Does one need to have SAS IML installed to use the SAS/R interface? or should/could one use the sas x command to run R and feed data to it?
If you want to actually use the SAS/R interface, then yes, you must license and have SAS/IML installed as it is specifically a feature of SAS/IML (which makes sense, as SAS/IML is SAS's matrix programming language, and R is a matrix programming language).
However, you're welcome to use R the way you describe (by submitting R programs via xcmd); you will, however, need to use a CSV file or similar to exchange data between the two programs. There are several ways to do it, so look at the different options available to see what's easiest for you.
If you're choosing between the different ways to do this, here is a list of the advantages of using IML which serves as a nice comparison between the two (perhaps a biased one (Rick is the lead developer of SAS/IML), but it is sufficiently detailed in what you won't have available to you running it as a separate program that it should be helpful in making the decision).
I'm looking for some kind of references which explain the pro's en con's of using Rcpp when compared to using rdyncall.
Can someone who has used both explain the basic differences from an R package developers perspective who is interested in providing R wrappers to C++ code.
Thanks.
I think we mention rdyncall in the (brief) comparison to other approaches the intro vignette / JSS paper. It is a neat package, but aims for a much lower-level connection. As I understand it, it gives you C-level APIs with least amount of fuzz, as motivated by say, the rgl package. there is very good and detailed paper about rdyncall in a recent R Journal issue.
And unless I miss something, it does nothing for you on the C++ side. Whereas Rcpp makes use as .Call() to pass complete R objects back and forth, and manages to map a wide variety of R and C++ types automatically for you---with the possibility add your own mappers.
Over at http://scicomp.stackexchange.com I asked this question about parallel matrix algorithms in IDL. The answers suggest using a multi-threaded LAPACK implementation and suggest some hacks to get IDL to use a specific LAPACK library. I haven't been able to get this to work.
I would ideally like the existing LAPACK DLM to simply be able to use a multi-threaded LAPACK library and it feels like this should be possible but I have not had any success. Alternatively I guess the next simplest step would be to create a new DLM to wrap a matrix inversion call in some C code and ensure this DLM points to the desired implementation. The documentation for creating DLMs is making me cross-eyed though, so any pointers to doing this (if it is required) would also be appreciated.
What platform are you targeting?
Looking at idl_lapack.so with nm on my platform (Mac OS X, IDL 8.2.1) seems to indicate that the LAPACK routines are directly in the .so, so my (albeit limited) understanding is that it would not be simple to swap out (i.e., by setting LD_LIBRARY_PATH).
$ nm idl_lapack.so
...
000000000023d5bb t _dgemm_
000000000023dfcb t _dgemv_
000000000009d9be t _dgeqp3_
000000000009e204 t _dgeqr2_
000000000009e41d t _dgeqrf_
000000000023e714 t _dger_
000000000009e9ad t _dgerfs_
000000000009f4ba t _dgerq2_
000000000009f6e1 t _dgerqf_
Some other possibilities...
My personal library has a directory src/dist_tools/bindings containing routines for automatically creating bindings for a library given "simple" (i.e., not using typedefs) function prototypes. LAPACK would be fairly easy to create bindings for (the hardest part would probably be to build the package you want to use ATLAS, PLAPACK, ScaLAPACK, etc.). The library is free to use, a small consulting contract could be done if you would like it done for you.
The next version of GPULib will contain a GPU implementation of LAPACK, using the MAGMA library. This is effectively a highly parallel option, but only works on CUDA graphics cards. It would also work best if other operations besides the matrix inversion could be done on the GPU to minimize memory transfer. This option costs money.
I am doing matrix operations on large matrices in my C++ program. I need to verify the results that I get, I used to use WolframAlpha for the task up until now. But my inputs are very large now, and the web interface does NOT accept such large values (textfield is limited).
I am looking for a better solution to quickly cross-check/do math problems.
I know there is Matlab but I have never used it and I don't know if thats what will suffice my needs and how steep the learning curve would be?
Is this the time to make the jump? or there are other solutions?
If you don't mind using python, numpy might be an option.
Apart from the license costs, MATLAB is the state of the art numerical math tool. There is octave as free open source alternative, with a similar syntax. The learning curve is for both tools absolutely smooth!
WolframAlpha is web interface to Wolfram Mathematica. The command syntax is exactly the same. If you have access to Mathematica at your university, it would be most smooth choice for you since you already have experience with WolframAlpha.
You may also try some packages to convert Mathematica commands to MATLAB:
ToMatlab
Mathematica Symbolic Toolbox for MATLAB 2.0
Let us know in more details what is your validation process. How your data look like and what commands have you used in WolframALpha? Then we can help you with MATLAB alternative.
As it currently stands, this question is not a good fit for our Q&A format. We expect answers to be supported by facts, references, or expertise, but this question will likely solicit debate, arguments, polling, or extended discussion. If you feel that this question can be improved and possibly reopened, visit the help center for guidance.
Closed 11 years ago.
I have been working on a few R packages for some general tools that aren't currently available in R: blogging, report delivery, logging, and scheduling. This led me to wonder: what are the most important things that people wish existed in R that currently aren't available?
My hope is that we can use this to pinpoint some gaps, and possibly work on them collaboratively.
