Thread Subject: Random matrix/normal distribution

I'm trying to generate random matrices whose entries are from the normal distribution.

Something like this:

(1/(sqrt(2*pi)))*exp(-(1+99*rand(5))^2/2)

Isn't going to work because squaring a matrix doesn't square each entry. How can I correct this or am I going about this the wrong way.

I'm still a MATLAB beginner so go easy on me!

Michael Grosswald wrote:
> I'm trying to generate random matrices whose entries are from the normal distribution.
>
> Something like this:
>
> (1/(sqrt(2*pi)))*exp(-(1+99*rand(5))^2/2)
>
> Isn't going to work because squaring a matrix doesn't square each
> entry. How can I correct this or am I going about this the wrong way.

> I'm still a MATLAB beginner so go easy on me!


Well,

lookfor normal

returns

 >> lookfor normal
RANDN Normally distributed random numbers.
...
SPRANDN Sparse normally distributed random matrix.
...

which might look interesting... :)

As for the question on squaring terms in an array (or vector), look at
".*" instead of "*"

--

A .^ 2

This will take your matrix and square it element by element.

.* is similar.

Doug

"Michael Grosswald" <michaelgrosswald@yahoo.co.uk> wrote in message <h9tpd6$bh9$1@fred.mathworks.com>...
> I'm trying to generate random matrices whose entries are from the normal distribution.
>
> Something like this:
>
> (1/(sqrt(2*pi)))*exp(-(1+99*rand(5))^2/2)
>
> Isn't going to work because squaring a matrix doesn't square each entry. How can I correct this or am I going about this the wrong way.
>
> I'm still a MATLAB beginner so go easy on me!

Thanks a lot to both of you.

> I'm trying to generate random matrices whose entries are from the normal
> distribution.
>
> Something like this:
>
> (1/(sqrt(2*pi)))*exp(-(1+99*rand(5))^2/2)
>
> Isn't going to work because squaring a matrix doesn't square each entry.
> How can I correct this or am I going about this the wrong way.

Michael, if you square the matrix, this will give you values of the normal
density function evaluated at random points. I encourage you to try the
advice from dbp instead of squaring the matrix in this formula.

-- Tom