[
https://issues.apache.org/jira/browse/MATH-936?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
Thomas Neidhart updated MATH-936:
-
Attachment: MATH-936.patch
We can not ensure that the resulting value of nextLong will really be uniformly
distributed within the given bounds due to limited precision of the calculation
(we would need to do the scaling calculation with a Dfp field for example, but
this would be very slow).
The attached patch at least ensures that the resulting value will be strictly
inside the given bounds.
RandomDataGenerator#nextLong violates bounds
Key: MATH-936
URL: https://issues.apache.org/jira/browse/MATH-936
Project: Commons Math
Issue Type: Bug
Affects Versions: 3.1
Reporter: Ralf Wiebicke
Labels: random
Attachments: MATH-936.patch, RandomGeneratorLongTest.java
I attached a test.
If the underlying RandomGenerator returns 0.0, then nextLong returns
Long.MIN_VALUE, although the lower bound is Long.MIN_VALUE+1.
The javadoc of RandomGenerator#nextDouble does not clearly define, whether
the result includes the lower border of 0.0 or not.
In java.util.Random it clearly defined as included: uniformly from the range
0.0d (inclusive) to 1.0d (exclusive). And the existence of
JDKRandomGenerator suggests, that RandomGenerator should have the same
contract.
I tested with version 3.1.1 from mvnrepository
--
This message is automatically generated by JIRA.
If you think it was sent incorrectly, please contact your JIRA administrators
For more information on JIRA, see: http://www.atlassian.com/software/jira
