[Top][All Lists]

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

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

From: Tristan Bereau
Subject: Re: [ESPResSo] bug in random.tcl in mbtools
Date: Thu, 19 Mar 2009 06:46:58 -0400


I'm not sure whether a gaussian random number generator has already been implemented in the TCL shell of ESPResSo, but if not, there is a couple of methods described at the end of the "Old Testament" of molecular simulations (i.e. Allen and Tildesley), and Burkhard's is one of them.



On Mar 19, 2009, at 4:52 AM, Jacob Kirkensgaard wrote:

Hi again

Thanks to Tristan and Burkhard - you are of course right. For my purposes my suggestion is sufficient but I will see if I have time to implement a small routine to do it properly as outlined by Burkhard. Is there a routine for generating Gaussian random numbers? As Tristan mentioned this would be a quick way to solve the issue.


On Mar 19, 2009, at 9:40 AM, address@hidden wrote:


from Tristan's remarks I understand that there
is a question of how to get random numbers
uniformly in or on a sphere? If so, let me tell
you one possible answer:

1. You first draw 3 RNs u_1, u_2, u_3 from [0:1]
2. Then you transform u_1 = 2 * u_1 - 1 etc.,
 resulting in RNs in [-1,1]
3. Calculate r2 = u_1**2 + u_2**2 + u_3**2
4. If r2 > 1 --> throw everything away,
 go to step 1, try again
5. If r2 < 1: Either you are done (supposing
 that you want RN IN the sphere), or you set
 norm = 1 / sqrt(r2)
 u_1 = u_1 * norm etc.
 By this you project the output onto the
 surface of the unit sphere, so you get
 something ON the sphere.

If that was trivial: Please accept my apologies
for bothering you.


reply via email to

[Prev in Thread] Current Thread [Next in Thread]