OK, so I finally have to ask. What the hell *is* a chi-squared test? And 
why does POV-Ray need one?

Post a reply to this message

Le 26/02/2010 17:37, Invisible nous fit lire :
> OK, so I finally have to ask. What the hell *is* a chi-squared test? And
> why does POV-Ray need one?

chi-squared is mentionned in povray documentation about the media
sampling, and it's an heuristic to avoid taking far too much expansive
samples when the chi-squared test would indicate that "statically",
there is no point getting more samples.

Post a reply to this message

And lo On Fri, 26 Feb 2010 16:37:13 -0000, Invisible <voi### [at] devnull> did  
spake thusly:

> OK, so I finally have to ask. What the hell *is* a chi-squared test? And  
> why does POV-Ray need one?

It's a confidence test so as Le_Forgeron stated it's used to keep the  
processing down i.e. should I bother with this sample? Chi-squared returns  
that there's only a 1% chance there's something significant there, so no I  
won't bother.

-- 
Phil Cook

--
I once tried to be apathetic, but I just couldn't be bothered
http://flipc.blogspot.com

Post a reply to this message

Phil Cook v2 wrote:
> And lo On Fri, 26 Feb 2010 16:37:13 -0000, Invisible <voi### [at] devnull> did 
> spake thusly:
> 
>> OK, so I finally have to ask. What the hell *is* a chi-squared test? 
>> And why does POV-Ray need one?
> 
> It's a confidence test so as Le_Forgeron stated it's used to keep the 
> processing down i.e. should I bother with this sample? Chi-squared 
> returns that there's only a 1% chance there's something significant 
> there, so no I won't bother.

OK, so how do you determine that an arbitrarily discontinuous function 
has been sampled to within an acceptable level of confidence? What's the 
algorithm here?

Post a reply to this message