[
https://issues.apache.org/jira/browse/MATH-1422?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16061738#comment-16061738
]
Gilles commented on MATH-1422:
------------------------------
It was certainly the other main goal.
See the Javadoc of the {{FastMath}} class which mentioning a best effort to
reach a 0.5 ulp precision, and the unit tests that validate the claim.
I've suggested a rename to {{AccurateMath}}, since the only benchmarks we have
do not seem to support the "Fast" qualifier...
However I indeed do not know for sure whether there is still an advantage to
using {{FastMath}}: see also
* MATH-1114
> How accurate is FastMath javadoc?
> ---------------------------------
>
> Key: MATH-1422
> URL: https://issues.apache.org/jira/browse/MATH-1422
> Project: Commons Math
> Issue Type: Bug
> Reporter: Roman Leventov
>
> Javadoc to FastMath states that it is a "faster, more accurate alternative to
> java.lang.Math" since at least 2012:
> https://github.com/apache/commons-math/blob/696be68b5dc8872a818a24d8a377158314b54e31/src/main/java/org/apache/commons/math3/util/FastMath.java
> How accurate is this assertion in 2017, Java 8?
--
This message was sent by Atlassian JIRA
(v6.4.14#64029)