It is probably important to note that the WRE with the new section on C API
compliance is
in the R-Devel docs, not the current ones.
JN
On 2024-06-25 12:10, luke-tierney--- via R-devel wrote:
On Tue, 25 Jun 2024, Josiah Parry wrote:
Hey folks,
I'm sure many of you all woke to the same messa
On 2024-06-25 12:25, Josiah Parry wrote:
"Stop using them" is pithy advice but far easier said than done!
With respect to NOTES and WARN on CRAN, these do not result in any
package maintainer notifications. The only notification that the developers
receive is the threatening one that states t
t, getting the indicated result of 0 for (sum(vv1) - 1.0e0), with
non-zero on my
Ryzen 7 laptop.
JN
# FPExtendedTest.R J C Nash
loopsum <- function(vec){
n <- length(vec)
vsum<-0.0
for (i in 1:n) { vsum <- vsum + vec[i]}
vsum
}
small<-.Machine$double.eps/4 # 1/4 of the
build R with
--disable-long-double to get closer to the arm64 results if that is your worry.
Cheers,
Simon
On 04/02/2024 4:47 p.m., J C Nash wrote:
Slightly tangential: I had some woes with some vignettes in my
optimx and nlsr packages (actually in examples comparing to OTHER
packages) bec
och
On 04/02/2024 4:47 p.m., J C Nash wrote:
Slightly tangential: I had some woes with some vignettes in my
optimx and nlsr packages (actually in examples comparing to OTHER
packages) because the M? processors don't have 80 bit registers of
the old IEEE 754 arithmetic, so some existing "
Slightly tangential: I had some woes with some vignettes in my
optimx and nlsr packages (actually in examples comparing to OTHER
packages) because the M? processors don't have 80 bit registers of
the old IEEE 754 arithmetic, so some existing "tolerances" are too
small when looking to see if is sma
Slightly tangential, but about two decades ago I was researching
how multimedia databases might be reasonably structured. To have a
concrete test case, I built a database of English Country (Playford)
dances, which I called Playford's Progeny. (Ben B. will be aware of
this, too.) This proved rathe
But why time methods that the author (me!) has been telling the community for
years have updates? Especially as optimx::optimr() uses same syntax as optim()
and gives access to a number of solvers, both production and didactic. This set
of solvers is being improved or added to regularly, with a ma
Better check your definitions of SVD -- there are several forms, but all I
am aware of (and I wrote a couple of the codes in the early 1970s for the SVD)
have positive singular values.
JN
On 2023-07-16 02:01, Durga Prasad G me14d059 wrote:
Respected Development Team,
This is Durga Prasad reac
Thanks Martin.
Following Duncan's advice as well as some textual input, I have put a proposed
Rd file for
optim on a fork of the R code at
https://github.com/nashjc/r/blob/master/src/library/stats/man/optim.Rd
This has the diffs given below from the github master. The suggested changes
primaril
A tangential email discussion with Simon U. has highlighted a long-standing
matter that some tools in the base R distribution are outdated, but that
so many examples and other tools may use them that they cannot be deprecated.
The examples that I am most familiar with concern optimization and non
I wrote first cut at unirootR for Martin M and he revised and put in
Rmpfr.
The following extends Ben's example, but adds the unirootR with trace
output.
c1 <- 4469.822
c2 <- 572.3413
f <- function(x) { c1/x - c2/(1-x) }; uniroot(f, c(1e-6, 1))
uniroot(f, c(1e-6, 1))
library(Rmpfr)
unirootR(f, c
nls() actually uses different modeling formulas depending on the 'algorithm',
and
there is, in my view as a long time nonlinear modeling person, an unfortunate
structural issue that likely cannot be resolved simply. This is because for
nonlinear
modeling programs we really should be using explic
FWIW, nlsr::nlxb() gives same answers.
JN
On 2023-01-25 09:59, Dave Armstrong wrote:
Dear Colleagues,
I recently answered [this question]() on StackOverflow that identified
what seems to be unusual behaviour with `stats:::nls.fitted()`. In
particular, a null model returns a single fitted value
Extreme scaling quite often ruins optimization calculations. If you think
available methods
are capable of doing this, there's a bridge I can sell you in NYC.
