Re: Deuterium Diffusion Query

[Tom Banks <tbanks@socrates.berkeley.edu>]

> I would
> - take a high statistics probability function calculated by
>    Brendan depending  on the variable r(range) and t(time of decay),
> - try to smooth this function with some root spline
> - and use the resulting 2d histogram for directly sampling the
>    probablity with the root arbitrary random function generator.
>
> Bernhard has implemented the diffusion in his MC,
> how did he do that?

Actually, Bernhard has *not* implemented mu-d diffusion in his MC yet.  He
has only implemented the root-mean-square approximation of thermal mu-p
radial diffusion: r_rms = sqrt(6Dt), where D is the diffusion constant.
However, I believe his intention is to one day implement mu-d diffusion
exactly as you describe: randomly generate a decay time, and then sample
from the appropriate diffusion distribution as provided by Brendan in ~ 1
us intervals.

That's probably the most accurate approach, but I was hoping that it might
be reasonable to model mu-d diffusion as some sort of "faster" version of
thermal mu-p diffusion.  In other words, if you had H2 gas w/3 ppm
deuterium, you would usually sample from the mu-p thermal diffusion
distribution, but once every 3.3E5 times you would sample from the mu-d
diffusion distribution.  That would certainly be simpler, but it might be
so far from reality as to be useless.

I'll talk to Bernhard about the functions he got from Brendan and how to
use them.  I'll contact you, Brendan, if I have any questions about them
that Bernhard can't answer.

Thanks,
Tom