• Status: Solved
• Priority: Medium
• Security: Public
• Views: 573

Hello,

The problem that I have is that I am trying to add Poisson distributed noise to a set of data, which is simply stored as a list of numbers. I was wondering if anyone knew any easy ways of performing this task.

Many thanks

0
FaustVs
• 3
• 2
1 Solution

Commented:
Can you explain more what you want? I don't understand your question, what is Poison Nose?
0

Commented:
Poisson distribution looks like:

P(X = m) = (a^m)*exp(-a)/(m!)
where a=M(X)

First, choose a range of X Poisson-distributed values. Because X allwais is a integer number, the range will contain a number of possible values. For example, if a=0.1 there is enough to calculate 5 values (X = 0 ... 4); if a=10, 30 values is enough. Probabilities of the other values will be very small (less then 0.000001). Build the table of (X, Y) values using the following rules.

P(0) = exp(-a);
P(X + 1) = P(X) * a / (X + 1);
Y(0) = P(0);
Y(X + 1) = Y(X) + P(X + 1)

Now suppose that we have random generator with uniform distribution (for example, pascal's Random function) which will generate continuous values from 0 to 1 exclusively. Call it R. Generate R = 0 ... 1 and using binary or straight search find the nearest Y(X) value. X will be the poisson-distributed value.

This method is called "table method" and could be used to generate a number X values with constant a (Poisson distributed noise).
0

Commented:
tilex: learn math ;)
0

Commented:
I'm only in secondary 3...I don't konw what's its like in US but well I'm 15, IV'E got math notions but I'm not that far...
0

Commented:
tilex: I hope you would like it ;)
0

Featured Post

• 3
• 2
Tackle projects and never again get stuck behind a technical roadblock.