[Rd] scan after seek in text files (PR#12640)
Full_Name: Dr. Alex Sheppard
Version: 2.7.1
OS: Linux Debian Lenny
Submission from: (NULL) (79.73.224.62)
After scanning from an open (text) connection, then seeking, then scanning
again, the second scan returns incorrect result. It looks like the first byte
scanned was from the pre-seek file position, then it continues to read from the
post-seek file position.
To reproduce:
#Put 3x3 matrix in a file
> write.matrix(t(matrix(1:9,nrow=3)),file="TEST.txt",sep="\t")
#Open file as text
> fd <- file("TEST.txt",open="rt")
#scan a couple of fields - this looks fine so far
> scan(file=fd,what=double(),n=2)
Read 2 items
[1] 1 2
#seek back to start of file
> seek(con=fd,where=0,origin="start")
[1] 5
#scan fields again - this doesn't work properly
> scan(file=fd,what=double(),n=2)
Read 2 items
[1] 31 2
This happens when either n or nmax arguments are used to control number of
fields read. Problem does not occur when using nlines argument instead. The seek
appears to work ok, as doing readChar(fd,n=1) after the seek operation correctly
returns "1".
Also, if the file is opened as binary, i.e. fd=file("TEST.txt",open="rb") , all
works fine.
__
R-devel@r-project.org mailing list
https://stat.ethz.ch/mailman/listinfo/r-devel
Re: [Rd] coxph function (Survival Library) (PR#12638)
[EMAIL PROTECTED] wrote: Hi, I am pretty sure that the results of Likelihood ratio test are bigger than the real value. They are multiplied for the factor Ln(10), in this way the p-values are smaller than what they should be. The reason is that you apply the formula -2*log(exp(L(0)/exp(L(B)) and not what it should be -2*log(L(0)/L(B)). Please if I am not right, tell me I don't think you are and please don't use the bug report system when you are not sure. Notice that the conventions for the default base in "log" differ between countries (and even between education levels with countries). There's no indication that base 10 logs are used anywhere, so where would a factor of ln(10) creep in? -- O__ Peter Dalgaard Øster Farimagsgade 5, Entr.B c/ /'_ --- Dept. of Biostatistics PO Box 2099, 1014 Cph. K (*) \(*) -- University of Copenhagen Denmark Ph: (+45) 35327918 ~~ - ([EMAIL PROTECTED]) FAX: (+45) 35327907 __ R-devel@r-project.org mailing list https://stat.ethz.ch/mailman/listinfo/r-devel
Re: [Rd] locale2charset CPU overhead (PR#12633)
On 8/28/2008 3:25 PM, [EMAIL PROTECTED] wrote:
Full_Name: R User
Version: 2.7.2 (default binary)
OS: Windows XP SP2
Submission from: (NULL) (83.25.29.163)
Hello!
I have conducted following experiment running simple R script:
for (i in 1:1) {
a<-0
for (j in 1:1000)
a<-c(a,j)
}
profiled R.dll using Intel Vtune and obtained following results!!!
Func. Name Clocticks
locale2charset 46,58%
R_RunWeakRefFinalizer 20,85%
Rf_eval 8,77%
Rf_eval takes only about 9% of CPU time, a huge overhead on checking
locale character set!!! It seems that is the problem with family of
functions: c(), rbind(), cbind()
I don't have Vtune, so I don't know that its clock counts are reliable.
Could you please identify where the calls to locale2charset are coming
from? I don't see anything obvious in the source that would explain
this, so I suspect Vtune is misreporting.
If I were to guess, I would say your loop would be spending most of its
time in memory management functions, since you allocate and reallocate
"a" so many times. Since those aren't even mentioned by Vtune, I think
it is confused.
Duncan Murdoch
__
R-devel@r-project.org mailing list
https://stat.ethz.ch/mailman/listinfo/r-devel
Re: [Rd] serialize() to via temporary file is heaps faster than doing it directly (on Windows)
I just want to re-post this thread in case it slipped through the
"summer sieve" of someone that might be interested and/or has a real
solution beyond my serialize2() patch.
Cheers
Henrik
On Thu, Jul 24, 2008 at 8:10 PM, Henrik Bengtsson <[EMAIL PROTECTED]> wrote:
> Hi,
>
> FYI, I just notice that on Windows (but not Linux) it is orders of
> magnitude (below it's 50x) faster to serialize() and object to a
> temporary file and then read it back, than to serialize to an object
> directly. This has for instance impact on how fast digest::digest()
> can provide a checksum.
