Re: [ESPResSo] bug in random.tcl in mbtools

[Tristan Bereau]

Dear Jacob,

Thanks for pointing this out. However, please keep in mind that the  
distribution of angles is still not homogeneous: this is effectively  
mapping a cube onto a sphere, and there will be an overemphasis along  
the corners of this cube. If you're interested in a more homogeneous  
distribution, a gaussian distribution may be more suitable.

I've still updated the CVS with your corrected code.

Best,

Tristan



On Mar 17, 2009, at 10:47 AM, Jacob Kirkensgaard wrote:

> Hi all mbtools users
>
> In the file random.tcl in the mbtools package the following lines are
> found
>
>         # Now choose a random orientation vector.  Actually it's not
>         # strictly random if we do it like this (fix)
>         lappend orient [expr [t_random]]
>         lappend orient [expr [t_random]]
>         lappend orient [expr [t_random]]
>
>
> I think replacing with these lines should fix it:
>
>         lappend orient [expr 2*[t_random]-1]
>         lappend orient [expr 2*[t_random]-1]
>         lappend orient [expr 2*[t_random]-1]
>
> This means we are now sampling all directions in space instead of only
> the cone consisting of vectors with positive coordinates.
>
> Best,
> Jacob Kirkensgaard
> Copenhagen University
>
>
>
> _______________________________________________
> ESPResSo mailing list
> address@hidden
> https://fias.uni-frankfurt.de/mailman/listinfo/espresso