MATLAB Central Newsreader - random number generation

Re: random number generation

Greg Heath — Wed, 19 Mar 2008 14:01:56 -0400

On Mar 18, 3:46=A0pm, richardsta...@comcast.net wrote:
> On Tue, 18 Mar 2008 19:42:01 +0000 (UTC), "Roger Stafford"
>
>
>
>
>
> <ellieandrogerxy...@mindspring.com.invalid> wrote:
> >"AMMAL " <alan...@ualberta.ca> wrote in message <frp525$8i
> >$...@fred.mathworks.com>...
> >> Hellow everybody,
>
> >> I need a hand on how to generate a 100 positive numbers
> >> that follow a normal distribution curve (Bell Shape).In
> >> otherwords, if i histogram the data, a bell shape will be
> >> generated. i am not transforming my data, instead, i want
> >> to generate a posirtive one hundred values that will follow
> >> a bell shape.
>
> >> Thanks,
> >> Ammal
> >------
> > =A0You can use "s*randn(1,100)+m" to generate a hundred normal distribut=
ion
> >numbers with mean m and standard deviation s, but you cannot be sure all =
of
> >them will be positive. =A0You cannot have a normal distribution and at th=
e same
> >time have only positive numbers. =A0That is a contradiction in terms. =A0=
The "bell
> >shape" extends all the way to minus and plus infinity.
>
> >Roger Stafford
>
> Roger's advice is correct of course.
> Having said that, you can insure that a *sample*is entirely positive.
> For example:
> x =3D s*randn(1,100)+m;
> if min(x)<0
> =A0x =3D x + abs(min(x));
> end

Alternatively, you can reject negative values
and keep generating until you have 100.

In either case, the resulting distribution is ...

ABNORMAL! (;>)

Greg

Re: random number generation

Vihang Patil — Wed, 19 Mar 2008 08:20:04 -0400

"AMMAL " <alanazi@ualberta.ca> wrote in message <frp525
$8i$1@fred.mathworks.com>...
> Hellow everybody,
>
> I need a hand on how to generate a 100 positive numbers
> that follow a normal distribution curve (Bell Shape).In
> otherwords, if i histogram the data, a bell shape will
be
> generated. i am not transforming my data, instead, i
want
> to generate a posirtive one hundred values that will
follow
> a bell shape.
>
> Thanks,
> Ammal

Hello
I am not sure about the Bell thing, but if you want to
generate 100 random positive numbers then you can use this
n = 100;
f = ceil(n.*rand(100,1));

HTH
Vihang

Re: random number generation

ammal — Tue, 18 Mar 2008 20:17:01 -0400

Thank you all for your answer,
I am wondering now how i would include the codition
statement you wrote in the block code to ensure the samples
are positive.Could you please show me how to include it in
a for loop maybe.

Thanks,
Ammal

richardstartz@comcast.net wrote in message
<8r60u3p6eauu99plunl4oo31dd1o242jq2@4ax.com>...
> On Tue, 18 Mar 2008 19:42:01 +0000 (UTC), "Roger Stafford"
> <ellieandrogerxyzzy@mindspring.com.invalid> wrote:
>
> >"AMMAL " <alanazi@ualberta.ca> wrote in message <frp525
$8i
> >$1@fred.mathworks.com>...
> >> Hellow everybody,
> >>
> >> I need a hand on how to generate a 100 positive
numbers
> >> that follow a normal distribution curve (Bell
Shape).In
> >> otherwords, if i histogram the data, a bell shape will
be
> >> generated. i am not transforming my data, instead, i
want
> >> to generate a posirtive one hundred values that will
follow
> >> a bell shape.
> >>
> >> Thanks,
> >> Ammal
> >------
> > You can use "s*randn(1,100)+m" to generate a hundred
normal distribution
> >numbers with mean m and standard deviation s, but you
cannot be sure all of
> >them will be positive. You cannot have a normal
distribution and at the same
> >time have only positive numbers. That is a
contradiction in terms. The "bell
> >shape" extends all the way to minus and plus infinity.
> >
> >Roger Stafford
>
> Roger's advice is correct of course.
> Having said that, you can insure that a *sample*is
entirely positive.
> For example:
> x = s*randn(1,100)+m;
> if min(x)<0
> x = x + abs(min(x));
> end
>
> -Dick Startz

