Re: random points with a given dsitribution in rectangle

"clausvcm" <clausvcm@xxxxxxxxxxx> wrote in message
news:1142349258.652163.269240@xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx
ups! I didn't know what you say in point 1). The the question is how to
do that?
...I don't understand 2), I am simply discarding any points outside the
box (acept - reject)
and for 3) the box is just a limited space where I wish to generate
points inside. I imagined that I could get the box as I said in the
last post, but probably I am wrong.
ok, well, probably all this is because I started following my empirical
thinking and after what you say I was walking in the wrong direction.
So, how can i do the things correctly?. I re-state: with a covmatrix
and a mean vector how to generate gaussian points in a box, where the
dimensions of the box is obtained from the covmatrix (and here now I am
not sure I have been done it right :( )

Thanks


Gaussian variates will not necessarily fit in a rectangle. The Gaussian
distribution is infinite in extent. If you throw away samples then the
result is no longer Gaussian. That being said, samples from an uncorrelated
bivariate Gaussian distribution can be obtained by using (possibly scaled)
results from randn for each coordinate. Samples from an correlated bivariate
Gaussian distribution can be obtained by using mvnrnd from the statistics
toolbox. If you dont have that toolbox then there are some simple tricks
using matrix decompositions that I've seen in this newsgroup, but I don't
remember. Perhaps someone else will chime in here.


.

Relevant Pages

  • Re: FFT test with few kbits
    ... ]>>> average to N/2 while the rest should average to zero. ... ]>>> able to test that they have the same gaussian distribution as the ... ]> output of the DFT is a weighted sum of N random inputs. ... ]> Gaussian distributed. ...
    (sci.crypt)
  • Re: Gaussian distribution
    ... How do you prove that a Gaussian distribution integrates to 1 over -oo.. ... I also tried integration by parts two times ... Gaussian filter so I can design them using a desired cutoff frequency. ...
    (sci.math)
  • Re: a random number obeying Gaussian + within a range
    ... I tried to set the random number out of the range to 0, but not sure whether the generated series also obeying Gaussian distribution.. ... Perhaps you're looking for uniformly distributed random numbers? ...
    (comp.soft-sys.matlab)
  • Re: Gaussian distribution
    ... How do you prove that a Gaussian distribution integrates to 1 over -oo.. ... I'm trying to find the fourier transform of a Gaussian ... I also tried integration by parts two times ...
    (sci.math)
  • Re: Gaussian distribution define in infinite-dimensional Hilbert space?
    ... There is a measure associated with an infinite-dimensional Hilbert space ... that is analogous to a multivariate Gaussian distribution. ...
    (sci.math)