mahout-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Jeff Eastman <j...@windwardsolutions.com>
Subject Re: Dirichlet
Date Wed, 04 Mar 2009 03:17:18 GMT
The problem with using Blog is that it is meant to be a stand-alone 
simulation environment and we just need two (one really) of its 
multitude of sampling classes. I also have not been able to compile it 
into a jar file, nor have I been able to find one. It does not seem to 
build on my Mac. Uncommons math already has Gaussian; has some good 
random number generators; and I just need to figure out how to sample 
from a Beta distribution with it.  The implementation seems simple 
enough: Beta(a, b) = Gam(1, a)/(Gam(1, a)*Gam(1, b)) - but I don't fully 
understand what I'm doing, yet.

Another alternative suggested by Ted is based upon commons math, but it 
requires some of Ted's personal extensions to support Beta.

Jeff




Grant Ingersoll wrote:
> Are they in the Maven repo?  If they are, then we can just add the 
> dependency that way.
> On Mar 3, 2009, at 6:42 PM, Jeff Eastman wrote:
>
>> Hi Ted,
>>
>> I'm still trying to resolve two external dependencies on the Blog 
>> package to get moving again on DP Clustering:
>>
>> import blog.distrib.Beta;
>> import blog.distrib.UnivarGaussian;
>>
>> I notice we already have uncommons math in trunk and so rNorm should 
>> be trivial. Can you suggest how to implement rBeta with it? From the 
>> API, it is not obvious to me.
>>
>> Jeff
>
>
>
>


Mime
View raw message