2.0.0 • Published 7 years ago
correl-z v2.0.0
correl-z
correlated standard normal distribution random number generator
Example
var correlZ = require('correlZ'),
randomZ = require('random-z'), // unit normal distribution random number generator
cholesky = require('cholesky') // cholesky decomposition
var correlMatrix = [[1], [0.2, 1], [0.2, 0.2, 1]] // lower triangle is enough
correlWeights = cholesky(correlMatrix),
correlSeeders = correlWeights.map(correlZ) // array of random number generators
var iids = correlWeights.map(randomZ) // shared identically distributed variables
// all 3 results below are different but correlated, E=0, V=1, Cor(i,j) = 0.2
var x0 = correlSeeder[0](iids),
x1 = correlSeeder[1](iids),
x2 = correlSeeder[2](iids)API
The module exports a single function that takes an array of linear factors or a key-factor object to be applied to shared standard normal Independent and identically distributed random_variables (iids) and returns a standard normal correlated random number generator.
| Name | Type | Notes / Examples |
|---|---|---|
correlZ | weights => randomFcn | rand = correlZ({2:0.1, 1:0.2}) |
weights | Object | Array | Σv²<1: {1:0.2}, [0, 0.2], {a: 0.2} |
randomFcn | (iidZs [,selfZ]) => Number | Zi<1: sample = randomFcn({a:0.7, b:0.4}) |
iidZs | Object | Array of zSeed | {a:0.7, b:0.4} |
selfZ | zSeed | optional standard normal seed for testing. Normally generated internally |
zSeed | Number | standard normal random seed -1 < v < 1 |
Note that the linear iid weights can obtained from the correlation matrix with the cholesky module.