Re: [Rd] Automatic Differentiation for R
In efforts to improve optimization tools for R, one of my interests has been getting automatic differentiation capabilities so that analytic rather than numerical derivatives can be used. They would be helpful in several other areas besides optimization, My timings show factors of the order of 1000s in time improvements by avoiding numerical derivatives in some cases. There has been some work in this e.g., http://code.google.com/p/pbs-software/ is an R interface to ADMB (Automatic Differentiation Model Builder). However, as far as I can see, this is directed essentially to nonlinear least squares modelling, an important but not general problem. Tom Coleman of Waterloo responded favourably with some advice, but the most enthusiastic answer came from Shaun Forth, which I have included below. I read this as an opportunity to develop what could be a profitable collaboration with the AD community. Unfortunately, I cannot take up the invitation to join the AD folk in Oxford due to a pre-existing obligation. Nor am I more than a complete novice with S3 and S4 classes etc. I am, nevertheless, willing to help organize the effort e.g., do some of the communications, chasing grant money, getting Google Summer of Code applications filled in etc. Can the R community come up with a few people who can provide the AD workers with appropriate information? If so, is there a reasonable chance to generate sufficient funding for a student? I suspect the answer in both cases is yes, but that we need some form of booster cables to get things going. (In Canada, booster cables are used to get cars started in winter by connecting a running vehicle's battery to that of a dead one.) I suggest communications off-list until there is progress to report. Possibly there is a better forum for this -- suggestions welcome. John Nash included msg from Shaun Forth --- Hi John, My computational statistics colleague Trevor Ringrose has asked me to consider AD in R in the past. As you may or may not be aware AD is implemented in one of two ways: overloading using OO features of the target language, or source transformation using compiler tools (after several man years of development) to read in the target code and spit out the differentiated code. Last time I looked I didn't think the object oriented features of R were up to overloading but on checking today I can see that this might now be possible (I can see overloading of arithmetic operators and functions for example now which I didn't see last time). I'd certainly be interested in following this up particularly on the overloading side but would need to get funding for a PhD student to do the graft. It would be particularly interesting doing this in an open source language because we could then perhaps tweak some of the core language features if they didn't seamlessly support the AD (we can't do this in Matlab and that is a pain!). My immediate suggestion is that you, or some other more local (to UK) R expert talks at the next European AD workshop in Oxford http://www.autodiff.org/?module=Workshopssubmenu=EuroAD/8/main We're a very friendly group and I'm sure there are others who might like to tackle R or perhaps we could put together a multigroup project. If someone could give a talk on R, its language features including the OO aspects, and some optimisation examples with associated code, the group there would be able to give you the best feedback on the planet on the possibilities. Please do treat this as a positive response and let's keep in touch on this. Regards Shaun Dr Shaun Forth Applied Mathematics Scientific Computation Cranfield Defence and Security Cranfield University, Shrivenham Campus Swindon SN6 8LA, England tel: +44 (0)1793 785311 fax: +44 (0)1793 784196 email: s.a.fo...@cranfield.ac.uk http://www.amorg.co.uk # __ R-devel@r-project.org mailing list https://stat.ethz.ch/mailman/listinfo/r-devel
Re: [Rd] Automatic Differentiation for R
