commons-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Phil Steitz <phil.ste...@gmail.com>
Subject [math] Fixing and then deprecating isSupportXxxInclusive in RealDistribution interface
Date Mon, 03 Sep 2012 15:01:26 GMT
The conclusion from [1] was never implemented and I was, once again,
confused by these properties when developing some density
integration tests last night.  I would like to deprecate these
properties from the RealDistribution interface, but since removal
will have to wait until 4.0, I would like to agree on a precise
definition and fix the code to match it in the mean time.

The definition that I propose is that isSupportXxxInclusive means
that when the density function is applied to the upper or lower
bound of support returned by getSupportXxxBound, a finite (i.e. not
infinite), not NaN value is returned.

The code is now consistent with this definition other than for the F
distribution, where the lower bound of support is marked included
but the density returns NaN there.

If others are in agreement, I will make the deprecations, clarify
the javadoc, change the value returned by FDistribution to match the
contract and add tests.

Phil

[1] http://markmail.org/message/dxuxh7eybl7xejde

---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@commons.apache.org
For additional commands, e-mail: dev-help@commons.apache.org


Mime
View raw message