yade-users team mailing list archive
Mailing list archive
Re: Uniform distribution
On 13.1.2010 20:25, Bruno Chareyre wrote:
>> It is the way it is supposed to work, and is documented as such.
>> rRelFuzz is radius fuzz ("dispersion" of the uniform distribution, i.e.
>> max-min) relative to rMean, that means it goes .5 above and below. E.g.
>> for rMean=0.1 and rRelFuzz=.5, you have range 0.1±½*.5*0.1.
> Mmmmh... ok.
> I considered makeCloud was a copy of the older generator from
> TriaxialTest, which was clearly not supposed to generate between +/-
> 0.5. You solved the problem the other way, by changing the meaning of
> the variable. For me, +/- deviation is more intuitive than
> +/-0.5dispersion, but it is just a matter of taste.
> Do we revert to 0.5 or do we update the doc?
Change the doc if you feel this is more appropriate, though for me
lexically fuzz==dispersion... (the old name, stDev, was actually not
meaninfgul, it wasn't "standard deviation" at all). It wasn't copy of
Triax code, I wrote it from scratch (but looking at triax code first).
In the future, we should provide way to pass distribution object to
makeCloud (uniform, normal, weibull), there seems to be demand for it