Re: [R] issue with Rcmdr
> Message: 14 > Date: Wed, 8 Jan 2020 09:59:56 -0800 > From: Bert Gunter > To: Norm Matloff > Cc: R-help > Subject: Re: [R] issue with Rcmdr > ... and even more generally, is generally misleading. ;-) > (search "problems with R^2" or similar for why). > Bert Gunter I was addressing the OP's issue of computing R^2. Its value in analysis is a separate question. If one's goal is prediction, it arguably is fairly good providing we avoid overfitting. If our goal is inference, R^2 still is informative, but limited. Norm __ R-help@r-project.org mailing list -- To UNSUBSCRIBE and more, see 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.
Re: [R] issue with Rcmdr
Glad to hear it now works for you. But speaking more generally, note that R-squared is the squared correlation between the predicted Y and actual Y values. E.g. lmout <- lm(y ~ x) print(cor(lmout$fitted.values,y)^2) One can use this in any regression setting, even machine learning methods. Norm __ R-help@r-project.org mailing list -- To UNSUBSCRIBE and more, see 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.
[R] book on parallel programming
I'm about halfway finished writing a book titled, Parallel Computation for Data Science: with Examples in R and Beyond, to be published later this year by Chapman and Hall. I'm making a rough draft of the first half of the book available on the Web: http://heather.cs.ucdavis.edu/paralleldatasci.pdf Your comments and suggestions are welcome, in fact very much hoped-for. I took this approach when I wrote my R programming book, and found it very helpful to me, and of much value to people who downloaded it. Hopefully the same will be true here. Norm Matloff __ R-help@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.
[R] novel graphics package
I've developed a new graphics library, BDGraphs, designed for big data, meaning any data large enough to cause major screen clutter and overplotting if the points are plotted individually. The library includes a novel approach to parallel coordinates, and some methods I introduced in my JSM talk last month. I'm posting this message because I'd like to get feedback from people before I put the package on CRAN. If you're interested, please go to http://heather.cs.ucdavis.edu/bdgraphs.html The package is there for downloading, and the Examples page shows several examples with pictures. Norm Matloff __ R-help@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.
Re: [R] Split the work for many cores
One replier worried about doing multiple operations on the graphics device in parallel. To avoid this, try saving the output of your plot function instead of immediately displaying it (if the function allows this). Then display everything when you're all done. Norm Matloff To: Alaios ala...@yahoo.com Subject: Re: Split the work for many cores In-Reply-To: 1336655568.56888.yahoomail...@web120106.mail.ne1.yahoo.com References: 1336655568.56888.yahoomail...@web120106.mail.ne1.yahoo.com In gmane.comp.lang.r.general, you wrote: Dear all, I am using my code the vgram.matrix of packets fields. I have around 500 matrices that I need to pass inside that function and then plot those results. Even though my system has 16 cores is quite clear that I am only using one of those. Would it be able to skip these 500 tasks to the 16 cores, with each processor having around 4 matrices to process? What would you suggest me doing? Regards Alex [[alternative HTML version deleted]] __ R-help@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.
Re: [R] Discrete Event Simulation problem
Unfortunately, I don't have time to read your code, but if it is any help, I have general discrete event simulation code as an example in my book. I've posted the code at http://heather.cs.ucdavis.edu/DES.R Norm Matloff __ R-help@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.
Re: [R] new R debugging tool
Rainer wrote: * On Thu, Oct 20, 2011 at 12:22 AM, Norm Matloff * matl...@cs.ucdavis.eduwrote: * * * I've developed a new R debugging tool, debugR, available at * http://heather.cs.ucdavis.edu/debugR.html * * This basically replaces my edtdbg, which I will no longer be * supporting. * The new tool is now decoupled from one's text editor, and has a lot * more * features than edtdbg did. * * * Sounds interesting. Do I have to write a script file for debugging, or * is * there an option to run it from within R to debug a function instead of a * script? * * Rainer One does indeed basically debug at the function level (or set of functions), as desired. The user specifies on the command line a file in which those functions reside. (In time, I can broaden this.) The best way to see the operation is to try the Quick Start. It really IS quick! Just go to http://heather.cs.ucdavis.edu/debugR.html where I've now included the Quick Start. I hope to write up a longer example for Tal's e-newsletter. Norm __ R-help@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.
