How do I generate a random number between two numbers with using a distribution

I I know that we can define distribution in the random command, but the random numbers I generate with the random command are integers. I'm using the rand command for decimal random numbers and this time I can't define a distribution. In short, how can I create a normal distribution or triangular distribution of 100 random numbers consisting of decimal numbers?

Answer 1

[Bruno Luong]

% define the siscrete value and their relative pdf
xval=(-10:10);
p=exp(-(xval/5).^2); % Gaussian function, triangular whatever you like

n = 1e6; % number of samples
c=cumsum(p);
c=c/c(end);
xrand=xval(discretize(rand(1,n ),[0; c(:)])); % here is the random

% Check graphically
p = p/sum(p);
figure
subplot(2,1,1);
plot(xval,p)
subplot(2,1,2);
histogram(xrand)

[Emre Can Yilmaz]

does the midpoint have to be 0? can I change that? and the values are not decimal, they are integers. this is what i was asking.

[Bruno Luong]

If you want continuous distribution (and not discrete, I don't know what means "decimal" and the computer is the same as me)

% define the siscrete value and their relative pdf
xval=(-10:10);
xmid = (xval(1:end-1) + xval(2:end))/2;
p=exp(-((xmid-3)/5).^2); % Gaussian function, triangular whatever you like

n = 1e6; % number of samples
c=cumsum(p);
c=c/c(end);
xrand=interp1([0 c], xval, rand(1,n));

% Check graphically
p = p/sum(p);
figure
ax1=subplot(2,1,1);
plot(xmid,p)
ax2=subplot(2,1,2);
histogram(xrand,100)
linkaxes([ax1 ax2], 'x')

[Emre Can Yilmaz]

What I was talking about was xrand values. When I looked at your first example, the xrand values were showing as integers like 1-2-3-4. I asked how this would become a decimal number. In your second example, xrand values became decimal numbers like 0.657. I got the answer I wanted, thank you.

Answer 2

[Paul]

Emre,

The statement that the the random() command only generates integers doesn't sound correct. The doc page shows an an example of generating random numbers from a Weibull diistribution which are not integers. Can you post an example that illustrates the behavior?

For a standard normal distribution you can use randn(), and then adjust the outputs for whatever mean and variance is desired. Or use normrnd() in the Statistics and Machine Learning Toolbox, or other options in that toolbox. That toolbox supports many distributions, including the Triangular distribution, for which you can get the pdf, cdf, random numbers, and lots of other interesting things.

[Paul]

Here's an example using Stats Toolbox functionality

pd = makedist('Triangular',0,20,100)

pd = 
  TriangularDistribution

A = 0, B = 20, C = 100

rng(100);
v = random(pd,100000,1);
histogram(v,'Normalization','pdf')

[Emre Can Yilmaz]

Thank you so much!