I've been trying for some years to develop a good check on scaling so I can
tell users
who provide functions like this to send (lots of)
concept of
inference can be when applied to parameters in such models.
JN
On 2021-08-20 11:35 a.m., Martin Maechler wrote:
>>>>>> J C Nash
>>>>>> on Fri, 20 Aug 2021 11:06:25 -0400 writes:
>
> > In our work on a Google Summer of Code proje
In our work on a Google Summer of Code project "Improvements to nls()",
the code has proved sufficiently entangled that we have found (so far!)
few straightforward changes that would not break legacy behaviour. One
issue that might be fixable is that nls() returns no result if it
encounters some co
In mentoring and participating in a Google Summer of Code project "Improvements
to nls()",
I've not found examples of use of the "subset" argument in the call to nls().
Moreover,
in searching through the source code for the various functions related to
nls(), I can't
seem to find where subset is
This message is to let R developers know that the project in the Subject is now
a Google Summer of Code project.
Our aim in this project is to find simplifications and corrections to the nls()
code, which has become heavily patched. Moreover, it has some deficiencies in
that
there is no Marquardt
One of the mechanisms by which R has been extended and improved has been through
the efforts of students and mentors in the Google Summer of Code initiatives.
This
year Toby Hocking (along with others) has continued to lead this effort.
This year, Google has changed the format somewhat so that th
Sorry, Martin, but I've NOT commented on this matter, unless someone has been
impersonating me.
Someone else?
JN
On 2021-01-11 4:51 a.m., Martin Maechler wrote:
>> Viechtbauer, Wolfgang (SP)
>> on Fri, 8 Jan 2021 13:50:14 + writes:
>
> > Instead of a separate file to sto
Is this a topic for Google Summer of Code? See
https://github.com/rstats-gsoc/gsoc2021/wiki
On 2021-01-09 12:34 p.m., Dirk Eddelbuettel wrote:
>
> The idea of 'white lists' to prevent known (and 'tolerated') issues, note,
> warnings, ... from needlessly reappearing is very powerful and general,
Does this issue fit in the more general one of centralized vs
partitioned checks? I've suggested before that the CRAN team
seems (and I'll be honest and admit I don't have a good knowledge
of how they work) to favour an all-in-one checking, whereas it
might be helpful to developers and also widen t
As with many areas of R usage, my view is that the concern is one
of making it easier to find appropriate information quickly. The
difficulty is that different users have different needs. So if
one wants to know (most of) what is available, the Time Series
Task View is helpful. If one is a novice,
chler wrote:
>>>>>> J C Nash
>>>>>> on Thu, 26 Mar 2020 09:29:53 -0400 writes:
>
> > Given that a number of us are housebound, it might be a good time to
> try to
> > improve the approximation. It's not an area where I have muc
Given that a number of us are housebound, it might be a good time to try to
improve the approximation. It's not an area where I have much expertise, but in
looking at the qbeta.c code I see a lot of root-finding, where I do have some
background. However, I'm very reluctant to work alone on this, an
The main thing is to post the "small reproducible example".
My (rather long term experience) can be written
if (exists("reproducible example") ) {
DeveloperFixHappens()
} else {
NULL
}
JN
On 2019-03-29 11:38 a.m., Saren Tasciyan wrote:
> Well, first I can't sign in bugzilla myse
Rereading my post below, I realize scope for misinterpretation. As I have said
earlier,
I recognize the workload in doing any streamlining, and also the immense
service to us
all by r-core. The issue is how to manage the workload efficiently while
maintaining
and modernizing the capability. That
by folk outside the core.
