[
https://issues.apache.org/jira/browse/STATISTICS-36?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17460726#comment-17460726
]
Gilles Sadowski commented on STATISTICS-36:
-------------------------------------------
{quote}
{code}
/** BrentSolver relative accuracy. Currently 1e-14. Change to 1.11e-16. */
private static final double SOLVER_RELATIVE_ACCURACY = 0x1.0p-53;
/** BrentSolver absolute accuracy. Currently 1e-9. */
private static final double SOLVER_ABSOLUTE_ACCURACY = 0;
{code}
{quote}
I seem to recall that the code listing in the book contained a caveat about one
of those value having to be larger than "sqrt(EPS)" where EPS is the
"epsilon-machine".
On the other hand, if your many additional tests do not signal any problem...
> Optimise the default inverse CDF function for continuous distributions
> ----------------------------------------------------------------------
>
> Key: STATISTICS-36
> URL: https://issues.apache.org/jira/browse/STATISTICS-36
> Project: Apache Commons Statistics
> Issue Type: Improvement
> Components: distribution
> Affects Versions: 1.0
> Reporter: Alex Herbert
> Assignee: Alex Herbert
> Priority: Minor
>
> Many distributions only compute the inverse CDF to 1e-9 relative error. Some
> fail to compute when the input p value maps to a very small x value (e.g.
> 1e-50). This is related to the defaults for the brent solver used in the
> default implementation.
> Investigation of tolerances in the brent solver discovered a bug when
> evaluating very small objective functions ([NUMBERS-168]).
> The brent solver requires a release to be used with this fix so work can use
> a local copy until all issues with it have been resolved, and the changes
> ported back to numbers.
>
--
This message was sent by Atlassian Jira
(v8.20.1#820001)