>
> Example:
> x <- 1:1e7;
> t1 <- system.time(raw1 <- serialize(x, connection=NULL));
> print(t1);
> #user system elapsed
> # 174.23 129.35 304.70 ## 5 minutes
> t2 <- system.time(raw2 <- serialize2(x, connection=NULL));
> print(t2);
> # user system elapsed
> # 2.190.185.72 ## 5 seconds
> print(t1/t2);
> # usersystem elapsed
> # 79.55708 718.6 53.26923
> stopifnot(identical(raw1, raw2));
>
> where serialize2() is serialize():ing to file and reading the results back:
>
> serialize2 <- function(object, connection, ...) {
> if (is.null(connection)) {
># It is faster to serialize to a temporary file and read it back
>pathname <- tempfile();
>con <- file(pathname, open="wb");
>on.exit({
> if (!is.null(con))
>close(con);
> if (file.exists(pathname))
>file.remove(pathname);
>});
>base::serialize(object, connection=con, ...);
>close(con);
>con <- NULL;
>fileSize <- file.info(pathname)$size;
>readBin(pathname, what="raw", n=fileSize);
> } else {
>base::serialize(object, connection=connection, ...);
> }
> } # serialize2()
>
> The above benchmarking was done in a fresh R v2.7.1 session on WinXP Pro:
>
>> sessionInfo()
> R version 2.7.1 Patched (2008-06-27 r46012)
> i386-pc-mingw32
>
> locale:
> LC_COLLATE=English_United States.1252;LC_CTYPE=English_United
> States.1252;LC_MON
> ETARY=English_United States.1252;LC_NUMERIC=C;LC_TIME=English_United
> States.1252
>
> attached base packages:
> [1] stats graphics grDevices utils datasets methods base
>
>
> When I do the same on a Linux machine there is no difference:
>
>> sessionInfo()
> R version 2.7.1 (2008-06-23)
> x86_64-unknown-linux-gnu
>
> locale:
> LC_CTYPE=en_US.UTF-8;LC_NUMERIC=C;LC_TIME=en_US.UTF-8;LC_COLLATE=en_US.UTF-8;LC_MONETARY=C;LC_MESSAGES=en_US.UTF-8;LC_PAPER=en_US.UTF-8;LC_NAME=C;LC_ADDRESS=C;LC_TELEPHONE=C;LC_MEASUREMENT=en_US.UTF-8;LC_IDENTIFICATION=C
>
> attached base packages:
> [1] stats graphics grDevices utils datasets methods base
>
> Is there an obvious reason (and an obvious fix) for this?
>
> Cheers
>
> Henrik
>
__
R-devel@r-project.org mailing list
https://stat.ethz.ch/mailman/listinfo/r-devel
Re: [Rd] Security issue with javareconf script (PR#12636)
On Fri, 2008-08-29 at 20:04 +0200, Peter Dalgaard wrote: > OK, committed. Not the easiest hole to exploit, I'd say (notice that > we > only compile something, not execute it). > > > . > > Oh, sh*! This is not portable! Needs code like INSTALL. Will refix. Sorry about that. I forgot that people cared about *nix that has a supported Java but no mktemp implementation... :) ~spot __ R-devel@r-project.org mailing list https://stat.ethz.ch/mailman/listinfo/r-devel
Re: [Rd] R 2.7.2 for Windows fails to install local zip package (PR#12639)
On Wed, Aug 27, 2008 at 06:54:27AM -0400, Duncan Murdoch wrote: > [EMAIL PROTECTED] wrote: >> R version 2.7.2 for Windows fails to install local (zipped) packages >> with following messages: >> >> >>> utils:::menuInstallLocal() >>> >> updating HTML package descriptions >> Error in .readRDS(pfile) : error reading from connection >> > It works for me. I suspect you have permissions problems. It appears that it may have been a problem in how I was compiling the Windows-version of the package. I'll let you know if the error rears its head again. Claude __ R-devel@r-project.org mailing list https://stat.ethz.ch/mailman/listinfo/r-devel
[Rd] coxph function (Survival Library) (PR#12638)
Hi, I am pretty sure that the results of Likelihood ratio test are bigger than the real value. They are multiplied for the factor Ln(10), in this way the p-values are smaller than what they should be. The reason is that you apply the formula -2*log(exp(L(0)/exp(L(B)) and not what it should be -2*log(L(0)/L(B)). Please if I am not right, tell me Thanks Ana __ R-devel@r-project.org mailing list https://stat.ethz.ch/mailman/listinfo/r-devel
Re: [Rd] R 2.7.2 for Windows fails to install local zip package (PR#12612)
On Wed, Aug 27, 2008 at 06:54:27AM -0400, Duncan Murdoch wrote: > [EMAIL PROTECTED] wrote: >> R version 2.7.2 for Windows fails to install local (zipped) packages >> with following messages: >> >> >>> utils:::menuInstallLocal() >>> >> updating HTML package descriptions >> Error in .readRDS(pfile) : error reading from connection >> > It works for me. I suspect you have permissions problems. It appears that it may have been a problem in how I was compiling the Windows-version of the package. I'll let you know if the error rears its head again. Claude __ R-devel@r-project.org mailing list https://stat.ethz.ch/mailman/listinfo/r-devel
Re: [Rd] Security issue with javareconf script (PR#12636)
[EMAIL PROTECTED] wrote:
Full_Name: Tom Callaway
Version: 2.7.2
OS: Fedora 10 (Linux/x86_64)
Submission from: (NULL) (96.233.67.230)
Recently, Debian identified a security issue with the javareconf script in R. I
confirmed that this is still unfixed in R 2.7.2.
The following patch resolves the issue:
diff -up R-2.7.2/src/scripts/javareconf.BAD R-2.7.1/src/scripts/javareconf
--- R-2.7.2/src/scripts/javareconf.BAD 2008-08-29 11:04:21.0 -0400
+++ R-2.7.2/src/scripts/javareconf 2008-08-29 11:05:34.0 -0400
@@ -125,16 +125,17 @@ fi
javac_works='not present'
if test -n "$JAVAC"; then
javac_works='not functional'
-rm -rf /tmp/A.java /tmp/A.class
-echo "public class A { }" > /tmp/A.java
-if test -e /tmp/A.java; then
- if "${JAVAC}" /tmp/A.java >/dev/null; then
- if test -e /tmp/A.class; then
+tempdir=`mktemp -d`
+echo "public class A { }" > ${tempdir}/A.java
+if test -e ${tempdir}/A.java; then
+ if "${JAVAC}" ${tempdir}/A.java >/dev/null; then
+ if test -e ${tempdir}/A.class; then
javac_works=yes
fi
fi
fi
-rm -rf /tmp/A.java /tmp/A.class
+rm -rf ${tempdir}
+
fi
if test "${javac_works}" = yes; then
echo "Java compiler: ${JAVAC}"
OK, committed. Not the easiest hole to exploit, I'd say (notice that we
only compile something, not execute it).