JN
On 2019-03-04 9:12 a.m., Avraham Adler wrote:
> On Mon, Mar 4, 2019 at 5:01 PM J C Nash <mailto:profjcn...@gmail.com>> wrote:
>
> As the original coder (in mid 1970s) of BFGS, CG and Nelder-Mead in
> optim(), I've
> been pu
As the original coder (in mid 1970s) of BFGS, CG and Nelder-Mead in optim(),
I've
been pushing for some time for their deprecation. They aren't "bad", but we have
better tools, and they are in CRAN packages. Similarly, I believe other
optimization
tools in the core (optim::L-BFGS-B, nlm, nlminb)
10. 0.06783655
> [4] 5.03391827 0.49004503 2.76198165
> [7] 1.09760394 1.92979280 1.34802525
> [10] 1.38677998 1.38628970 1.38635074
> [13] 1.38628970
>
> This will not tell you why the objective function is being called (e.g. in a
> line search
>
In https://cran.r-project.org/bin/linux/ubuntu/
Administration and Maintances of R Packages
^^
Minor stuff, but if someone who can edit is on the page,
perhaps it can be changed to "Maintenance"
Best, JN
__
R-devel@r-project
In looking at rootfinding for the histoRicalg project (see
gitlab.com/nashjc/histoRicalg),
I thought I would check how uniroot() solves some problems. The following short
example
ff <- function(x){ exp(0.5*x) - 2 }
ff(2)
ff(1)
uniroot(ff, 0, 10)
uniroot(ff, c(0, 10), trace=1)
uniroot(ff, c(0, 10
After some thought, I decided r-devel was probably the best of the R lists
for this item. Do feel free to share, as the purpose is to improve documentation
and identify potential issues.
John Nash
The R Consortium has awarded some modest funding for "histoRicalg",
a project to document and tran
Having worked with optim() and related programs for years, it surprised me
that I haven't noticed this before, but optim() is inconsistent in how it
deals with bounds constraints specified at infinity. Here's an example:
# optim-glitch-Ex.R
x0<-c(1,2,3,4)
fnt <- function(x, fscale=10){
yy <- len
a bug, but it is
asking for trouble.
JN
On 15-04-25 07:57 AM, peter dalgaard wrote:
>
>> On 25 Apr 2015, at 13:11 , Prof J C Nash (U30A) wrote:
>>
>> Hendrik pointed out it was the parentheses that gave the complaint.
>> Single quotes and no parentheses seem to satisfy R
.
Is this something CRAN should be thinking about? I would argue greater
benefit to users than title case.
JN
On 15-04-24 06:17 PM, Uwe Ligges wrote:
>
>
> On 24.04.2015 22:44, Ben Bolker wrote:
>> Prof J C Nash (U30A uottawa.ca> writes:
>>
>>>
>>> I
I was preparing a fix for a minor glitch in my optimx package and R CMD
check gave an error that the title was not in title case. It is
A Replacement and Extension of the optim() Function
R CMD check suggests the incorrect form
A Replacement and Extension of the Optim() Function
'Writing R Exte
nls() is using
1) only a Gauss-Newton code which is prone to some glitches
2) approximate derivatives
Package nlmrt uses symbolic derivatives for expressions (you have to
provide Jacobian code for R functions) and an aggressive Marquardt
method to try to reduce the sum of squares. It does return m
//www.ubuntu.com/";
>> SUPPORT_URL="http://help.ubuntu.com/";
>> BUG_REPORT_URL="http://bugs.launchpad.net/ubuntu/";
>>
>> john@john-J6-2015 ~/current/nls14work $ cat /etc/linuxmint/info
>> RELEASE=17.1
>> CODENAME=rebecca
>> EDIT
Are other developers finding R 3.1.3 problematic because vignette
building requires pandoc 1.12.3, while Linux Mint 17 / Ubuntu 14.04 have
1.12.2.1? R 3.1.2 seems to work fine.
I'd very much like to avoid having to build as large a Linux package as
pandoc, which has given me issues outside of R (i
ink I need to do some more digging to
narrow down where this issue is lurking. It may be some local matter, as
with the r-recommended links failing.
