Set boundaries for norminv output

Hi.

I am trying to generate randomized data with a certain average and a standard deviation. However, the data has to be within a realistic range.

My attempt so far has been as follows:

JV = norminv(rand(dim),4700,1500); %dim is the dimensions of the output

The problem with this method is that I have to manually limit the output:

JV(JV < 1000) = 1000;
JV(JV > 9000) = 9000;

The mean and standard deviation are still within an acceptable range. However, due to the manual limitation, the histogram shows sudden "jumps" in the amounts at the bins 1000 and 9000.

Is there a way I can set the boundaries within norminv, so that I get a "clean" histogram?

Thanks!

- Marc

UPDATE: I found a solution:

P1 = normcdf(1000,4700,1500);
P2 = normcdf(9000,4700,1500);
JV = norminv(P1+(P2-P1).*rand(dim),4700,1500);

This "cleans up" the histogram. But of course the standard deviation is slightly smaller than 1500, since a strictly gaussian distribution is not possible within a limited interval.

Answer 1

[Torsten]

Use a truncated normal distribution:

http://www.mathworks.com/matlabcentral/fileexchange/23832-truncated-gaussian

Best wishes

Torsten.