R help - Apr 2009 - Need Advice on Matrix Not Positive Semi-Definite with cholesky decomposition

Dear fellow R Users:

I am doing a Cholesky decomposition on a correlation matrix and get error
message
the matrix is not semi-definite.

Does anyone know: 
1- a work around to this issue?
2- Is there any approach to try and figure out what vector might be co-linear
with another in thr Matrix?
3- any way to perturb the data to work around this?

Thanks for any suggestions.

Look at the nearPD() function in the package "Matrix".

	require(Matrix)
	?nearPD

In particular, pay attention to the arguments "eig.tol" and
"posd.tol",
which you can tweak to define how much "positiveness" you would like
to
have.

Ravi. 


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

Ravi Varadhan, Ph.D.

Assistant Professor, The Center on Aging and Health

Division of Geriatric Medicine and Gerontology 

Johns Hopkins University

Ph: (410) 502-2619

Fax: (410) 614-9625

Email: rvaradhan at jhmi.edu

Webpage:  http://www.jhsph.edu/agingandhealth/People/Faculty/Varadhan.html

 

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


-----Original Message-----
From: r-help-bounces at r-project.org [mailto:r-help-bounces at r-project.org]
On
Behalf Of Gottlieb, Neil
Sent: Wednesday, April 01, 2009 9:57 AM
To: 'r-help at r-project.org'
Subject: [R] Need Advice on Matrix Not Positive Semi-Definite with cholesky
decomposition

Dear fellow R Users:

I am doing a Cholesky decomposition on a correlation matrix and get error
message the matrix is not semi-definite.

Does anyone know: 
1- a work around to this issue?
2- Is there any approach to try and figure out what vector might be
co-linear with another in thr Matrix?
3- any way to perturb the data to work around this?

Thanks for any suggestions.
______________________________________________
R-help at r-project.org mailing list
https://stat.ethz.ch/mailman/listinfo/r-help
PLEASE do read the posting guide http://www.R-project.org/posting-guide.html
and provide commented, minimal, self-contained, reproducible code.

Neil,
   pls  tell why do you need the correlation matrix? if you are trying to
   simulate correlated variables then you can go around the cholesky by using
   svd.
   if you really need the correlation ( I think it is always possible to avoid
   it  )  then Rmetrics have a function to turn your matrix into positive
   semi-definite.
   you can also consider a factor model which will reduce the dimensionality
   tremendously.
   Stephen C. Bond

   On Apr 1, 2009, ngottlieb at marinercapital.com wrote:

     Dear fellow R Users:
     I am doing a Cholesky decomposition on a correlation matrix and get error
     message
     the matrix is not semi-definite.
     Does anyone know:
     1- a work around to this issue?
     2- Is there any approach to try and figure out what vector might be
     co-linear with another in thr Matrix?
     3- any way to perturb the data to work around this?
     Thanks for any suggestions.
     ______________________________________________
     R-help at r-project.org mailing list
     https://stat.ethz.ch/mailman/listinfo/r-help
     PLEASE do read the posting guide
     http://www.R-project.org/posting-guide.html
     and provide commented, minimal, self-contained, reproducible code.