Best, JN
On 15-01-18 09:27 AM, Kurt Hornik wrote:
Prof J C Nash (U30A) writes:
I've been implementing a wrapper to the 2011 Fortran ve
I've been implementing a wrapper to the 2011 Fortran version of
L-BFGS-B. In optim(), R uses a C translation of a Fortran version (the
version number does not appear to be documented by the original
authors). The authors of the original Fortran code have updated it and
published the reasons in
As the author of 3 of the 5 methods in optim, I think you may be wasting
your time if this is for performance. My reasons are given in
http://www.jstatsoft.org/v60/i02
Note that most of the speed benefits of compilation are found in the
objective and gradient function, with generally more minor i
I won't comment on the C/C++ option, as I'm not expert in that. However,
R users and developers should know that Nocedal et al. who developed
L-BFGS-B released an update to correct a fault in 2011. It was important
enough that an ACM TOMS article was used for the announcement.
I recently implement
As you dig deeper you will find vmmin.c, cgmin.c and (I think) nmmin.c
etc. Those were, as I understand, converted by p2c from my Pascal codes
that you can find in the pascal library on netlib.org. These can be run
with the Free Pascal compiler.
Given how long ago these were developed (>30 years i
I noted Duncan's comment that an answer had been provided, and went to
the archives to find his earlier comment, which I am fairly sure I saw a
day or two ago. However, neither May nor June archives show Duncan in
the thread except for the msg below (edited for space). Possibly tech
failures are ca
Over the years, this has been useful to me (not just in R) for many
nonlinear optimization tasks. The alternatives often clutter the screen.
> On 13-11-06 06:00 AM, r-devel-requ...@r-project.org wrote:
> People do sometimes use this pattern for displaying progress (e.g. iteration
> counts).
>>
t; cmpfun(tfor)
>> cmpfun(twhile)
>> timforc<-microbenchmark(tfor(n))
>> timwhilec<-microbenchmark(twhile(n))
>> timforc
>> timwhilec
>> looptimes<-data.frame(timfor$time, timforc$time, timwhile$time,
>> timwhilec$time)
>> colMeans(looptimes)
&
ooted
that partition for quite a while.
Here is the for-while test code:
# forwhiletime.R
library(microbenchmark)
require(compiler)
tfor <- function(n){
for (i in 1:n) {
xx<-exp(sin(cos(as.double(i
}
xx
}
twhile <- function(n){
i<-0
while (i On 13-11-03
I had a bunch of examples of byte code compiles in something I was
writing. Changed to 3.0.2 and the advantage of compiler disappears. I've
looked in the NEWS file but do not see anything that suggests that the
compile is now built-in. Possibly I've just happened on a bunch of
examples where it doe
This issue has been known for some time and I've had "why don't you fix
this?" queries. However, I'm not one of the R-core folk who could do so,
and don't code in C. Moreover, as far as I can tell, the version of
L-BFGS-B in R is not one of the standard releases from Morales and Nocedal.
As m
There is quite a literature on related methods for variance. If anyone
is interested, I did some work (and even published the code in the
magazine Interface Age in 1981) on some of these. I could probably put
together scans of relevant materials, some of which are not easily
available. It would
While as a Linux user who has not so far been banished to Winland I have
not experienced this problem, it seems to be the type of issue where a
"how to", for example, on the R Wiki, would be helpful. Moreover, surely
this is a name conflict on different platforms, so possibly a list of
these
The message below came to me from the Getting Open Source Logic INto
Government list. I'm passing it on to the devel list as the infoworld
article may have some ideas of relevance to the R project, mainly
concerning build and test issues and tracking changes in the code base.
While the LibreOff
For info, I put a little study I did about the byte code compiler and
other speedup approaches (but not multicore) on the Rwiki at
http://rwiki.sciviews.org/doku.php?id=tips:rqcasestudy
which looks at a specific problem, so may not be relevant to everyone.
However, one of my reasons for doing i
If CRAN were a passive repository, the discussion about its policies
would not be relevant to this list e.g., SourceForge. However, the
development of R and its packages are very intimately connected to the
CRAN repository policy.
I doubt any of the players in building our current R ecosystem
und in
estimation levels of
factor ", nn[i])
dostop = TRUE
}
}
} # end loop over names
} # end part where we DO have new data
return(dostop) # TRUE if we should stop
}
# end of predict.warn code
Test example:
58 matches
Mail list logo
