Harry Garst

219 Reputation

5 Badges

16 years, 337 days

MaplePrimes Activity


These are replies submitted by Harry Garst

Yes I was aware of this post and I tried to include points in their solutions, but I did not succeed.

Thanks anyway,

Harry Garst

One way to explain it is that I do  not want to take a random sample from the MVN distribution, but I want to choose values from the MVN in such a way that the scores are as close as possible to the MVN distribution.

 
One possible way might be to obtain N values from the univariate distribution by constructing an N-vector with the z-values that are associated with p = 1/(N+1), 2/(N+1), ..., N/(N+1). Such a vector has an exact univariate normal distribution with m = 0 and s = 1.
 
For a k-variate normal distribution we could construct k such vectors, and reorder their values in such a way that the vectors become orthogonal, in order to obtain k vectors of length N that exactly follow the MVN distribution with m = 0 and S = identity.
 
Thanks Axel, I did not tried it yet, but it looks good. Harry
Thanks Axel, I did not tried it yet, but it looks good. Harry
okay, I finally found the bookmark http://personal.kenyon.edu/hartlaub/MellonProject/mellon.html You can download the library: http://personal.denison.edu/~karian/maple_supplement.html Does anybody know how to use this library in Maple 10 or 11? Maybe it only works in previous versions of Maple. Harry Garst
okay, I finally found the bookmark http://personal.kenyon.edu/hartlaub/MellonProject/mellon.html You can download the library: http://personal.denison.edu/~karian/maple_supplement.html Does anybody know how to use this library in Maple 10 or 11? Maybe it only works in previous versions of Maple. Harry Garst
1 2 3 4 Page 4 of 4