mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Benjamin McCann (JIRA)" <j...@apache.org>
Subject [jira] Updated: (MATH-293) Matrix's "OutOfBoundException" in SimplexSolver
Date Thu, 10 Sep 2009 07:29:57 GMT

     [ https://issues.apache.org/jira/browse/MATH-293?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]

Benjamin McCann updated MATH-293:
---------------------------------

    Attachment: SimplexSolverTest.patch
                SimplexTableau.patch

Thanks for the bug report Andrea.  I was expecting this problem when I submitted the patch
for MATH-286, but didn't have a good test case to work against and verify the fix.  This is
largely what I was thinking of when I mentioned we should test some more, but didn't think
you'd beat me to it in finding a good example to work off of :o)  When I changed the columns
being dropped in the patch for MATH-286 it meant there wasn't always a way to always tell
which variable each column of the tableau represented.  This patch creates a mapping between
column index and variable label so that when we can read the final solution out of the tableau.
 I feel much better about this now.  Thanks!

> Matrix's "OutOfBoundException" in SimplexSolver
> -----------------------------------------------
>
>                 Key: MATH-293
>                 URL: https://issues.apache.org/jira/browse/MATH-293
>             Project: Commons Math
>          Issue Type: Bug
>    Affects Versions: Nightly Builds
>         Environment: java 1.6 on Windows XP 32-Bit
>            Reporter: Andrea
>         Attachments: SimplexSolverTest.patch, SimplexTableau.patch
>
>
> Hi all,
> This bug is somehow related to incident MATH-286, but not necessarily...
> Let's say I have an LP and I solve it using SimplexSolver. Then I create a second LP
similar to the first one, but with "stronger" constraints. The second LP has the following
properties:
> * the only point in the feasible region for the second LP is the solution returned for
the first LP
> * the solution returned for the first LP is also the (only possible) solution to the
second LP
> This shows the problem:
> {code:borderStyle=solid}
> LinearObjectiveFunction f = new LinearObjectiveFunction(new double[] { 0.8, 0.2, 0.7,
0.3, 0.4, 0.6}, 0 );
> Collection<LinearConstraint> constraints = new ArrayList<LinearConstraint>();
> constraints.add(new LinearConstraint(new double[] { 1, 0, 1, 0, 1, 0 }, Relationship.EQ,
30.0));
> constraints.add(new LinearConstraint(new double[] { 0, 1, 0, 1, 0, 1 }, Relationship.EQ,
30.0));
> constraints.add(new LinearConstraint(new double[] { 0.8, 0.2, 0.0, 0.0, 0.0, 0.0 }, Relationship.GEQ,
10.0));
> constraints.add(new LinearConstraint(new double[] { 0.0, 0.0, 0.7, 0.3, 0.0, 0.0 }, Relationship.GEQ,
10.0));
> constraints.add(new LinearConstraint(new double[] { 0.0, 0.0, 0.0, 0.0, 0.4, 0.6 }, Relationship.GEQ,
10.0));
> RealPointValuePair solution = new SimplexSolver().optimize(f, constraints, GoalType.MAXIMIZE,
true);
> double valA = 0.8 * solution.getPoint()[0] + 0.2 * solution.getPoint()[1];
> double valB = 0.7 * solution.getPoint()[2] + 0.3 * solution.getPoint()[3];
> double valC = 0.4 * solution.getPoint()[4] + 0.6 * solution.getPoint()[5];
> f = new LinearObjectiveFunction(new double[] { 0.8, 0.2, 0.7, 0.3, 0.4, 0.6}, 0 );
> constraints = new ArrayList<LinearConstraint>();
> constraints.add(new LinearConstraint(new double[] { 1, 0, 1, 0, 1, 0 }, Relationship.EQ,
30.0));
> constraints.add(new LinearConstraint(new double[] { 0, 1, 0, 1, 0, 1 }, Relationship.EQ,
30.0));
> constraints.add(new LinearConstraint(new double[] { 0.8, 0.2, 0.0, 0.0, 0.0, 0.0 }, Relationship.GEQ,
valA));
> constraints.add(new LinearConstraint(new double[] { 0.0, 0.0, 0.7, 0.3, 0.0, 0.0 }, Relationship.GEQ,
valB));
> constraints.add(new LinearConstraint(new double[] { 0.0, 0.0, 0.0, 0.0, 0.4, 0.6 }, Relationship.GEQ,
valC));
> solution = new SimplexSolver().optimize(f, constraints, GoalType.MAXIMIZE, true);
> {code} 
> Instead of returning the solution, SimplexSolver throws an Exception:
> {noformat} Exception in thread "main" org.apache.commons.math.linear.MatrixIndexException:
no entry at indices (0, 7) in a 6x7 matrix
> 	at org.apache.commons.math.linear.Array2DRowRealMatrix.getEntry(Array2DRowRealMatrix.java:356)
> 	at org.apache.commons.math.optimization.linear.SimplexTableau.getEntry(SimplexTableau.java:408)
> 	at org.apache.commons.math.optimization.linear.SimplexTableau.getBasicRow(SimplexTableau.java:258)
> 	at org.apache.commons.math.optimization.linear.SimplexTableau.getSolution(SimplexTableau.java:336)
> 	at org.apache.commons.math.optimization.linear.SimplexSolver.doOptimize(SimplexSolver.java:182)
> 	at org.apache.commons.math.optimization.linear.AbstractLinearOptimizer.optimize(AbstractLinearOptimizer.java:106){noformat}

> I was too optimistic with the bug MATH-286 ;-)

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


Mime
View raw message