.
Oh, sh*! This is not portable! Needs code like INSTALL. Will refix.
--
O__ Peter Dalgaard Øster Farimagsgade 5, Entr.B
c/ /'_ --- Dept. of Biostatistics PO Box 2099, 1014 Cph. K
(*) \(*) -- University of Copenhagen Denmark Ph: (+45) 35327918
~~ - ([EMAIL PROTECTED]) FAX: (+45) 35327907
__
R-devel@r-project.org mailing list
https://stat.ethz.ch/mailman/listinfo/r-devel
Re: [Rd] Security issue with javareconf script (PR#12636)
On 29 August 2008 at 17:35, [EMAIL PROTECTED] wrote:
| Full_Name: Tom Callaway
| Version: 2.7.2
| OS: Fedora 10 (Linux/x86_64)
| Submission from: (NULL) (96.233.67.230)
|
|
| Recently, Debian identified a security issue with the javareconf script in R.
Yes, somewhat launched a massive list of bug reports against all script using
plain /tmp. The fact that javareconf already rm's the file just before
creation leaves just a tiny tiny tiny window -- but I didn't argue this with
our folks either as the patch (almost like yours) is easy enough.
| I confirmed that this is still unfixed in R 2.7.2.
|
| The following patch resolves the issue:
|
| diff -up R-2.7.2/src/scripts/javareconf.BAD R-2.7.1/src/scripts/javareconf
| --- R-2.7.2/src/scripts/javareconf.BAD 2008-08-29 11:04:21.0 -0400
| +++ R-2.7.2/src/scripts/javareconf2008-08-29 11:05:34.0 -0400
| @@ -125,16 +125,17 @@ fi
| javac_works='not present'
| if test -n "$JAVAC"; then
| javac_works='not functional'
| -rm -rf /tmp/A.java /tmp/A.class
| -echo "public class A { }" > /tmp/A.java
| -if test -e /tmp/A.java; then
| - if "${JAVAC}" /tmp/A.java >/dev/null; then
| - if test -e /tmp/A.class; then
| +tempdir=`mktemp -d`
"mktemp -d -t" is preferable, and that is what out patch does.
I reported this to Simon off-list. The trouble is that such a patch, at the R
source level, would require mktemp to be present on all system which is not a
given. So I suggested to Simon to add a function that will use mktemp where
available and 'does something else' otherwise. I have not heard back yet.
Dirk
| +echo "public class A { }" > ${tempdir}/A.java
| +if test -e ${tempdir}/A.java; then
| + if "${JAVAC}" ${tempdir}/A.java >/dev/null; then
| + if test -e ${tempdir}/A.class; then
| javac_works=yes
| fi
| fi
| fi
| -rm -rf /tmp/A.java /tmp/A.class
| +rm -rf ${tempdir}
| +
| fi
| if test "${javac_works}" = yes; then
| echo "Java compiler: ${JAVAC}"
|
| __
| R-devel@r-project.org mailing list
| https://stat.ethz.ch/mailman/listinfo/r-devel
--
Three out of two people have difficulties with fractions.
__
R-devel@r-project.org mailing list
https://stat.ethz.ch/mailman/listinfo/r-devel
Re: [Rd] specifying compiler path in configure
After reading over the relevant sections of the manual more carefully,
I now see where the information is presented and I was able to get R
to compile with an alternate version of gcc. On thing, however,
remains unclear to me. If R is compiled with a non-default version of
a compiler, are subsequent add-on packages and updated automatically
directed to use the alternate or is it necessary to include this
information in 'install.packages' and 'update.packages'? If the
latter, then it is unclear to me how this is accomplished. Below is my
best attempt at putting this instruction into install.packages and it
failed to produce the intended result, with gcc-4.3 being used instead
of 4.2.
install.packages("affxparser", repos=repos,configure.args =
c('CC=/usr/bin/gcc-4.2', 'CXX=/usr/bin/g++-4.2'))
Thanks, Mark
On Wed, Aug 27, 2008 at 4:38 PM, Prof Brian Ripley
<[EMAIL PROTECTED]> wrote:
> On Wed, 27 Aug 2008, Mark Kimpel wrote:
>
>> I'm having trouble building some packages while running Debian Lenny
>> (testing) and suspect that the issues are related to the default use
>> of gcc-4.3.
>
> You might want to compare your problems with the CRAN checks at
>
> http://cran.r-project.org/web/checks/check_summary.html
>
> as those are run with gcc 4.3.x on Debian testing. From memory, the
> problems are confined to C++-(mis)using packages.
>
>> With Lenny, build-essentials depends on 4.3, so I'd like
>> to leave it installed but have also installed 4.2.1. How do I tell
>> ./configure the path to 4.2.1 ? I"m sure it's an option, but I don't
>> see it documented in the R-admin manual.
>
> From configure --help:
>
> CC C compiler command
> CFLAGS C compiler flags
>
> There are many examples of setting CC in the R-admin manual, and the first
> para of 'Essential Programs' seems to me to tell you that as well.
