Normal Random number generation

I want to generate random numbers from a standard normal distribution with decreasing standard deviation that lies between 0 and 1. First, what is meant by decreasing standard deviation? Secondly, randn generates normal distribution random numbers. But how to limit them between 0 and 1?

Answer 1

[James Tursa]

Maybe this link will help you:

https://www.mathworks.com/matlabcentral/answers/325625-how-do-i-create-random-normally-distributed-number-array-within-a-set-range

[Rakesh Jain]

But what is meant by decreasing standard deviation

[James Tursa]

I don't know. Maybe your professor wants you to make several different runs with various values of standard deviation and then make some conclusions about the results you get. What is the actual wording of the problem you are working?

[Rakesh Jain]

I had to minimise errror using a optimisation algorithm (BBBC) which I had to code using Matlab. There was a statement :'r' is a random number from a standard normal distribution with decreasing standard deviation

[Rakesh Jain]

r lies between 0 and 1

[Walter Roberson]

I do not seem to find any optimization algorithm abbreviated as BBBC. Perhaps you were referring to Broad Bioimage Benchmark Collection ?

[Rakesh Jain]

@Walter Roberson, BBBC is Big Bang Big Crunch Optimisation Algorithm by Osman K. Erol and Ibrahim Eksin discovered in 2006

[Walter Roberson]

Decreasing standard deviation would result in an algorithm with similarities to Simulated Annealing.

Answer 2

[Walter Roberson]

My interpretation is that the normal distribution is to be unbounded, not restricted to the range [0 1], but that the standard deviation used is to start and 1 and decrease to 0. For example,

SD = linspace(1, 0, 20)
results = randn(1,20) .* SD

That would give you 20 results with stand deviation decreasing from 1 to 0.

As for why... I don't know?

[Rakesh Jain]

Got it. Thanks