Not sure if this is sufficient for your needs but R does include symbolic differentiation, see ?D, and the Ryacas and rSymPy packages interface R to the yacas and sympy computer algebra systems (CAS) and those system include symbolic differentiation. http://ryacas.googlecode.com http://rsympy.googlecode.com Note that Ryacas communicates with yacas via XML but recent versions of the XML package changed in a way that breaks Ryacas so you will likely have to use an old version of XML and Ryacas if you want to try that one -- see home page. The rsympy interface is early stage but its functional and is easier to install since it includes the entire CAS right in the R package. On Wed, Apr 15, 2009 at 9:30 AM, John C Nash nas...@uottawa.ca wrote: In efforts to improve optimization tools for R, one of my interests has been getting automatic differentiation capabilities so that analytic rather than numerical derivatives can be used. They would be helpful in several other areas besides optimization, My timings show factors of the order of 1000s in time improvements by avoiding numerical derivatives in some cases. There has been some work in this e.g., http://code.google.com/p/pbs-software/ is an R interface to ADMB (Automatic Differentiation Model Builder). However, as far as I can see, this is directed essentially to nonlinear least squares modelling, an important but not general problem. Tom Coleman of Waterloo responded favourably with some advice, but the most enthusiastic answer came from Shaun Forth, which I have included below. I read this as an opportunity to develop what could be a profitable collaboration with the AD community. Unfortunately, I cannot take up the invitation to join the AD folk in Oxford due to a pre-existing obligation. Nor am I more than a complete novice with S3 and S4 classes etc. I am, nevertheless, willing to help organize the effort e.g., do some of the communications, chasing grant money, getting Google Summer of Code applications filled in etc. Can the R community come up with a few people who can provide the AD workers with appropriate information? If so, is there a reasonable chance to generate sufficient funding for a student? I suspect the answer in both cases is yes, but that we need some form of booster cables to get things going. (In Canada, booster cables are used to get cars started in winter by connecting a running vehicle's battery to that of a dead one.) I suggest communications off-list until there is progress to report. Possibly there is a better forum for this -- suggestions welcome. John Nash included msg from Shaun Forth --- Hi John, My computational statistics colleague Trevor Ringrose has asked me to consider AD in R in the past. As you may or may not be aware AD is implemented in one of two ways: overloading using OO features of the target language, or source transformation using compiler tools (after several man years of development) to read in the target code and spit out the differentiated code. Last time I looked I didn't think the object oriented features of R were up to overloading but on checking today I can see that this might now be possible (I can see overloading of arithmetic operators and functions for example now which I didn't see last time). I'd certainly be interested in following this up particularly on the overloading side but would need to get funding for a PhD student to do the graft. It would be particularly interesting doing this in an open source language because we could then perhaps tweak some of the core language features if they didn't seamlessly support the AD (we can't do this in Matlab and that is a pain!). My immediate suggestion is that you, or some other more local (to UK) R expert talks at the next European AD workshop in Oxford http://www.autodiff.org/?module=Workshopssubmenu=EuroAD/8/main We're a very friendly group and I'm sure there are others who might like to tackle R or perhaps we could put together a multigroup project. If someone could give a talk on R, its language features including the OO aspects, and some optimisation examples with associated code, the group there would be able to give you the best feedback on the planet on the possibilities. Please do treat this as a positive response and let's keep in touch on this. Regards Shaun Dr Shaun Forth Applied Mathematics Scientific Computation Cranfield Defence and Security Cranfield University, Shrivenham Campus Swindon SN6 8LA, England tel: +44 (0)1793 785311 fax: +44 (0)1793 784196 email: s.a.fo...@cranfield.ac.uk http://www.amorg.co.uk # __ R-devel@r-project.org mailing list https://stat.ethz.ch/mailman/listinfo/r-devel __
Re: [Rd] HTML help