>
> --
> Brian D. Ripley, [EMAIL PROTECTED]
> 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 272866 (PA)
> Oxford OX1 3TG, UKFax: +44 1865 272595
>
--
Mark W. Kimpel MD ** Neuroinformatics ** Dept. of Psychiatry
Indiana University School of Medicine
15032 Hunter Court, Westfield, IN 46074
(317) 490-5129 Work, & Mobile & VoiceMail
(317) 663-0513 Home (no voice mail please)
__
R-devel@r-project.org mailing list
https://stat.ethz.ch/mailman/listinfo/r-devel
[Rd] Security issue with javareconf script (PR#12636)
Full_Name: Tom Callaway
Version: 2.7.2
OS: Fedora 10 (Linux/x86_64)
Submission from: (NULL) (96.233.67.230)
Recently, Debian identified a security issue with the javareconf script in R. I
confirmed that this is still unfixed in R 2.7.2.
The following patch resolves the issue:
diff -up R-2.7.2/src/scripts/javareconf.BAD R-2.7.1/src/scripts/javareconf
--- R-2.7.2/src/scripts/javareconf.BAD 2008-08-29 11:04:21.0 -0400
+++ R-2.7.2/src/scripts/javareconf 2008-08-29 11:05:34.0 -0400
@@ -125,16 +125,17 @@ fi
javac_works='not present'
if test -n "$JAVAC"; then
javac_works='not functional'
-rm -rf /tmp/A.java /tmp/A.class
-echo "public class A { }" > /tmp/A.java
-if test -e /tmp/A.java; then
- if "${JAVAC}" /tmp/A.java >/dev/null; then
- if test -e /tmp/A.class; then
+tempdir=`mktemp -d`
+echo "public class A { }" > ${tempdir}/A.java
+if test -e ${tempdir}/A.java; then
+ if "${JAVAC}" ${tempdir}/A.java >/dev/null; then
+ if test -e ${tempdir}/A.class; then
javac_works=yes
fi
fi
fi
-rm -rf /tmp/A.java /tmp/A.class
+rm -rf ${tempdir}
+
fi
if test "${javac_works}" = yes; then
echo "Java compiler: ${JAVAC}"
__
R-devel@r-project.org mailing list
https://stat.ethz.ch/mailman/listinfo/r-devel
Re: [Rd] Defining environments within functions
If you want to use this pattern repeatedly you can define something like
makeFunWithCounter <- function(fun) {
counter <- 0
list(count = function() counter,
fun = function(...) { counter <<- counter + 1; fun(...)})
}
and then do
> fwc <- makeFunWithCounter(function() print("Hello"))
> f2 <- fwc$fun
> f2()
[1] "Hello"
> f2()
[1] "Hello"
> f2()
[1] "Hello"
> fwc$count()
[1] 3
If you only want to do it once you can use local,
fwc <- local({
counter <- 0
fun <- function() { counter <<- counter + 1; print("Hello") }
list(count = function() counter, fun = fun)
})
Best,
luke
On Fri, 29 Aug 2008, Giles Hooker wrote:
Thanks,
I think I over-emphasized the secondary function, but I can generate the
scoping problem as follows. First, at the command line, I can get a function
to access objects that were not in its arguments by
ProfileEnv = new.env()
hello.world = "Hello World"
assign('hello.world',hello.world,3,envir=ProfileEnv)
fn1 = function()
{
hw = get('hello.world',envir=ProfileEnv)
print(hw)
}
and then call
fn1()
[1] "Hello World"
Now I want to define a wrapper function
fn2 = function()
{
ProfileEnv = new.env()
hello.world = "Hello World"
assign('hello.world',hello.world,3,envir=ProfileEnv)
fn1()
}
and if I try
rm(ProfileEnv) # Just to be safe
rm(hello.world)
fn2()
Error in get("hello.world", envir = ProfileEnv) :
object "ProfileEnv" not found
In my actual code, fn1() is really a call to
optim(pars,ProfileErr,)
and hello.world are quantities that were calculated the last time that
ProfileErr was called and that I want to keep track of.
As an alternative simple example, how would I keep a counter for the number
of times that optim (or any other generic optimizer) has called ProfileErr?
giles
How can I define environments within a function so that they are visible
to calls to a sub-function?
I think you need to give a simplified, runnable example. (Or at least
runnable until it hits the scoping problem you've got.) "Sub-function"
isn't R terminology, and it's not clear what you mean by it.
--
Luke Tierney
Chair, Statistics and Actuarial Science
Ralph E. Wareham Professor of Mathematical Sciences
University of Iowa Phone: 319-335-3386
Department of Statistics andFax: 319-335-3017
Actuarial Science
241 Schaeffer Hall email: [EMAIL PROTECTED]
Iowa City, IA 52242 WWW: http://www.stat.uiowa.edu
__
R-devel@r-project.org mailing list
https://stat.ethz.ch/mailman/listinfo/r-devel
Re: [Rd] non user-friendly error for chol2inv functions
> "TH" == Ted Harding <[EMAIL PROTECTED]>
> on Fri, 29 Aug 2008 14:21:05 +0100 (BST) writes:
TH> On 29-Aug-08 13:00:01, Martin Maechler wrote:
>>> "cd" == christophe dutang <[EMAIL PROTECTED]>
>>> on Fri, 29 Aug 2008 14:28:42 +0200 writes:
>>
cd> Yes, I do not cast the first argument as a matrix with
cd> as.matrix function.
cd> Maybe we could detail the error message if the first
cd> argument
cd> is a numeric?
