2

I have a problem with my statistics task. It sounds like this: "Using the Monte Carlo method, plot the dependence of a on b for a from -5 to 5, where $P (a < x-y < b) = 0.95$, where $x\sim N(0,1)$, and y is a uniformly distributed random variable from -2 to 2." what should I do, where to start? until I have no ideas.

To be honest, I did not understand anything in the following answer: "First find the probability distribution of random variable X−Y using montecarlo method.

1.Pick an x and y randomly according to their densities and then compute x−y. Do this N times.

2.Make a list x−y values.

3.Sort the list in ascending order.

4.Fix an b which is in [−5,5].

5.From count number of values less than b, call this k.

$\cdot$ If k/N>0.95: Count from smallest value till you get the k/N less more or less equal to 0.95. The value you find is the a

$\cdot$ If k/N<0.95:

No such a exists."

I tried to code it and I have a lot of questions.

First of all, I created a list for $x-y$.

 dif<-x-y

Then, I created a list of values of a, which is from -5 to 5

a<-seq(-5, 5, 0.05)

But now I must go to the 5-th item. It sounds like this: From count number of values more than a, call this k. What should I do here? I must take values from $x-y$, which are more than a? Then I should count $k/N$ and if it is more than 0.95 what is next? "Count from smallest value", but what smallest value?

Thanks for help

1 Answers1

4

First find the probability distribution of random variable $X-Y$ using montecarlo method.

  1. Pick an $x$ and $y$ randomly according to their densities and then compute $x-y$. Do this $N$ times.
  2. Make a list $x-y$ values.
  3. Sort the list in ascending order.
  4. Fix an b which is in $[-5,5]$.
  5. From count number of values less than $b$, call this $k$.
    • If $k/N>0.95$

      • Count from smallest value till you get the $k/N$ less more or less equal to 0.95.
      • The value you find is the $a$
    • If k/N<0.95

      • No such a exists.
Sonal_sqrt
  • 4,711