[jira] [Updated] (MATH-1269) FastMath.exp may return NaN for non-NaN arguments

["Thomas Neidhart (JIRA)" <ji...@apache.org>]

     [ https://issues.apache.org/jira/browse/MATH-1269?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ]

Thomas Neidhart updated MATH-1269:
----------------------------------
    Attachment: MATH-1269_fix_z.patch

I propose another patch for this issue.
The polynomial expansion for z = exp(epsilon) - 1.0 is actually returning wrong results in case epsilon = 0.

The patch handles this case and further fixes the calculation of (1+z)(tempA+tempB) in case z is 0 as well (in which case we only have to add tempA and tempB).

This would fix your case as well and other potential cases in the range [0, 709] for which the epsilon is also 0.

> FastMath.exp may return NaN for non-NaN arguments
> -------------------------------------------------
>
>                 Key: MATH-1269
>                 URL: https://issues.apache.org/jira/browse/MATH-1269
>             Project: Commons Math
>          Issue Type: Bug
>    Affects Versions: 4.0
>            Reporter: Otmar Ertl
>         Attachments: MATH-1269.patch, MATH-1269.patch, MATH-1269_fix_z.patch
>
>
> I have observed that FastMath.exp(709.8125) returns NaN. However, the exponential function must never return NaN (if the argument is not NaN). The result must always be non-negative or positive infinity.



--
This message was sent by Atlassian JIRA
(v6.3.4#6332)