>>
cd> error(_("'a' is a numeric and must be coerced to a numeric
cd> matrix"));
>>
>> Merci, Christophe. Yes, we *could* do that.
>> Alternatively, I think I will just make it work in that case,
>> since I see that
>> qr(), chol(), svd(), solve() all
>> treat a numeric (of length 1) as a 1 x 1 matrix automatically.
TH> I was about to draw attention to this "inconsistency"!
TH> While one is about it, might it not be useful also to do
TH> the converse: Treat a 1x1 matrix as a scalar in appropriate
TH> contexts?
TH> E.g.
TH> a <- 4
TH> A <- matrix(4,1,1)
TH> B <- matrix(c(1,2,3,4),2,2)
TH> a*B
TH> # [,1] [,2]
TH> # [1,]4 12
TH> # [2,]8 16
TH> a+B
TH> # [,1] [,2]
TH> # [1,]57
TH> # [2,]68
TH> A*B
TH> # Error in A * B : non-conformable arrays
TH> A+B
TH> # Error in A + B : non-conformable arrays
TH> Ted.
I can only speak for myself and my own taste,
but I'd clearly not want that the above would just work.
I could agree of having it work *WITH A WARNING* ...
particularly if there were more good arguments along your line
of thought..
Martin
>>
cd> Thanks for your answer
>> De rien!
>> Martin
>>
cd> 2008/8/29 Martin Maechler <[EMAIL PROTECTED]>
>>
>> >> > "cd" == christophe dutang <[EMAIL PROTECTED]>
>> >> > on Fri, 29 Aug 2008 12:44:18 +0200 writes:
>> >>
cd> Hi,
cd> In function chol2inv with the option LINPACK set to false
>> (default),
>> >> it
cd> raises an error when the matrix is 1x1 matrix (i.e. just a
>> real)
>> >> saying
>> >>
cd> 'a' must be a numeric matrix
>> >>
>> >> It is very helpful, but you have to read and understand it.
>> >> I'm pretty sure you did not provide a 1 x 1 matrix.
>> >>
>> >> Here's an example showing how things works :
>> >>
>> >> > m <- matrix(4,1,1)
>> >> > cm <- chol(m)
>> >> > cm
>> >> [,1]
>> >> [1,]2
>> >> > chol2inv(cm)
>> >> [,1]
>> >> [1,] 0.25
>> >> >
>> >>
>> >> Martin Maechler, ETH Zurich
>> >>
>> >>
cd> This error is raised by the underlying C function
>> (modLa_chol2inv in
cd> function Lapack.c). Everything is normal, but I wonder if we
>> could
>> >> have
cd> another behavior when we pass a 1x1 matrix. I spent time this
>> >> morning
cd> finding where was the error, and it was this "problem".
>> >>
cd> Thanks in advance
>> >>
cd> Christophe
>> >>
cd> [[alternative HTML version deleted]]
>> >>
cd> __
cd> R-devel@r-project.org mailing list
cd> https://stat.ethz.ch/mailman/listinfo/r-devel
>> >>
>>
cd> [[alternative HTML version deleted]]
>>
cd> __
cd> R-devel@r-project.org mailing list
cd> https://stat.ethz.ch/mailman/listinfo/r-devel
>>
>> __
>> R-devel@r-project.org mailing list
>> https://stat.ethz.ch/mailman/listinfo/r-devel
TH>
TH> E-Mail: (Ted Harding) <[EMAIL PROTECTED]>
TH> Fax-to-email: +44 (0)870 094 0861
TH> Date: 29-Aug-08 Time: 14:16:03
TH> -- XFMail --
TH> __
TH> R-devel@r-project.org mailing list
TH> https://stat.ethz.ch/mailman/listinfo/r-devel
__
R-devel@r-project.org mailing list
https://stat.ethz.ch/mailman/listinfo/r-devel
Re: [Rd] non user-friendly error for chol2inv functions
On 29-Aug-08 13:00:01, Martin Maechler wrote:
>> "cd" == christophe dutang <[EMAIL PROTECTED]>
>> on Fri, 29 Aug 2008 14:28:42 +0200 writes:
>
> cd> Yes, I do not cast the first argument as a matrix with
> cd> as.matrix function.
> cd> Maybe we could detail the error message if the first
> cd> argument
> cd> is a numeric?
>
> cd> error(_("'a' is a numeric and must be coerced to a numeric
> cd> matrix"));
>
> Merci, Christophe. Yes, we *could* do that.
> Alternatively, I think I will just make it work in that case,
> since I see that
> qr(), chol(), svd(), solve() all
> treat a numeric (of length 1) as a 1 x 1 matrix automatically.
I was about to draw attention to this "inconsistency"!
While one is about it, might it not be useful also to do
the converse: Treat a 1x1 matrix as a scalar in appropriate
contexts?
E.g.
a <- 4
A <- matrix(4,1,1)
B <- matrix(c(1,2,3,4),2,2)
a*B
# [,1] [,2]
# [1,]4 12
# [2,]8 16
a+B
# [,1] [,2]
# [1,]57
# [2,]68
A*B
# Error in A * B : non-conformable arrays
A+B
# Error in A + B : non-conformable arrays
Ted.
>
> cd> Thanks for your answer
> De rien!
