Skip to content

quantile function for noncentral f-distribution

12 messages · Juergen Hedderich, Agus Salim, Peter Dalgaard +3 more

Juergen Hedderich
# 
hello R-friends,
I'm looking for a quantile function for the noncentral
f-distribution in the area of equivalence hypotheses testing.
Can somebody help me?

Many thanks 

-----------------------------------------------------------------
Dipl. Inform. J. Hedderich              
Institut f?r Medizinische Informatik       Phone : 0431 /
5973182               
und Statistik im Klinikum an der CAU       Fax   : 0431 / 5973193
               
Brunswiker Str.
10                                                          
24105 Kiel                  
E-mail:hedderich at medinfo.uni-kiel.de                                            
-----------------------------------------------------------------
-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-
r-help mailing list -- Read http://www.ci.tuwien.ac.at/~hornik/R/R-FAQ.html
Send "info", "help", or "[un]subscribe"
(in the "body", not the subject !)  To: r-help-request at stat.math.ethz.ch
_._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._
Brian Ripley
#
On Tue, 9 Jan 2001, Juergen Hedderich wrote:

            

      
      
      
    

        
Since pf has a non-centrality parameter, you can easily invert it.

myqf <- function(p, df1, df2, ncp)
{
uniroot(function(x) pf(x, df1, df2, ncp) - p, , 0, 100)$root
}

adjust 100 to suit (one can surely come up with an useful computed upper
bound if ncp might be large), and add a tol arg to uniroot if you care.

            

      
      
      
    

        

  
    

      
      
    

  


  


      

    

    

      
      

        


  

        

      Agus Salim
    

    

      
      #
    

  


  

      
Hi R-users!

I'm building R package using R 1.2.0 for windows. To check
the completeness of the program I used the following command:

from c:/rw1020/rw1020
cd src/library
Rcmd check meanscore

(# `meanscore' is the name of a subdirectory contains the package.)

However, I got the following error message (see the last line)

* using log directory C:/rw1020/rw1020/src/meanscore.Rcheck
* checking for file `library/meanscore/DESCRIPTION' ... OK
* checking package directory ... OK
* checking DESCRIPTION Package entry ... OK
* checking DESCRIPTION Version entry ... OK
* checking DESCRIPTION Title entry ... OK
* checking DESCRIPTION License entry ... OK
* checking DESCRIPTION Author entry ... OK
* checking Rd files ... OK
* checking for undocumented objects ... OK
* creating meanscore-Ex.R ... ERROR

But when I'm using the primitive command `make pkgcheck'
from c:/rw1020/rw1020/src/gnuwin32, the meanscore-Ex.R file
is successfully created and the output is correct.

I've already checked the AUTOEXEC file and all the required paths 
are there. I'm using Perl 5.6.0 build 623 on windows 98 OS. 

The set of rtools I'm using is the one from
http://www.stats.ox.ac.uk/pub/Rtools/tools.zip
(downloaded 09/01/01). And I don't have C/Fortran codes in the package.

Is this something to do with the `check' perlscript at R_HOME/bin
or do I need to edit the Mkrules file?

Any suggestion will be very appreciated,


Agus
------------------------
Agus Salim
PhD student
Dept. of Statistics
University College Cork,
Ireland
Tel: 353-21-4902197
------------------------
-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-
r-help mailing list -- Read http://www.ci.tuwien.ac.at/~hornik/R/R-FAQ.html
Send "info", "help", or "[un]subscribe"
(in the "body", not the subject !)  To: r-help-request at stat.math.ethz.ch
_._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._

  
  


  


      

    

    

      
      

        


  

        

      Peter Dalgaard
    

    

      
      #
    

  


  

      
Agus Salim <a.salim at ucc.ie> writes:

            

      
      
      
    

        
I think the thing to do at this point is to run something like

rterm.exe --vanilla < meanscore-Ex.R

(except that one or both of rterm.exe and meanscore.R probably need to
have a full path). Then watch the output and see where the error
comes from.

