1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
|
--[[ https://en.wikipedia.org/wiki/Normal_distribution ]]
-- The Box–Muller method
local function gaussian(mean, variance)
local U = math.random()
local V = math.random()
return math.sqrt(-2.0 * variance * math.log(U)) *
math.cos(2.0 * math.pi * V) + mean
end
local function mean(t)
local sum = 0
local count = #t
for i = 1, count do
sum = sum + t[i]
end
return sum / count
end
local function std(t, mean)
local squares = 0.0
for i = 1, #t do
local deviation = math.abs(mean - t[i])
squares = squares + deviation * deviation
end
local variance = squares / #t
return math.sqrt(variance)
end
local function do_the_call()
local t = {}
local mu = 34.0
local sigma = 10.0
for i = 1, 5 do
table.insert(t, gaussian(mu, sigma))
end
return string.format("Got mean: %1.5f, mu: %1.5f\nstd deviance:%1.5f, expected: %1.5f",
mean(t), mu,
std(t, mu), math.sqrt(sigma))
end
math.randomseed(os.time())
return do_the_call
|