> Martin
>
> cd> 2008/8/29 Martin Maechler <[EMAIL PROTECTED]>
>
> >> > "cd" == christophe dutang <[EMAIL PROTECTED]>
> >> > on Fri, 29 Aug 2008 12:44:18 +0200 writes:
> >>
> cd> Hi,
> cd> In function chol2inv with the option LINPACK set to false
> (default),
> >> it
> cd> raises an error when the matrix is 1x1 matrix (i.e. just a
> real)
> >> saying
> >>
> cd> 'a' must be a numeric matrix
> >>
> >> It is very helpful, but you have to read and understand it.
> >> I'm pretty sure you did not provide a 1 x 1 matrix.
> >>
> >> Here's an example showing how things works :
> >>
> >> > m <- matrix(4,1,1)
> >> > cm <- chol(m)
> >> > cm
> >> [,1]
> >> [1,]2
> >> > chol2inv(cm)
> >> [,1]
> >> [1,] 0.25
> >> >
> >>
> >> Martin Maechler, ETH Zurich
> >>
> >>
> cd> This error is raised by the underlying C function
> (modLa_chol2inv in
> cd> function Lapack.c). Everything is normal, but I wonder if we
> could
> >> have
> cd> another behavior when we pass a 1x1 matrix. I spent time this
> >> morning
> cd> finding where was the error, and it was this "problem".
> >>
> cd> Thanks in advance
> >>
> cd> Christophe
> >>
> cd> [[alternative HTML version deleted]]
> >>
> cd> __
> cd> R-devel@r-project.org mailing list
> cd> https://stat.ethz.ch/mailman/listinfo/r-devel
> >>
>
> cd> [[alternative HTML version deleted]]
>
> cd> __
> cd> R-devel@r-project.org mailing list
> cd> https://stat.ethz.ch/mailman/listinfo/r-devel
>
> __
> R-devel@r-project.org mailing list
> https://stat.ethz.ch/mailman/listinfo/r-devel
E-Mail: (Ted Harding) <[EMAIL PROTECTED]>
Fax-to-email: +44 (0)870 094 0861
Date: 29-Aug-08 Time: 14:16:03
-- XFMail --
__
R-devel@r-project.org mailing list
https://stat.ethz.ch/mailman/listinfo/r-devel
Re: [Rd] Defining environments within functions
Thanks,
I think I over-emphasized the secondary function, but I can generate the
scoping problem as follows. First, at the command line, I can get a
function to access objects that were not in its arguments by
ProfileEnv = new.env()
hello.world = "Hello World"
assign('hello.world',hello.world,3,envir=ProfileEnv)
fn1 = function()
{
hw = get('hello.world',envir=ProfileEnv)
print(hw)
}
and then call
> fn1()
[1] "Hello World"
Now I want to define a wrapper function
fn2 = function()
{
ProfileEnv = new.env()
hello.world = "Hello World"
assign('hello.world',hello.world,3,envir=ProfileEnv)
fn1()
}
and if I try
> rm(ProfileEnv) # Just to be safe
> rm(hello.world)
> fn2()
Error in get("hello.world", envir = ProfileEnv) :
object "ProfileEnv" not found
In my actual code, fn1() is really a call to
optim(pars,ProfileErr,)
and hello.world are quantities that were calculated the last time that
ProfileErr was called and that I want to keep track of.
As an alternative simple example, how would I keep a counter for the
number of times that optim (or any other generic optimizer) has called
ProfileErr?
giles
How can I define environments within a function so that they are visible
to calls to a sub-function?
I think you need to give a simplified, runnable example. (Or at least
runnable until it hits the scoping problem you've got.) "Sub-function"
isn't R terminology, and it's not clear what you mean by it.
--
Giles Hooker
Assistant Professor:
Department of Biological Statistics and Computational Biology
Department of Statistical Science
1186 Comstock Hall
Cornell University
Ithaca, NY, 14853
Ph: (+1 607) 255 1638
Fax: (+1 607) 255 4698
Email: [EMAIL PROTECTED]
__
R-devel@r-project.org mailing list
https://stat.ethz.ch/mailman/listinfo/r-devel
Re: [Rd] non user-friendly error for chol2inv functions
> "cd" == christophe dutang <[EMAIL PROTECTED]>
> on Fri, 29 Aug 2008 14:28:42 +0200 writes:
cd> Yes, I do not cast the first argument as a matrix with as.matrix
function.
cd> Maybe we could detail the error message if the first argument is a
numeric?
cd> error(_("'a' is a numeric and must be coerced to a numeric matrix"));
Merci, Christophe. Yes, we *could* do that.
Alternatively, I think I will just make it work in that case,
since I see that
qr(), chol(), svd(), solve() all
treat a numeric (of length 1) as a 1 x 1 matrix automatically.
cd> Thanks for your answer
De rien!
Martin
cd> 2008/8/29 Martin Maechler <[EMAIL PROTECTED]>
>> > "cd" == christophe dutang <[EMAIL PROTECTED]>
>> > on Fri, 29 Aug 2008 12:44:18 +0200 writes:
>>
cd> Hi,
cd> In function chol2inv with the option LINPACK set to false (default),
>> it
cd> raises an error when the matrix is 1x1 matrix (i.e. just a real)
>> saying
>>
cd> 'a' must be a numeric matrix
>>
>> It is very helpful, but you have to read and understand it.
>> I'm pretty sure you did not provide a 1 x 1 matrix.