[R] new R debugging tool
I've developed a new R debugging tool, debugR, available at http://heather.cs.ucdavis.edu/debugR.html This basically replaces my edtdbg, which I will no longer be supporting. The new tool is now decoupled from one's text editor, and has a lot more features than edtdbg did. Try it! Feedback is encouraged, of course. Norm Matloff __ R-help@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.
Re: [R] running R commands asynchronously
On 06/10/2011 02:29 PM, Richard M. Heiberger wrote: I am interested in running R commands asynchronously... You can do this with my Rdsm package on CRAN. Set up 2 Rdsm clients, which I'll call A and B. Use client A as you main R session, where you do most of your work, but start your asynchronous process in B, writing to memory shared by A and B. The code at B would look like: run task B, writing results to shared variable X[1] when done The code at A would look like: do various unrelated tasks while (X[1] == 0) ; use X[1] Norm Matloff __ R-help@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.
Re: [R] conversion of matrix into list
Martin Spindler wrote: I have a matrix X which consists of 2 columns. I would like to convert this matrix into a list where every entry of the list consists of a single row of the matrix. Here's another way besides split(): # returns a list of the matrix m's rows (rowcol=1) or columns mat2lst - function(m,rowcol=1) { if (rowcol == 1) m - t(m) dm - as.data.frame(m) as.list(dm) } This seems to be faster than the split() approach for columns, but slower for rows. Apparently the transpose operation makes the difference. (You could try investigating via Rprof().) Norm Matloff __ R-help@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.
Re: [R] GUI's and R background processes
Anne, you can accomplish your goal by using my Rdsm package, which adds a threads-like capability to R. You can download it from CRAN. Look in particular in the examples/ directory. The file WebProbe.R is pretty much exactly the same usage that you want. Look at Auction.R too. You may also find my UseR! presentation on Rdsm to be helpful, user2010.org/slides/Matloff.pdf You could do the same thing, though less directly and I believe less conveniently, using some of the packages Louis mentioned, as well as bigmemory. Norm Matloff __ R-help@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.
Re: [R] Significance of the difference between two correlation coefficients
Adaikalavan Ramasamy wrote: Using the info from that website, I can code up the following to give the two-tailed p-value of difference in correlations: diff.corr - function( r1, n1, r2, n2 ){ ... William Revelle also mentioned the r.test in the psych package. I would add here that inference on second-order quantities, such as correlation coefficients and variances, is not robust to the assumption of a normally-distributed population. (Inference on first-order quantities such as means and regression coefficients, IS pretty robust to that assumption.) A good general alternative is the bootstrap, implemented in R in the boot package. Norm Matloff __ R-help@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.
Re: [R] binary tree construction in R
On Tue, Oct 05, 2010 at 06:13:04PM -0400, Mike Marchywka wrote: I guess it wouldn't be too far a field to discuss benefits of data stucture exploration in R vs cpp or java- Especially for something like this where you may want to time it in a multithreaded setting- you can always instrument something like that, collect lots of monte carlo results, and then import the statistical data into R for analysis I would think. This is fairly common--do your data collection in another language, then analyze it in R. R has nice interfaces to other languages, e.g. Python and C, so you can do this directly rather than writing to a file in one language and then reading it from R. For a multithreaded setting, see the various approaches that have been developed, at http://cran.r-project.org/web/views/HighPerformanceComputing.html However, in my opinion, these work well only for embarrassingly parallel applications. For the other apps, probably including yours, you may need to go to C. Norm __ R-help@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.
