r = randi(10,100,1) but using non-uniform distribution

How to create random number from a range non-uniformly distributed?

Thanks!

[Jan]

There is an infinite number of non-uniform distributions. Which one do you want?

[Vikas Tiwari]

how to get random normally distributed numbers in 5by7matrix?

[Walter Roberson]

randn(5,7)

Answer 1

[the cyclist]

As Jan comments, there are literally an infinite number of possible distributions. (In particular, do you want integers or real numbers in that range?)

If you have the Statistics Toolbox, there are several random number generators available. You can type

>> doc random

to see a list of possibilities. Another command that might be handy is randsample(). You can specify a particular frequency distribution for the random numbers. See

>> doc randsample

for details.

Answer 2

[Steven Lord]

One way to generate numbers from a "weighted die" is to use discretize with a vector of probabilities. In this example, the weight vector P indicates that I want to generate numbers between 1 and 4 but 1 is twice as likely as each of the other three numbers.

P = [2 1 1 1];

Scale the weight so they sum to 1.

prob = normalize(P, 'scale', sum(P));

Make a vector of probability edges

cprob = cumsum([0 prob]);

Generate some sample data and discretize that sample data to determine into which bin (defined by the probability edges) each of the sample data point falls.

x = rand(1, 1e5);
roll = discretize(x, cprob);

Show a histogram of the rolls and you should see the bar for 1 is twice as high as the bars for 2, 3, and 4.

histogram(roll, 'Normalization', 'probability')