Generation of undesired small grains

[Ava-22-neper]

Hi @rquey

I am trying to generate a distribution like the attached file in which the blue and orange graphs show lognormal and beta distributions. Using both types of distributions, I got a large number of very small grains with the order of 1e-5 mm and the max grain size was around 0.002 mm which is far smaller than my peak (0.01 mm). I also tried using "from=" in my command according to Q&A 950. For the beta distribution I got error and for the lognormal distribution to obvious change was found. Here is my parameters for both fits and my last command for lognormal distribution including "from". I really appreciate if you can kindly provide me with some solutions for this, please.

Beta:
a = 1.97 b = 90.30

Lognormal:
mu = -4.11 sigma = 0.77

My command for Lognormal

neper -T -n 2261 -reg 1 -dim 2 -domain "square(1,1)" -morpho "diameq:lognormal(-4.11,0.77,from=0.006),sphericity:lognormal(0.85,0.01)" -statcell "id,x,y,z,diameq,area" -morphooptistop itermax=10000 -o Lo5_2D

[rquey]

lognormal(-4.11,0.77,from=0.006): you are supposed to provide the average value of diameq (not its log). The "standard deviation" should be the standard deviation of the diameqs.

[Ava-22-neper]

Sorry can you please expand this a bit more?
I have the diameters and the number of grains of each diameter. In this case how I can get the average value and standard deviation of the diameters for a lognormal distribution for instance? Here, I generated a vector from the information that I mentioned and fitted a lognormal distribution to them using Matlab and got the parameters from that.

[rquey]

As you can see above, the value you provided to diameq:lognormal() isn't the actual average diameter value, since it is negative.
From you data, I would just compute the (weighted) average and standard deviation of the grain diameters in the standard way, without going into fitting a lognormal distribution.

[Ava-22-neper]

Hi @rquey

Thank you very much for your answer. I also had some issues regarding the outputs and the way I compared them with experimental data. Using this and changing my strategy, the outcome has become much better.