Re: [R] binary tree construction in R
MK wrote: Hi all, I'm very new to R and I'm trying to construct a threaded binary tree using recursive functions. I'm very confused was wondering if anyone had any R sample code they would share. I've come across a lot of C++ code(nothing in R) and this is not helping. best, MK Not sure what you mean by a threaded binary tree, but I am enclosing code below. It is from my forthcoming book on software development in R. Two caveats: 1. It hasn't been checked yet. There may be bugs, inefficiencies etc. 2. It does search and insert, not delete. Norm Matloff # routines to create trees and insert items into them are included # below; a deletion routine is left to the reader as an exercise # storage is in a matrix, say m, one row per node of the tree; a link i # in the tree means the vector m[i,] = (u,v,w); u and v are the left and # right links, and w is the stored value; null links have the value NA; # the matrix is referred to as the list (m,nxt,inc), where m is the # matrix, nxt is the next empty row to be used, and inc is the number of # rows of expansion to be allocated when the matrix becomes full # initializes a storage matrix, with initial stored value firstval newtree - function(firstval,inc) { m - matrix(rep(NA,inc*3),nrow=inc,ncol=3) m[1,3] - firstval return(list(mat=m,nxt=2,inc=inc)) } # inserts newval into nonempty tree whose head is index hdidx in the # storage space treeloc; note that return value must be reassigned to # tree; inc is as in newtree() above ins - function(hdidx,tree,newval,inc) { tr - tree # check for room to add a new element tr$nxt - tr$nxt + 1 if (tr$nxt nrow(tr$mat)) tr$mat - rbind(tr$mat,matrix(rep(NA,inc*3),nrow=inc,ncol=3)) newidx - tr$nxt # where we'll put the new tree node tr$mat[newidx,3] - newval idx - hdidx # marks our current place in the tree node - tr$mat[idx,] nodeval - node[3] while (TRUE) { # which direction to descend, left or right? if (newval = nodeval) dir - 1 else dir - 2 # descend # null link? if (is.na(node[dir])) { tr$mat[idx,dir] - newidx break } else { idx - node[dir] node - tr$mat[idx,] nodeval - node[3] } } return(tr) } # print sorted tree via inorder traversal printtree - function(hdidx,tree) { left - tree$mat[hdidx,1] if (!is.na(left)) printtree(left,tree) print(tree$mat[hdidx,3]) right - tree$mat[hdidx,2] if (!is.na(right)) printtree(right,tree) } __ R-help@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.
Re: [R] Why is vector assignment in R recreates the entire vector ?
Tal wrote: A friend recently brought to my attention that vector assignment actually recreates the entire vector on which the assignment is performed. ... I brought this up in r-devel a few months ago. You can read my posting, and the various replies, at http://www.mail-archive.com/r-de...@r-project.org/msg20089.html Some of the replies not only explain the process, but list lines in the source code where this takes place, enabling a closer look at how/when duplication occurs. Norm Matloff __ R-help@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.
Re: [R] Question regarding significance of a covariate in a coxme survival
In 2010-08-30, C. Peng button...@hotmail.com wrote: What statistical measure(s) tend to be answering ALL(?) question of practical interest? None. All I had said was that significance testing doesn't really answer any questions of practical interest. Unfortunately, that doesn't mean there's something to answer all such questions. In the regression case brought up by the original poster (that was Cox regression, but the principle is the same), prediction-oriented measures such AIC or cross-validation directly address the question of interest, if that question is predictive ability. In general, one should at the very least form confidence intervals instead of significance tests. In the case in which tests are demanded, e.g. medical journals, replace instead of by in addition to. Norm __ R-help@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.
Re: [R] Question regarding significance of a covariate in a coxme survival
Using a p-value to make any kind of decision is questionable to begin with, and especially unreliable in choosing covariates in regression. Old studies, e.g. by Walls and Weeks and by Bendel and Afifi, have shown that if predictive ability is the criterion of interest and one wishes to use p-values for deciding whether to include a covariate, one should set the p-value bar very large, at 0.25 and even 0.40. By contrast, methods such as AIC are aimed at avoiding overfitting, by penalizing models with large numbers of covariates. Same for Mallows' Cp, cross validation etc. So, the p-value and AIC are answering quite different questions, and thus should not be expected to give the same or even similar results. But, worse than that, many point out that p-values tend not to be answering ANY question of practical interest. It's a shame that the use of p-values is so entrenched. I can expand on this, with references, if there is interest. Norm Matloff Professor of Computer Science (formerly Statistics) University of California, Davis __ R-help@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.