  
    

      
      
    

  


  


      

    

    

      
      

        


  

        

      Brian Ripley
    

    

      
      #
    

  


  

      
On 9 Jan 2001, Peter Dalgaard BSA wrote:

        

            

      
      
      
    

        
The error is in creating meanscore-Ex.R.  Agus Salim is asking the list
what he asked me privately a while ago.  I have no idea, and him to debug
the perl on his system.  I got no reply!  The error would appear
to be from Rcmd massage-Examples.sh.

I have used this literally hundreds of times and never seen this.

  
    

      
      
    

  


  


      

    

    

      
      

        


  

        

      Peter Dalgaard
    

    

      
      #
    

  


  

      
Prof Brian D Ripley <ripley at stats.ox.ac.uk> writes:

            

      
      
      
    

        
^^^^^^^^

            

      
      
      
    

        
Argh! Sorry. But the strategy persists: Try doing the relevant steps
manually and get more clues, either in the form of error messages or
in pinpointing which spot in the input files triggers the error.

A "set -x" inserted near the beginning of the massage-Examples script
might be helpful (um, this is Bash, right?).

  
    

      
      
    

  


  


      

    

    

      
      

        


  

        

      Duncan Murdoch
    

    

      
      #
    

  


  

      
On Tue, 9 Jan 2001 18:27:28 +0000 (GMT), Prof Brian D Ripley
<ripley at stats.ox.ac.uk> wrote in message
<Pine.GSO.4.31.0101091821550.10815-100000 at auk.stats>:

            

      
      
      
    

        
I get the same error with "Rcmd CHECK", and not with "make pkgcheck".
How do I debug the perl?.  Is there a "verbose" option that would help
narrow down where the error is happening? 

Duncan Murdoch
-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-
r-help mailing list -- Read http://www.ci.tuwien.ac.at/~hornik/R/R-FAQ.html
Send "info", "help", or "[un]subscribe"
(in the "body", not the subject !)  To: r-help-request at stat.math.ethz.ch
_._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._

  
  


  


      

    

    

      
      

        


  

        

      Brian Ripley
    

    

      
      #
    

  


  

      
On Tue, 9 Jan 2001, Duncan Murdoch wrote:

        

            

      
      
      
    

        
No. You need to add some print statements. It's interpreted code.
However, the log statements give a pretty good idea of progress.

  
    

      
      
    

  


  


      

    

    

      
      

        


  

        

      Agus Salim
    

    

      
      #
    

  


  

      
At 19:51 09/01/01 +0100, Peter Dalgaard BSA wrote:

            

      
      
      
    

        
Thanks to Prof. Dalgaard and Prof. Ripley. Honestly, I'm new to Perl
so I couldn't check the script file to locate the problem. However,
I just managed to get an introductory book on Perl, I'll see what
I can learn from it.


Thanks a lot,

Agus
Agus Salim
PhD student
Dept. of Statistics
University College Cork
Ireland
Tel:353-21-4902197


-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-
r-help mailing list -- Read http://www.ci.tuwien.ac.at/~hornik/R/R-FAQ.html
Send "info", "help", or "[un]subscribe"
(in the "body", not the subject !)  To: r-help-request at stat.math.ethz.ch
_._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._

  
  


  


      

    

    

      
      

        


  

        

      Brian Ripley
    

    

      
      #
    

  


  

      
Thanks to some debugging by Duncan Murdoch, it looks like this is a Perl
problem on Win98/ME at least.  As far as I can tell Perl's system command
calls command.com on that system, and at times that gets caught out by
line lengths and redirections.

So please let me know which version of Windows you have.

It's pretty unpredictable, and this did work on my Win98 mini-notebook
with a very small example (ash) but the undoc check failed there.
I was able to get it all to work by changing shell using COMSPEC, but
I cannot distribute the shell I did use.  Anyone know of a reliable
shell foo.exe for Win98 that uses `foo.exe /c commands...' ?  Otherwise
I can write a wrapper for sh.  Actually, I might be able to avoid the
shell here by using a list argument for system....

