[Date Index] [Thread Index] [Author Index]

Re: Re: Simulating Correlated non-Normal Random Variables

• To: mathgroup at smc.vnet.net
• Subject: [mg32594] Re: [mg32476] Re: Simulating Correlated non-Normal Random Variables
• From: "Johannes Ludsteck" <johannes.ludsteck at wiwi.uni-regensburg.de>
• Date: Thu, 31 Jan 2002 01:45:16 -0500 (EST)
• Organization: Universitaet Regensburg
• Sender: owner-wri-mathgroup at wolfram.com

Erich Neuwirth wrote:

>>you can apply the same method to any set of independent variables
>>
>>when x1 ... x1 are independent variables
>>and A is a matrix
>>
>>A.(x1,..xn)
>>
>>has covariance matrix A'A
>>
>>no normality assumptions are needed for that.

The statement concerning the covariance Matrix ist correct,
but it is not clear in general,
whether you obtain a multivariate beta-distribution
if you apply this transformation to a vector of univariate
beta-distributed random variables.
The invariance property (linear combinations of normal deviates
are again normal) is a special property of normal deviates.
If this is not clear to you, please try adding two uniform
deviates with different support.

Best regards,
	Johannes

<><><><><><><><><><><><><><><><><><>
Johannes Ludsteck
Institut fuer Volkswirtschaftslehre
Lehrstuhl Prof. Dr. Moeller
Universitaet Regensburg
Universitaetsstrasse 31
93053 Regensburg
Tel +49/0941/943-2741