Re: [R] ANNOUNCE--Rdsm package, a threads-like environment for R
Guo-Hao Huang wrote: Date: Thu, 11 Mar 2010 14:55:35 +0800 From: Guo-Hao Huang guohao.hu...@gmail.com To: Norm Matloff matl...@cs.ucdavis.edu Cc: r-help@r-project.org Subject: Re: [R] ANNOUNCE--Rdsm package, a threads-like environment for R I am interested in Rdsm package, but I have no idea about how to use it. Where can I find examples? regards Guo-Hao Huang There are examples there in the package, in the examples/ directory. The file examples/README will tell you what to do in quick-start form. Also, after loading the Rdsm library, type help(Rdsm) at the R interactive prompt, for further information. If you have further questions, feel free to contact me. Norm __ R-help@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.
[R] ANNOUNCE--Rdsm package, a threads-like environment for R
My long-promised Rdsm package is now on CRAN. Some of you may recall that I made a prototype available on my own Web page last July. This is the official version, much evolved since I released the prototype. The CRAN description states: Provides a threads-like programming environment for R, usable both on a multicore machine and across a network of multiple machines. The package gives the illusion of shared memory, again even across multiple machines on a network. This is based on a similar package I wrote for Perl back in 2002. Rdsm should complement the bigmemory package written by Mike Kane and Jay Emerson. As noted above and in my July announcement for my prototype, Rdsm works across a network of machines, while bigmemory is limited to multicore platforms. On the other hand, bigmemory is capable of handling huge data structures, which Rdsm is not. Rdsm is intended for parallel/distributed computing, while bigmemory is currently used primarily for the problems its name implies. However, if bigmemory is used as a parallel R vehicle on a multicore machine, it may be faster than Rdsm for some applications, given its direct use of bare shared memory; typically, though, they should provide similar performance in such contexts. By the way, I have a prototype of an infrastructure package to facilitate using bigmemory as a parallel R engine. Norm Matloff University of California, Davis __ R-help@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.
[R] ANNOUNCE: edtdbg debugging tool
My edtdbg debugging tool is now on CRAN, at http://cran.r-project.org/web/packages/edtdbg/index.html I've added a few new commands since the last time I announced the package here. I'll enclose command list at the end of this message. Currently the package is implemented for Vim. I have a volunteer to adapt it to Emacs, so I'm hopeful it will be usable for Emacs fans later. I'm still looking for time to prepare my Rdsm parallel R package for submission to CRAN. The code is all ready, though, so if anyone wishes to obtain it, just let me know. Norm Here are the edtdbg commands, typed within the editor: * ,dbg Start edtdbg. * ,dbgu Exit ‘edtdbg’. * ,src Tell R to source the file you have in your current Vim buffer. * [ If you are in the browser in your R window, hitting the left-bracket key runs the browser's n command. Your Vim cursor will then automatically move to the new line, i.e. to the one that the browser will now be on. * ] Same as [, but runs the c command instead of n. * ,dsp This toggles display. When display is on, the values of the current function's arguments and local variables will appear in the R window after each debug step. * ,mom This displays the arguments and local variables in the parent function, i.e. the next-higher frame in the call stack. The action occurs once per time you invoke the command. * ,glb This displays all global variables. The action occurs once per time you invoke the command. * ,br This sets a breakpoint at the line after the current cursor line of the editor. This is done by inserting a call to browser() and then re-sourcing the code at R. * ,bru This undoes the last action of ,br thus removing the last breakpoint. If you wish to remove a different breakpoint, you need to do that manually. * ,dbga This tells R to set the debug status on all of your functions, i.e. call debug() on them. * ,dbgf This tells R to display a list of our functions with the status on each, regarding whether the function is currently the subject of a debug() call. The user can then specify on which functions to toggle that status. * ,q This exits the browser, i.e. emits a Q command to the browser. * ,dt This executes the R call you've previously stored in the Vim variable g:TestRun. For example, suppose you are testing your code via a call test(5,12,13). You can store that by typing :let g:TestRun = test(5,12,13) in Vim. Then anytime you want to run the test, hit ,dt. __ R-help@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.