JAck, On Apr 14, 2009, at 22:50 , Jack (Zhan, Hua Ping) wrote: Hi,guys, HTML help is not right for math formula: For example: I got the following formula in help page of ca.jo in urca package: p align=centeriBX/B_t = BPi/B_1 BX/B_{t-1} + ... + BPi/B_k BX/B_{t-k} + Bmu;/B + BPhi D/B_t + Bvarepsilon/B_t , quad (t = 1, ..., T),/i/p p This presentation is just not good, however, in most browser, with tag math/math, the formula in the tag works well. I don't see what you mean - just putting math tags around the above doesn't change rendering at all. Moreover math is not a valid tag in the HTML standard. such as in wikipedia. please check the same formula: http://en.wikipedia.org/wiki/Johansen_test Wikipedia uses bitmaps to represent those formulae, there are no HTML math tags involved there. Can someone please fix this? help(ca.jo,offline=TRUE), we can get the formula right. The problem seems to occur at the building process. Latex output is right, but the HTML output is not good for math formula. I guess this issue is not so difficult. Well, then fix it and send us the patch please ... Cheers, Simon __ R-devel@r-project.org mailing list https://stat.ethz.ch/mailman/listinfo/r-devel
[Rd] Correction to documentation of print.ls_str in utils package
In the documentation for print.ls_str (on the ls.str page) in the utils package, the description for max.level is not correct. It says Default 0: Display all nesting levels. The default, according to the Usage on that page and the source code, is 1. Also, a value of 0 does not display all the nesting levels; NA does that (based on the documentation of str). A value of 0 does not display any nested levels. I recommend changing that sentence to If NA, display all nesting levels. Default 1: Display only the first nested level. -- Brian Diggs, Ph.D. Senior Research Associate, Department of Surgery, Oregon Health Science University __ R-devel@r-project.org mailing list https://stat.ethz.ch/mailman/listinfo/r-devel
Re: [Rd] HTML help
Hi Jack, I'm not sure about the math tag either, and it didn't work when I tried it. I think the approach this web page uses may help in writing mathematical symbols to HTML: http://comers.citadel.edu/math_sym2005.htm Cheers, Mike Jack (Zhan, Hua Ping) wrote: HTML help is not right for math formula: For example: I got the following formula in help page of ca.jo in urca package: p align=centerX_t = Pi_1 X_{t-1} + ... + Pi_k X_{t-k} + mu; + Phi D_t + varepsilon_t , quad (t = 1, ..., T),/p p This presentation is just not good, however, in most browser, with tag math/math, the formula in the tag works well. such as in wikipedia. please check the same formula: http://en.wikipedia.org/wiki/Johansen_test -- with best regards Jack (Zhan, Hua Ping) +1-514-8800518 -- View this message in context: http://www.nabble.com/HTML-help-tp23054298p23063793.html Sent from the R devel mailing list archive at Nabble.com. __ R-devel@r-project.org mailing list https://stat.ethz.ch/mailman/listinfo/r-devel
Re: [Rd] HTML help
Hi, if the system that builds the html help page is latex capable, then it would not be too difficult to use the same trick wikipedia is using (generating a png file for each equation). Translating latex style markup into mathml is another game, see this: http://www1.chapman.edu/~jipsen/mathml/asciimath.html for example. Romain mnorton52 wrote: Hi Jack, I'm not sure about the math tag either, and it didn't work when I tried it. I think the approach this web page uses may help in writing mathematical symbols to HTML: http://comers.citadel.edu/math_sym2005.htm Cheers, Mike Jack (Zhan, Hua Ping) wrote: HTML help is not right for math formula: For example: I got the following formula in help page of ca.jo in urca package: p align=centerX_t = Pi_1 X_{t-1} + ... + Pi_k X_{t-k} + mu; + Phi D_t + varepsilon_t , quad (t = 1, ..., T),/p p This presentation is just not good, however, in most browser, with tag math/math, the formula in the tag works well. such as in wikipedia. please check the same formula: http://en.wikipedia.org/wiki/Johansen_test -- with best regards Jack (Zhan, Hua Ping) +1-514-8800518 -- Romain Francois Independent R Consultant +33(0) 6 28 91 30 30 http://romainfrancois.blog.free.fr __ R-devel@r-project.org mailing list https://stat.ethz.ch/mailman/listinfo/r-devel
[Rd] (Not quite) Automatic Differentiation
Many thanks to Gabor Grothendieck for responding to my posting about Automatic Differentiation (invite from Shaun Forth for interaction with R developers) showing how one might use rSymPy and symbolic (rather than automatic) differentiation to get a function that computes gradients. See http://code.google.com/p/rsympy/#Automatic_Differentiation_(well,_sort_of) for a worked example on the Broyden test function. This is a big step forward. There's still a way to go before we can produce a vectorized gradient code automatically when the size of the problem is variable, but the example may serve to incite some imaginative coders to action. Thanks again Gabor. JN __ R-devel@r-project.org mailing list https://stat.ethz.ch/mailman/listinfo/r-devel