Re: random number generation

richardstartz@comcast.net — Tue, 18 Mar 2008 19:46:06 -0400

On Tue, 18 Mar 2008 19:42:01 +0000 (UTC), "Roger Stafford"
<ellieandrogerxyzzy@mindspring.com.invalid> wrote:

>"AMMAL " <alanazi@ualberta.ca> wrote in message <frp525$8i
>$1@fred.mathworks.com>...
>> Hellow everybody,
>>
>> I need a hand on how to generate a 100 positive numbers
>> that follow a normal distribution curve (Bell Shape).In
>> otherwords, if i histogram the data, a bell shape will be
>> generated. i am not transforming my data, instead, i want
>> to generate a posirtive one hundred values that will follow
>> a bell shape.
>>
>> Thanks,
>> Ammal
>------
> You can use "s*randn(1,100)+m" to generate a hundred normal distribution
>numbers with mean m and standard deviation s, but you cannot be sure all of
>them will be positive. You cannot have a normal distribution and at the same
>time have only positive numbers. That is a contradiction in terms. The "bell
>shape" extends all the way to minus and plus infinity.
>
>Roger Stafford

Roger's advice is correct of course.
Having said that, you can insure that a *sample*is entirely positive.
For example:
x = s*randn(1,100)+m;
if min(x)<0
x = x + abs(min(x));
end

-Dick Startz

Re: random number generation

Roger Stafford — Tue, 18 Mar 2008 19:42:01 -0400

"AMMAL " <alanazi@ualberta.ca> wrote in message <frp525$8i
$1@fred.mathworks.com>...
> Hellow everybody,
>
> I need a hand on how to generate a 100 positive numbers
> that follow a normal distribution curve (Bell Shape).In
> otherwords, if i histogram the data, a bell shape will be
> generated. i am not transforming my data, instead, i want
> to generate a posirtive one hundred values that will follow
> a bell shape.
>
> Thanks,
> Ammal
------
You can use "s*randn(1,100)+m" to generate a hundred normal distribution
numbers with mean m and standard deviation s, but you cannot be sure all of
them will be positive. You cannot have a normal distribution and at the same
time have only positive numbers. That is a contradiction in terms. The "bell
shape" extends all the way to minus and plus infinity.

Roger Stafford

Re: random number generation

Randy Poe — Tue, 18 Mar 2008 19:34:08 -0400

On Mar 18, 3:27 pm, "AMMAL " <alan...@ualberta.ca> wrote:
> Hellow everybody,
>
> I need a hand on how to generate a 100 positive numbers
> that follow a normal distribution curve (Bell Shape).In
> otherwords, if i histogram the data, a bell shape will be
> generated. i am not transforming my data, instead, i want
> to generate a posirtive one hundred values that will follow
> a bell shape.
>
> Thanks,
> Ammal

The function RANDN will generate samples from a
standard normal distribution. Note that if you histogram
the results, you won't get a perfect bell shape since samples,
being random, have a little variation in them. The mean
of your samples won't be exactly 0, but something close.
The variance won't be exactly 1, but something close.

That's the nature of random variables and samples.

- Randy

random number generation

ammal — Tue, 18 Mar 2008 19:27:01 -0400

Hellow everybody,

I need a hand on how to generate a 100 positive numbers
that follow a normal distribution curve (Bell Shape).In
otherwords, if i histogram the data, a bell shape will be
generated. i am not transforming my data, instead, i want
to generate a posirtive one hundred values that will follow
a bell shape.

Thanks,
Ammal