I'm a former Mathematica junkie, and one thing that I really miss is the notebook style interface. When I did my research with notebooks, papers would almost write themselves as I did my analysis. But now that I'm using R, I find that documenting my work to be quite tedious.
For people that are not so familiar with Mathematica, you have documents called "notebooks" that can contain code, text, equations, and the results from executed code (which can be equations, text, graphics, or interactive tools). Everything can be neatly organized into styled subsections or sections that are collapsable. You can have multiple open documents that integrate with a single shared kernel.
While I don't think a full-blown Mathematica style interface is entirely necessary, some interactive document system that would support text (for description), code, code output, and embedded image output would be a real boon to researchers.
A Real-Time R package would be my choice, using C Streaming perhaps.
Also I'd like a more robust web development package. Nothing as extensive as Ruby on Rails but something a bit better than Sweave combined with R2HTML, that can run on RApache. I think this needs to be a huge area of emphasis for R in general.
I realize LaTeX is better markup for certain academia but in general I think HTML should be the markup language of choice. More needs to be done in terms of R Web Apps, so applications can be hosted on huge RAM remotely and R can start being used for SaaS data applications and other graphics choices.
Interfaces to any of the new-fangled 'Web 2.0' databases that use key-value pairs rather than the standard RDMS. A non-exhaustive list (in alphabetical order) would be
Cassandra Project
CouchDB
MongoDB
Project Voldemort
Redis
Tokyo Cabinet
and it would of course be nice if we had a DBI-alike abstraction on top of this. Jeff has started with RBerkeley but that use the older-school Oracle BerkeleyDB backend rather than one of those new things.
An output device which produces Javascript code, perhaps using the protovis library.
as a programmer and writer of libraries for colleagues, I was definitely missing a logging package, I googled and asked around, here too, then wrote one myself. it is on r-forge, here, and it s called "logging" :)
I use it and I'm obviously still developing it.
There are few libraries to interface with database in general, and there is not ORM library.
RMySQL is useful, but you have to write the SQL queries manually and there is not a way to generate them as in a ORM. Morevoer, it is only specific to MySQL.
Another library set that R still doesn't have, for me, it is a good system for reading command line arguments: there is R getopt but it is nothing like, for example, argparse in python.
A natural interface to the .NET framework would be awesome, though I suspect that that might be a lot of work.
EDIT:
Syntax highlighting from within RGui would also be wonderful.
ANOTHER EDIT:
R.NET now exists to integrate R with .NET.
A FRAQ package for FRequently Asked Questions, a la fortune(). R-help would be so much fun: "Try this, library(FRAQ); faq("lattice won't print"), etc.
See also.
A wiki package that adds wiki-like documentation to R packages. You'd have a inst/wiki subdirectory with plain text files in markdown, asciidoc, textile, with embedded R code. With the right incantation, these files would be executed (think brew and/or asciidoc packages), and the relevant output uploaded to a given repository online (github, googlecode, etc.). Another function could take care of synchronizing the changes made online, typically via svn or git.
Suddenly you have a wiki documentation for your package with reproducible examples (could even be hooked to R CMD check).
EDIT 2012:
... and now the knitr package would make this process even easier and neater
I would like to see a possibility to embed another programming language within R in a more straightforward way by the users. I give this as an example in some common-lisp implementations one could write a function with embedded C code like this:
(defun sample (x)
(ffi:c-inline (n1 n2) (:int :int) (values :int :int) "{
int n1 = #0, n2 = #1, out1 = 0, out2 = 1;
while (n1 <= n2) {
out1 += n1;
out2 *= n1;
n1++;
}
#(return 0)= out1;
#(return 1)= out2;
}"
:side-effects nil))
It would be good if one could write an R function with embedded C or lisp code (more interested in the latter) in a similar way.
A native .NET interface to RGUI. R(D)Com is based on COM, and it only allows to exchange matrices, not more complex structures.
I would very much like a line profiler. This exists in Matlab and Python, and is very useful for finding bits of code that take a lot of time or are executed more (or less) than expected. A lot of my code involves function optimizations and how many times something iterates may not be known in advance (though most iterations are constrained or specified).
The call stack is useful if all of your code is in R and is very simple, but as I recently posted about it, it takes a painstaking effort if your code is complex.
It's quite easy to develop a line profiler for a given bit of code. A naive way is to index every line (or just pre-specified sections) and insert a call to log proc.time() that line. In a loop, I simply enumerate sections of code and store in a 2 dimensional list the proc.time values for section i in iteration k. [See update below: this isn't actually a way to do a line profiler for all kinds of code.]
One can use such a tool to find hotspots, anomalies (e.g. code that should be O(n) but is really O(n^2)), code that may benefit from memoization (a line profiler doesn't tell you this, but it lets you know where to look), code that is mistakenly inside a loop, and more.
Update 1: Inserting a timing line between every function line is slightly erroneous: the definition of a line of code is not simply code separated by whitespace. Being able to parse the code into an AST is necessary for knowing where operations begin and end. As discussed in some of the answers to this question, there are some tools (namely, showTree and walkCode in the codetools package) for doing this. Simply applying a regular expression to source code would be a very bad thing to do.