>>
>> Here's an example showing how things works :
>>
>> > m <- matrix(4,1,1)
>> > cm <- chol(m)
>> > cm
>> [,1]
>> [1,]2
>> > chol2inv(cm)
>> [,1]
>> [1,] 0.25
>> >
>>
>> Martin Maechler, ETH Zurich
>>
>>
cd> This error is raised by the underlying C function (modLa_chol2inv in
cd> function Lapack.c). Everything is normal, but I wonder if we could
>> have
cd> another behavior when we pass a 1x1 matrix. I spent time this
>> morning
cd> finding where was the error, and it was this "problem".
>>
cd> Thanks in advance
>>
cd> Christophe
>>
cd> [[alternative HTML version deleted]]
>>
cd> __
cd> R-devel@r-project.org mailing list
cd> https://stat.ethz.ch/mailman/listinfo/r-devel
>>
cd> [[alternative HTML version deleted]]
cd> __
cd> R-devel@r-project.org mailing list
cd> https://stat.ethz.ch/mailman/listinfo/r-devel
__
R-devel@r-project.org mailing list
https://stat.ethz.ch/mailman/listinfo/r-devel
Re: [Rd] non user-friendly error for chol2inv functions
Yes, I do not cast the first argument as a matrix with as.matrix function.
Maybe we could detail the error message if the first argument is a numeric?
error(_("'a' is a numeric and must be coerced to a numeric matrix"));
Thanks for your answer
2008/8/29 Martin Maechler <[EMAIL PROTECTED]>
> > "cd" == christophe dutang <[EMAIL PROTECTED]>
> > on Fri, 29 Aug 2008 12:44:18 +0200 writes:
>
>cd> Hi,
>cd> In function chol2inv with the option LINPACK set to false (default),
> it
>cd> raises an error when the matrix is 1x1 matrix (i.e. just a real)
> saying
>
>cd> 'a' must be a numeric matrix
>
> It is very helpful, but you have to read and understand it.
> I'm pretty sure you did not provide a 1 x 1 matrix.
>
> Here's an example showing how things works :
>
> > m <- matrix(4,1,1)
> > cm <- chol(m)
> > cm
> [,1]
> [1,]2
> > chol2inv(cm)
> [,1]
> [1,] 0.25
> >
>
> Martin Maechler, ETH Zurich
>
>
>cd> This error is raised by the underlying C function (modLa_chol2inv in
>cd> function Lapack.c). Everything is normal, but I wonder if we could
> have
>cd> another behavior when we pass a 1x1 matrix. I spent time this
> morning
>cd> finding where was the error, and it was this "problem".
>
>cd> Thanks in advance
>
>cd> Christophe
>
>cd> [[alternative HTML version deleted]]
>
>cd> __
>cd> R-devel@r-project.org mailing list
>cd> https://stat.ethz.ch/mailman/listinfo/r-devel
>
[[alternative HTML version deleted]]
__
R-devel@r-project.org mailing list
https://stat.ethz.ch/mailman/listinfo/r-devel
Re: [Rd] non user-friendly error for chol2inv functions
> "cd" == christophe dutang <[EMAIL PROTECTED]> > on Fri, 29 Aug 2008 12:44:18 +0200 writes: cd> Hi, cd> In function chol2inv with the option LINPACK set to false (default), it cd> raises an error when the matrix is 1x1 matrix (i.e. just a real) saying cd> 'a' must be a numeric matrix It is very helpful, but you have to read and understand it. I'm pretty sure you did not provide a 1 x 1 matrix. Here's an example showing how things works : > m <- matrix(4,1,1) > cm <- chol(m) > cm [,1] [1,]2 > chol2inv(cm) [,1] [1,] 0.25 > Martin Maechler, ETH Zurich cd> This error is raised by the underlying C function (modLa_chol2inv in cd> function Lapack.c). Everything is normal, but I wonder if we could have cd> another behavior when we pass a 1x1 matrix. I spent time this morning cd> finding where was the error, and it was this "problem". cd> Thanks in advance cd> Christophe cd> [[alternative HTML version deleted]] cd> __ cd> R-devel@r-project.org mailing list cd> https://stat.ethz.ch/mailman/listinfo/r-devel __ R-devel@r-project.org mailing list https://stat.ethz.ch/mailman/listinfo/r-devel
Re: [Rd] Defining environments within functions
On 29/08/2008 6:52 AM, Giles Hooker wrote:
Thanks,
I think I over-emphasized the secondary function, but I can generate the
scoping problem as follows. First, at the command line, I can get a
function to access objects that were not in its arguments by
ProfileEnv = new.env()
hello.world = "Hello World"
assign('hello.world',hello.world,3,envir=ProfileEnv)
fn1 = function()
{
hw = get('hello.world',envir=ProfileEnv)
print(hw)
}
and then call
> fn1()
[1] "Hello World"
Now I want to define a wrapper function
fn2 = function()
{
ProfileEnv = new.env()
hello.world = "Hello World"
assign('hello.world',hello.world,3,envir=ProfileEnv)
fn1()
}
and if I try
> rm(ProfileEnv) # Just to be safe
> rm(hello.world)
> fn2()
Error in get("hello.world", envir = ProfileEnv) :
object "ProfileEnv" not found
In fn2, you have a local variable called ProfileEnv. You defined fn1 in
the global environment, so it can't see it. Everything would work if
you defined fn1 within fn2.
A more natural way to do this is not to explicitly use environments.
Just use the natural lexical scoping in R. In the example below, I've
also added a counter variable, to answer your other question.
