particletemplate gaussian radius distribution

Submitted by ahobbs on Thu, 12/06/2012 - 14:33

Another beginner question: I'd like a normal distribution of particle sizes with a mean of 0.000925m and a standard deviation of 0.1.

I've tried this:
fix pts1 all particletemplate/sphere 1 atom_type 1 density constant 2900 radius gaussian number 0.009525 0.1
fix pdd1 all particledistribution/discrete 1. 1 pts1 1.0

but I get this:
ERROR: Probablity distribution: mu-3*sigma < 0, please increase mu or decrease sigma (probability_distribution.h:169)

Do I need to put sigma = (0.009525*0.1)?

Forums:

ckloss's picture

ckloss | Thu, 12/06/2012 - 15:34

Hi Andrew,

"gaussian" has to be cut-off because particles can not have negative diameters. In order to ensure that the cut-off distribution still resembles a Gaussian, LIGGGHTS requires mu-3*sigma >= 0

In your case, mu = 0.009525; sigma = 0.1 would not work for this reason

Christoph

deepakpawar.2310 | Thu, 12/14/2017 - 06:19

Any comment on the follwoing error , i am using the following command for particle distribution

# Particle distributions
fix pts1 all particletemplate/sphere 15485863 atom_type 1 density constant 2500 radius gaussian number 0.001 0.0001 # mu (mean) and sigma (standard devation)
fix pdd1 all particledistribution/discrete 15485867 1 pts1 1.0

ERROR: Fix particletemplate/sphere (id pts1): invalid radius random style (../fix_template_sphere.cpp:199)
(../fix_template_sphere.cpp:199)

Daniel Queteschiner | Thu, 12/14/2017 - 09:29

The gaussian option has been removed in LIGGGHTS 3.7.0 (as well as uniform and lognormal). Thus the only valid style is constant.

msandli | Fri, 01/12/2018 - 08:05

*I know the default answer is to make the code myself, but I'm not nearly good enough at c++ to make it myself, so I figure I'll ask anyways*

I seem to recall that the gaussian and the lognormal distributions were removed due to the issues with generating negative particles. Is there an easy way to hard code a minimum particle radius? I wouldn't think this would be an issue with the uniform distribution (except in the case of typos).