Efficient allocation of random numbers(U(0,1)) into categories

I am trying to create a function which takes two arguments as input, one a vector of values drawn from a uniform distribution and the other input a number of values between 0 ad 1. It should then index the values from the first input with a number corresponding to its category in the second input.

For example Input1 = [0.2 0.5 0.90 0.995] input2 = [0 0.81 0.99 1] Output = [0 0 1 2]

At present i am using a double for loop with an if statement

    for i=1:length(randoms)
    for r = 1:(length(ShockProbabilities)-1)
       if  randoms(i) >= (ShockProbabilities(r)) && randoms(i) < (ShockProbabilities(r+1));  randoms(i)=(r-1); end 
    end

end

Could any one suggest how i might speed this up ? i am sure there is a way and i dont like using 2 for loops and an if statement. I am just curious, i dont really need it as such

Thanks :)

Answer 1

[Matt Fig]

How about the HISTC function?

Also, in your code there is no reference to either input1 or input2, so I cannot run it...

[DoVile Last Name:]

HISTC doesnt quite work, this code should run:

%Initializations
clear; close; clc  

randoms =[0 0.5 0.91 0.995];
ShockProbabilities = [0 0.81 0.99 1];

for i=1:length(randoms)
    for r = 1:(length(ShockProbabilities)-1)
       if  randoms(i) >= (ShockProbabilities(r)) && randoms(i) < (ShockProbabilities(r+1));  randoms(i)=(r-1); end 
    end

end
randoms

randoms =[0 0.5 0.91 0.995];
ShockProbabilities = [0 0.81 0.99 1];

histc(randoms,ShockProbabilities)

So when i try to use HISTC i sort of get the answer but backwards and not quite

[Matt Fig]

Try this:

[J,K] = histc(randoms,ShockProbabilities);
K = K-1

By the way, to make your code faster you could put a break after the randoms(i) reassignment since it is presumed that only one fit will be found per element. As it is you continue to check even after you find where randoms(i) goes! You could also only run the inner loop from the last r found to the end-1 because randoms is sorted. These two hints would speed things up considerably for large arrays. Still, I doubt you will beat HISTC.

[DoVile Last Name:]

Great i got it working, and the break tip was a real eye opener for me! in the actual program i will be using quasirandom numbers instead of the ones from the example so they wont be sorted, also i kind of want a more robust function.

Thanks Matt :)