You might begin to see why Guido and I changed to NT about 2 years' ago.

        
On Wed, 10 Jan 2001, Agus Salim wrote:

        

            

      
      
      
    

        

  
    

      
      
    

  


  


      

    

    

      
      

        


  

        

      Duncan Murdoch
    

    

      
      #
    

  


  

      
On Wed, 10 Jan 2001 17:49:01 +0000 (GMT), you wrote in message
<Pine.GSO.4.31.0101101733200.8914-100000 at toucan.stats>:

            

      
      
      
    

        
I've been reading up on Perl, and a workaround that looks possible is
to open a file and print `some command` into it, instead of calling
system("some command > file").  You lose the possibility of seeing an
error code.

Duncan Murdoch
-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-
r-help mailing list -- Read http://www.ci.tuwien.ac.at/~hornik/R/R-FAQ.html
Send "info", "help", or "[un]subscribe"
(in the "body", not the subject !)  To: r-help-request at stat.math.ethz.ch
_._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._

  
  


  


      

    

    

      
      

        


  

        

      M. Edward (Ed) Borasky
    

    

      
      #
    

  


  

      
Perl on Win95/Win98/WinME is limited by the capabilities of the command-line
editor. There are applications that flat-out can't use it; check the Perl
Data Language (PDL) pages for the gory details. I have personally given up
on Windows ME, though for different reasons: the worst thing that happened
to me was the Visual Studio blue-screened when trying to do post-build
commands. This is probably something to do with the lame command line
interpreter. I went out the next day and spent $330 -- $300 for Windows 2000
Professional and $30 for Red Hat Linux. I am now a happy dual-booter :-).

-----Original Message-----
From: owner-r-help at stat.math.ethz.ch
[mailto:owner-r-help at stat.math.ethz.ch]On Behalf Of Prof Brian Ripley
Sent: Wednesday, January 10, 2001 9:49 AM
To: Agus Salim
Cc: r-help at stat.math.ethz.ch
Subject: Re: [R] problem with Rcmd check


Thanks to some debugging by Duncan Murdoch, it looks like this is a Perl
problem on Win98/ME at least.  As far as I can tell Perl's system command
calls command.com on that system, and at times that gets caught out by
line lengths and redirections.

So please let me know which version of Windows you have.

It's pretty unpredictable, and this did work on my Win98 mini-notebook
with a very small example (ash) but the undoc check failed there.
I was able to get it all to work by changing shell using COMSPEC, but
I cannot distribute the shell I did use.  Anyone know of a reliable
shell foo.exe for Win98 that uses `foo.exe /c commands...' ?  Otherwise
I can write a wrapper for sh.  Actually, I might be able to avoid the
shell here by using a list argument for system....

You might begin to see why Guido and I changed to NT about 2 years' ago.

        
On Wed, 10 Jan 2001, Agus Salim wrote:

        

            

      
      
      
    

        
to

            

      
      
      
    

        
debug

            

      
      
      
    

        
--
Brian D. Ripley,                  ripley at stats.ox.ac.uk
Professor of Applied Statistics,  http://www.stats.ox.ac.uk/~ripley/
University of Oxford,             Tel:  +44 1865 272861 (self)
1 South Parks Road,                     +44 1865 272860 (secr)
Oxford OX1 3TG, UK                Fax:  +44 1865 272595

-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.
-.-
r-help mailing list -- Read http://www.ci.tuwien.ac.at/~hornik/R/R-FAQ.html
Send "info", "help", or "[un]subscribe"
(in the "body", not the subject !)  To: r-help-request at stat.math.ethz.ch
_._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._.
_._


-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-
r-help mailing list -- Read http://www.ci.tuwien.ac.at/~hornik/R/R-FAQ.html
Send "info", "help", or "[un]subscribe"
(in the "body", not the subject !)  To: r-help-request at stat.math.ethz.ch
_._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._