Binomial Simulator in Matlab with M= number of experiments

Hello, I must make a simulator that uses binomial distribution, where

• n= number of tries in each experiment (is either 0(false) or 1(true)
• p= success probability
• m= number of experiments

If for example, let's say n=10, and m= 5, it means the simulator must run 10 tries of F/T events with the given probability (lets say its 0.3) 5 times, which means 50 individual events, and I must graph the results in a histogram with X= number of successes in every N and y= pdf

The thing is, I'm getting confused as how to handle the mix of n and m! Please help

[Matt Trenton]

My first idea was to make a cycle like this:

for j=1:m
    for i=1:n
        r=rand;
        if r <= p
            B(i)=1;
        else
            B(i)=0;
        end
    end
    k(j)=B(1);
end

what the part that says k(j)=B(1) is trying to do it's to count all the successes (1s) of the first iteration of the cycle n, but pretty much to no avail. Any ideas?

Answer 1

[Jeff Miller]

Instead of

      k(j)=B(1);

you need

      k(j)=sum(B);

[Matt Trenton]

It does help a lot!! thank you, but it brings out some (bigger) issues my simulator is having...

In the end I need to graph the simulator's histogram and the theoric binomial results pdf and they must be similar.... which does not happen right now.

-------

I don't even get what's so terribly wrong

[Jeff Miller]

You are only running the simulator m=5 times. That's not nearly enough to ensure a good approximation to the theoretical distribution. Try m=5000.

[Matt Trenton]

Such an idiotic mistake, thank you!