> buildfn1 <- function() {
+ hello.world <- "Hello World"
+ counter <- 0
+
+ fn1 <- function() {
+ print(hello.world)
+ counter <<- counter + 1 # note superassignment <<-
+ print(counter)
+ }
+
+ return(fn1)
+ }
>
> fn1 <- buildfn1()
> fn1()
[1] "Hello World"
[1] 1
> fn1()
[1] "Hello World"
[1] 2
Duncan Murdoch
In my actual code, fn1() is really a call to
optim(pars,ProfileErr,)
and hello.world are quantities that were calculated the last time that
ProfileErr was called and that I want to keep track of.
As an alternative simple example, how would I keep a counter for the
number of times that optim (or any other generic optimizer) has called
ProfileErr?
giles
How can I define environments within a function so that they are visible
to calls to a sub-function?
I think you need to give a simplified, runnable example. (Or at least
runnable until it hits the scoping problem you've got.) "Sub-function"
isn't R terminology, and it's not clear what you mean by it.
__
R-devel@r-project.org mailing list
https://stat.ethz.ch/mailman/listinfo/r-devel
[Rd] non user-friendly error for chol2inv functions
Hi, In function chol2inv with the option LINPACK set to false (default), it raises an error when the matrix is 1x1 matrix (i.e. just a real) saying 'a' must be a numeric matrix This error is raised by the underlying C function (modLa_chol2inv in function Lapack.c). Everything is normal, but I wonder if we could have another behavior when we pass a 1x1 matrix. I spent time this morning finding where was the error, and it was this "problem". Thanks in advance Christophe [[alternative HTML version deleted]] __ R-devel@r-project.org mailing list https://stat.ethz.ch/mailman/listinfo/r-devel
Re: [Rd] Defining environments within functions
Matt Calder wrote:
Duncan,
Sorry to be, well, nit picking, but I can't get your example to run. I
am a longtime Splus user and am only using R on the side. The concept of
environments is one I have yet to grasp. I think the example you are trying
to provide will illuminate things for me, but it doesn't source. The
definition of fnBuilder sources fine, though I am not sure what the
"commonArgs <- ..." is meant to do. Anyway, calling:
That wasn't R code, it was pseudocode. That's why I asked the original
poster for a real example: pseudocode examples aren't useful. But I
shouldn't have posted one myself, I should just have stopped at the
first paragraph.
Duncan Murdoch
both <- fnBuilder(...)
Error: '...' used in an incorrect context
and so I though, maybe you meant:
both <- fnBuilder(c(1,2,3))
Error in fnBuilder(c(1, 2, 3)) : '...' used in an incorrect context
But then I thought you were perhaps using "..." as a placeholder not as
syntax and so I have little idea what it is you are trying to show. Could
you clarify what your example is demonstrating in regards to scoping?
So far, I have found the only part of R my Splus knowledge doesn't
translate is in regard to this sort of thing. I can't tell you how much pain
translating "substitute" idioms into R has caused me. Thanks,
Matt
- Original Message -
From: "Duncan Murdoch" <[EMAIL PROTECTED]>
To: <[EMAIL PROTECTED]>
Cc:
Sent: Thursday, August 28, 2008 5:42 AM
Subject: Re: [Rd] Defining environments within functions
Giles Hooker wrote:
How can I define environments within a function so that they are visible
to calls to a sub-function?
I think you need to give a simplified, runnable example. (Or at least
runnable until it hits the scoping problem you've got.) "Sub-function"
isn't R terminology, and it's not clear what you mean by it.
In R, you rarely need to work with environments explicitly. You just
define functions in the same location and they share the same environment.
For example,
fnBuilder <- function(commonArgs) {
commonVars <- ...
ProfileErr <- function(params, ...) {}
coefs <- function(params, ...) {}
return(list(ProfileErr, coefs))
}
both <- fnBuilder(...)
ProfileErr <- both[[1]]
coefs <- both[[2]]
Now ProfileErr and coefs share the same environment, and both can see (and
modify) commonArgs and commonVars.
Duncan Murdoch
I have defined an objective function,
ProfileErr = function(params,...)
which I would like to optimize using standard routines (optim,
nlminb,) but which contains auxiliary variables which need to be
updated along with params. No optimization routine in R that I have
found has facilities for this.
Specifically, within ProfileErr, I need to calculate
coefs(params,...)
This a function which requires a further optimization, and I can achieve
significant efficiency gains by starting where the last optimization
ended, so I would like to keep track of it.
At the command line, I get around this by
ProfileEnv = new.env()
assign('coefs',coefs,3,ProfileEnv)
and within ProfileErr, I can call
startcoefs = get('coefs',envir=ProfileEnv)
* do the optimization to get newcoefs *
assign('coefs',newcoefs,3,ProfileEnv)
Then calling
optim(pars,ProfileErr,)
works fine. However, when I try to wrap all of that in its own function
profile.estimate = fn(pars,...){
ProfileEnv = new.env()
assign('coefs',coefs,3,ProfileEnv)
res = optim(pars,ProfileErr,)
}
ProfileErr no longer sees ProfileEnv. I haven't been able to make much
sense out of the documentation on environments, but is there a way to
make this work? Otherwise I'm back to writing variables out to files.
Many thanks,
Giles
__
R-devel@r-project.org mailing list
https://stat.ethz.ch/mailman/listinfo/r-devel
__
R-devel@r-project.org mailing list
https://stat.ethz.ch/mailman/listinfo/r-devel
