Multivariate random number generators

Multivariate random number generators

[Przemyslaw Sliwa]

Hi,

I have coded four random number generators for multivariate distributions.
They include:
   Multivariate (elliptical) t-distribution
   Multivariate generalized Laplace distribution,
   Multivariate skew-normal distribution
   Multivariate skew-t student distribution.

All distributions are expressed in terms of univariate distribution (normal, gamma and chi^2), so that no special inverse transformation or similar method is used. I have done it for my thesis  and thought it can be useful for others. If you are interested please let me know and I will code it using the GSL standards.

By the way: what is the problem to put the complete MS Visual build on the webpage?
I have managed to compile the library, it was straightforward to me but other people have lots of problems and this topic appears very often on the list.

Thanks,

Przem


--------------------------------------------------
Deutsche Bank AG
Risk Controlling
Credit Risk Engineering & Implementation
            /  Daily Credit Risk

Tel.:  (+44 20) 754 59676 London
Fax:   (+44 20) 754  71060 London

----------------------------------------------------


--

This e-mail may contain confidential and/or privileged information. If you are not the intended recipient (or have received this e-mail in error) please notify the sender immediately and destroy this e-mail. Any unauthorized copying, disclosure or distribution of the material in this e-mail is strictly forbidden.

Re: Multivariate random number generators

[Brian Gough]

Przemyslaw Sliwa writes:
 > All distributions are expressed in terms of univariate distribution
 > (normal, gamma and chi^2), so that no special inverse
 > transformation or similar method is used. I have done it for my
 > thesis and thought it can be useful for others. If you are
 > interested please let me know and I will code it using the GSL
 > standards.

Note: Check with your employer that you are allowed to release your
own software under the GPL -- unfortunately many banks are quite
strict about not allowing their employees to do that.

 > By the way: what is the problem to put the complete MS Visual build
 > on the webpage?  I have managed to compile the library, it was
 > straightforward to me but other people have lots of problems and
 > this topic appears very often on the list.

My tolerance for dealing with Microsoft Windows and Microsoft Visual
C++ has been completely used up.  Clint Whaley, the author of ATLAS,
describes the phenomenon on his site:
  http://math-atlas.sourceforge.net/faq.html#utone
 
-- 
Brian Gough

Re: Multivariate random number generators

[Robert G. Brown]

On Fri, 5 Mar 2004, Brian Gough wrote:

> My tolerance for dealing with Microsoft Windows and Microsoft Visual
> C++ has been completely used up.  Clint Whaley, the author of ATLAS,
> describes the phenomenon on his site:
>   http://math-atlas.sourceforge.net/faq.html#utone

What a great URL.  I'll have to keep this one handy!

And just to comply with its content, I personally absolutely love the
GSL and both use it in my own work, teach students to turn to it first
when looking for numerical solutions, and encourage others to use it for
all their GPL work.

Thanks!

   rgb

-- 
Robert G. Brown	                       http://www.phy.duke.edu/~rgb/
Duke University Dept. of Physics, Box 90305
Durham, N.C. 27708-0305
Phone: 1-919-660-2567  Fax: 919-660-2525     email:rgb@phy.duke.edu

Re: Multivariate random number generators

[linas]

On Fri, Mar 05, 2004 at 06:57:09PM +0000, Brian Gough wrote:
> Przemyslaw Sliwa writes:
>  > By the way: what is the problem to put the complete MS Visual build
>  > on the webpage?  I have managed to compile the library, it was
>  > straightforward to me but other people have lots of problems and
>  > this topic appears very often on the list.
> 
> My tolerance for dealing with Microsoft Windows and Microsoft Visual
> C++ has been completely used up.  Clint Whaley, the author of ATLAS,
> describes the phenomenon on his site:
>   http://math-atlas.sourceforge.net/faq.html#utone

I agree/support Clint/Brian.  This is a cultural divide between 
Windows users and all free software projects, not just atlas/gsl.  
I suppose ESR probably has an anthropological analysis of it 
somewhere.

--linas

Re: Multivariate random number generators

[Brian Gough]

Robert G. Brown writes:
 > On Fri, 5 Mar 2004, Brian Gough wrote:
 > 
 > > My tolerance for dealing with Microsoft Windows and Microsoft Visual
 > > C++ has been completely used up.  Clint Whaley, the author of ATLAS,
 > > describes the phenomenon on his site:
 > >   http://math-atlas.sourceforge.net/faq.html#utonew
 > 
 > What a great URL.  I'll have to keep this one handy!

Indeed, as he says, "I am not confident that users will read this, but
I am confident that I will use its URL in replying to a lot of future
requests".

I should make clear that in this case it only applies to the download
for Microsoft Visual C++.... somehow a 1-click Windows install set the
barrier too low, and attracted way too many people who thought they
did not need to read beyond page 1 of K&R (let alone the GSL Manual)
because they could just email me instead.

In contrast I've never encountered that problem with people using
Unix, thankfully.

-- 
Brian 

Re: Multivariate random number generators

[Przemyslaw Sliwa]

Robert G. Brown writes:
 > On Fri, 5 Mar 2004, Brian Gough wrote:
 >
 > > My tolerance for dealing with Microsoft Windows and Microsoft Visual
 > > C++ has been completely used up.  Clint Whaley, the author of ATLAS,
 > > describes the phenomenon on his site:
 > >   http://math-atlas.sourceforge.net/faq.html#utonew
 >
 > What a great URL.  I'll have to keep this one handy!

Indeed, as he says, "I am not confident that users will read this, but
I am confident that I will use its URL in replying to a lot of future
requests".

I should make clear that in this case it only applies to the download
for Microsoft Visual C++.... somehow a 1-click Windows install set the
barrier too low, and attracted way too many people who thought they
did not need to read beyond page 1 of K&R (let alone the GSL Manual)
because they could just email me instead.

In contrast I've never encountered that problem with people using
Unix, thankfully.

--
Brian


Hmmmm,

Maybe I shall remove my previous email from the list to avoid lots of emails from the Win32 users ???

Regards




--------------------------------------------------
Deutsche Bank AG
Risk Controlling
Credit Risk Engineering & Implementation
            /  Daily Credit Risk

Tel.:  (+44 20) 754 59676 London
Fax:   (+44 20) 754  71060 London

----------------------------------------------------


--

This e-mail may contain confidential and/or privileged information. If you are not the intended recipient (or have received this e-mail in error) please notify the sender immediately and destroy this e-mail. Any unauthorized copying, disclosure or distribution of the material in this e-mail is strictly forbidden.

Re: Multivariate random number generators

[Rodney Sparapani]

>Hi,
>
>I have coded four random number generators for multivariate distributions.
>They include:
>   Multivariate (elliptical) t-distribution
>   Multivariate generalized Laplace distribution,
>   Multivariate skew-normal distribution
>   Multivariate skew-t student distribution.
>
>All distributions are expressed in terms of univariate distribution (normal, 
gamma and chi^2), so that no special inverse transformation or similar method is 
used. I have done it for my thesis  and thought it can be useful for others. If 
you are interested please let me know and I will code it using the GSL 
standards.

>Przem
>

Przem:

I would love to see it in GSL.  Let me know if I can help.

Rodney Sparapani              Medical College of Wisconsin
Sr. Biostatistician           Patient Care & Outcomes Research
rsparapa@mcw.edu              http://www.mcw.edu/pcor
Was 'Name That Tune' rigged?  WWLD -- What Would Lombardi Do