[R] new, much improved version of edtdbg debugging tool
A few weeks ago, I released my package edtdbg, which integrates R's debug() with your text editor. At the time, I said I'd release a version with more features in a couple of days. Well, it's taken a while, as I've experimented with various approaches, but I'm now releasing the new version, introducing the new features I had in mind then and some I've thought of since then. The current features include: * Code Tracking: As one steps from line to line of code using the debugger, the cursor in the editor will always follow, i.e. it will alway be on whatever line the debugger is on. * Variable Display: One can enable automatic display of the current arguments and local variables in the R window. * Quick Breakpoints: One can quickly and conveniently set breakpoints. * Toggling of Function Debugging Status: Turn debugging on or off for several functions at once. The current version is for Vim with vim-r only. I did write an Emacs/ESS version, but it started to hang mysteriously, so I am releasing the Vim/vim-r version only. The package is downloadable at http://heather.cs.ucdavis.edu/~matloff/R/edtdbg/edtdbg.zip I'd like to thank Jakson and Duncan M. for some useful e-mail exchanges. Now that this digression is done, I can get back to finishing my Rdsm parallel R package and uploading it to CRAN. (Not sure whether edtdbg is appropriate for CRAN?) Norm Matloff __ R-help@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.
Re: [R] Announce: edtdbg, integrating R's debug() with your text editor
The files seem to be inaccessible, and I can't see why. Neither can my department Web technician. So, for the time being, simply go to http://heather.cs.ucdavis.edu/~matloff/R/edtdbg/ and then download the 3 files as directed. Meanwhile, I've thought of an interesting enhancement, and a couple of people have made suggestions as well. I'll post an enhanced version in a day or so. Norm Date: Sun, 6 Dec 2009 23:39:06 -0800 From: Norm Matloff matl...@cs.ucdavis.edu Subject: [R] Announce: edtdbg, integrating R's debug() with your text editor To: r-help@r-project.org I've just developed edtdbg, a small package that integrates R's debug() with one's text editor. Excerpt from the README file: ... I'm placing the current version of the package at: http://heather.cs.ucdavis.edu/~matloff/R/edtdbg/edtdbg.tar.gz http://heather.cs.ucdavis.edu/~matloff/R/edtdbg/edtdbg.zip __ R-help@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.
[R] Announce: edtdbg, integrating R's debug() with your text editor
I've just developed edtdbg, a small package that integrates R's debug() with one's text editor. Excerpt from the README file: Goal The debug() function in R is primitive. My goal was to make it more usable by integrating it with one's text editor. Hence I wrote the package here, edtdbg. Its features include: * As one steps from line to line of code using the debugger, the cursor in the editor will always follow, i.e. it will alway be on whatever line the debugger is on. * One can enable automatic display of user-specified variables. * One can quickly and conveniently set breakpoints. Currently the package is implemented only for Vim. In addition, the second and third features require the vim-r plugin for Vim. Emacs/ESS users should easily be able to convert the code; some of the machinery is already in the code and in the instructions below. I'm placing the current version of the package at: http://heather.cs.ucdavis.edu/~matloff/R/edtdbg/edtdbg.tar.gz http://heather.cs.ucdavis.edu/~matloff/R/edtdbg/edtdbg.zip Please give me your comments. I probably will upload to CRAN at some point, possibly after an ESS guru contributes the ESS code. :-) A note on my Rdsm package for parallel R: The new, much improved version is just about ready. I'll be uploading Rdsm to CRAN very soon. Norm Matloff __ R-help@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.
