[
https://issues.apache.org/jira/browse/MATH-1549?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17160726#comment-17160726
]
Gilles Sadowski commented on MATH-1549:
---------------------------------------
Thanks for confirming that the problem stems from floating-point accuracy!
However, among the possible improvements (see MATH-1550), could we perhaps
redesign the representation of the input data (to {{SimplexTableau}} and
{{SimplexSolver}}) so that the scaling can be performed on that, *prior* to the
solving itself? [IMHO, that would prevent making the code even more complex
than it is already.]
> Simplex solver report unbounded solution when solvable constraints scaling up
> -----------------------------------------------------------------------------
>
> Key: MATH-1549
> URL: https://issues.apache.org/jira/browse/MATH-1549
> Project: Commons Math
> Issue Type: Bug
> Affects Versions: 3.6.1
> Environment: macOS catalina
> Reporter: Fan Zhang
> Priority: Major
> Attachments: CommonsMathSimplexSolverTest.java
>
> Time Spent: 0.5h
> Remaining Estimate: 0h
>
> *scenario 1:*
> {code:java}
> min: x + y
> x, y >= 0
> 9000x + y >= 0
> 10000x + y >= 2000
> x >= 2{code}
> *scenario 2:*
> {code:java}
> min: x + y
> x, y >= 0
> 9.0E15x + 1.0E12y >= 0
> 1.1e16x + 1.0e12y >= 2.0e15
> 1.0e12x >= 2e12{code}
> for scenario 1, it works out. but for scenario 2 whose constrains are just
> scaled up by 1e12, the solver reports unbounded solution.
>
--
This message was sent by Atlassian Jira
(v8.3.4#803005)