Re: [R] Book on R programming
ANJAN PURKAYASTHA wrote: Most books on R I come across describe running statistical procedures in R. Any suggestions on a good book that teaches *programming* in R? Thanks, Anjan I'm in the midst of writing such a book right now. It will be published early next year, by the whimsically named No Starch Press, a subsidiary of O'Reilly (the menagenary series on programming languages and software). You've seen the NSP series in bookstores, with the yellow and black motif. I published a book on debugging with them last year, with Peter Salzman. I just checked with NSP, and they say that I can provide the current manuscript if anyone is interested. Just sent me an e-mail message requesting it. That current manuscript is about 75% complete. Obviously, due to its incomplete state, it likely has various bugs in the code (though I've run all the code), missed opportunities (i.e. code that could be written better if I were to make better use of R constructs), and incorrect statements. But hopefully there are not many such cases, and it should be useful to R programmers, both beginning and experienced. Needless to say, I would like to hear of any such cases you find. A bit on the background I bring to this book project: My PhD was in Math at UCLA, with a thesis in probability theory and a lot of work in applied statistics. I was a founding member of the Dept. of Statistics at UC Davis, but later moved to form the Dept. of Computer Science. My CS research work has often been statistical in nature. I'm an R user going way back to the blue book days of S. Norm Matloff __ R-help@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.
[R] Rdsm, a DSM package for parallel R programming
As I mentioned last week, I've been developing a package that I call Rdsm (R distributed shared memory), modeled after a similar package, PerlDSM, I wrote for Perl some years ago. It is now in alpha form, so I'm not uploading to CRAN yet, but it is definitely usable, and I am releasing it at http://heather.cs.ucdavis.edu/~matloff/R/Rdsm I hope many try it out, and give me some feedback. Note that the word distributed here means that the memory is not really shared, but instead is an abstraction, to give the programmer a shared-memory view even though the program may be running on several separate machines. For C/C++ this is generally accomplished by manipulation of the virtual memory hardware. For R, I do this by redefining functions such as [ and [- for a new class. Rdsm is intended as an alternative for those who favor the shared-memory view of things. In the parallel processing community, there has always been a debate between advocates of the two main programming paradigms, shared memory and message passing. Shared memory advocates claim greater clarity of code, while the message passing people point to that paradigm's greater flexibility. I happen to be of the shared-memory school. Given the popularity of OpenMP for C/C++/FORTRAN, I believe Rdsm will be of interest to many for R. Indeed, in the next few months, I will be extending Rdsm with functions that give it the look and feel of OpenMP. Norm Matloff UC Davis - End forwarded message - __ R-help@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.
Re: [R] queue simulation
Date: 22-Dec-2008 10:11:28 GMT From: Gerard M. Keogh gmke...@justice.ie Subject: [R] queue simulation To: r-help@r-project.org Hi all, I have a multiple queing situation I'd like to simulate to get some idea of the distributions - waiting times and allocations etc. Does R has a package available for this - many years ago there used to be a language called simscript for discrete event simulation and I was wondering if R has an equivalent (or hopefully with graphics, something better!). To my knowledge, this doesn't exist, but one never knows. I look forward to hearing the other responses. Discrete-event simulation (DES) is generally done under one of two main world views--event-oriented, process-oriented. The more popular is probably the process-oriented view, but it requires something like threading, which would be problematic in R. It would be easy to take the event-oriented view, as it would just require coding up some kind of priority queue routine. In fact, a couple of weeks ago I made a note to myself to do this as an example of how one could do linked data structures in R. Again, this view is considered a poor way to program DES, but if you are interested, feel free to contact me. I very much like (and am somewhat involved in the development of) SimPy, a Python-based DES package. You could use SimPy for your simulation and use RPy to access R from SimPy, to take advantage of R's graphics and statistics facilities. SimPy is at http://simpy.sourceforge.net/ Also, I have a tutorial on it at http://heather.cs.ucdavis.edu/~matloff/simcourse.html Norm Matloff University of California, Davis __ R-help@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.