[ADMB Users] AD Model Builder and R

Mon Dec 15 13:32:45 PST 2008

All--

This is explain a little more about the software (ADMB2R) that coauthors 
and I have developed and that is available from my Web site, linked to 
by the ADMB Project site.  It is also available from CRAN.

Very briefly, ADMB2R provides an alternative to the REPORT_SECTION of 
ADMB.  In the ADMB template, a user writes a block of calls to 
high-level functions supplied by ADMB2R.  Those functions cause a 
special ASCII file to be written that can be imported by R's dget() 
function to form an R "list" object.  The calls are pretty simple, 
hardly more difficult than writing a standard output section. They do 
things like open the list, start a sublist (list within the list), store 
vectors or matrices, give names to the matrices, lists, and vectors, 
close the list, and that type of thing. 

So the output we provide is entirely independent of the REPORT_SECTION.  
It turns out, though, a good way to write your first ADMB2R function 
calls is to start with an existing REPORT_SECTION and incrementally add 
parallel code for R-compatible output.

The assessment group here now uses this as the main output from ADMB.  
We write a short standard report section as well.

The advantages of this approach are that all data from any ADMB run can 
be transferred to R in a structured way, complete with names, comments, 
etc.  The output is complete and does not require parsing code (beyond 
the R parser itself).  It eliminates the need for multiple output files 
per run, and the data within the list can be structured and named to 
facilitate automated graphing.  (We have something called FishGraph, 
also available in the same places, that we use for that.)

ADMB2R is OS-independent and quick -- it is compiled with the template.  
However, due to quirks in ADMB, it works with only some compilers.  We 
have tested it with  Borland and gcc.  We hope that updates to ADMB will 
remove this compiler dependence so that ADMB2R can be used with any 
compiler.

The download of ADMB2R includes detailed documentation and examples.

Mike

on 12/15/2008 3:10 PM John Sibert said the following:
> Hi Folks,
>
> Perhaps someone could clarify what is meant by an "R interface to ADMB"?
>
> Communication of data between R and admb applications is easily done 
> with ascii files. This is what Anders Nielsen and I do for all of our 
> track reconstruction software. Pierre Kleiber has written R packages 
> to interact with MULTIFAN-CL. This approach is fast and platform 
> independent. So I'm not sure what you are asking for.
>
> Mike Prager has created some software that he calls "middleware" for 
> examining ADMB output. I haven't taken the opportunity to examine 
> Mike's utilities, but the information on the website suggests it might 
> be very useful:  http://www.sefsc.noaa.gov/mprager/rinter.htm
>
> Creation of installation scripts and of an IDE for developing admb 
> application are separate issues that should not depend on R. The Inno 
> installer that Dave created is a promising start for Windows 
> installations and will be developed further. Also we have had some 
> positive results from preliminary tests of NetBeans for creation of an 
> ADMB IDE.
>
> So, lets hear some feedback on what is meant by an "R interface to ADMB".
>
> Regards,
> John
>
>
>
> Schnute, Jon wrote:
>> Mark and colleagues - This note is a reply to a message forwarded to 
>> me by Mark Maunder. I'm addressing a question by Marc Kery: "Any 
>> chance that someone might develop an R interface [to ADMB]?"
>>
>> You may or may not know that, with the help of colleagues here, I've 
>> developed several packages for R in recent years. Three of them are 
>> currently available on CRAN http://cran.r-project.org/. Each package 
>> name begins with "PBS" for the Pacific Biological Station, where I 
>> still work as a retired scientist.
>>
>> We're now developing a package (PBSadmb) designed to support ADMB 
>> within R. It will have a GUI interface, and it should make 
>> installation of ADMB fairly easy for R users. Hopefully, it will work 
>> on all platforms that support R, as long as the ADMB Project 
>> distributes the corresponding binary libraries.
>>
>> See the attached PDF file for a picture of the current (rather 
>> primitive) interface. It may take a while for us to support multiple 
>> version of Open ADMB, but I'll send a notice to all ADMB users 
>> (users at admb-project.org) when PBSadmb is available.
>>
>> Best wishes,
>> Jon
>>
>> ******************************************
>> Jon Schnute, Scientist Emeritus
>> Pacific Biological Station
>> 3190 Hammond Bay Road
>> Nanaimo, BC V9T 6N7
>> CANADA
>>
>> NEW email: Jon.Schnute at dfo-mpo.gc.ca  (formerly 
>> schnutej at pac.dfo-mpo.gc.ca)
>> ******************************************
>>
>>  
>> -----Original Message-----
>> From: Mark Maunder [mailto:mmaunder at iattc.org] Sent: Wednesday, 
>> December 10, 2008 4:13 PM
>> To: Kery Marc
>> Cc: Arni Magnusson; Schnute, Jon
>> Subject: RE: AD Model Builder now free
>>
>> Marc,
>>
>> I guess that there will be some form of integration with R because 
>> many of the current ADMB users also use R. We have a couple of people 
>> involved in the initial stages who have a lot more R knowledge than I 
>> do. Hopefully, they will be motivated to make the necessary 
>> developments.
>> Currently there are DLLs for R, but few people use them.
>>
>> Regards,
>>
>> Mark
>> -----Original Message-----
>> From: Kery Marc [mailto:marc.kery at vogelwarte.ch] Sent: Wednesday, 
>> December 10, 2008 12:37 PM
>> To: Mark Maunder
>> Subject: AW: AD Model Builder now free
>>
>> Hi Mark,
>>
>> any chance that someone might develop an R interface ?
>>
>> Regards  --  Marc
>>
>> -----Ursprüngliche Nachricht-----
>> Von: bounce-3381192-3455202 at list.cornell.edu
>> [mailto:bounce-3381192-3455202 at list.cornell.edu]Im Auftrag von Mark
>> Maunder
>> Gesendet: Mittwoch, 10. Dezember 2008 21:31
>> An: euring-l at cornell.edu
>> Betreff: AD Model Builder now free
>>
>>
>> The most widely used software package for the development of
>> state-of-the-art fisheries stock assessment methods, AD Model Builder,
>> or ADMB, can now be downloaded without charge from a public website,
>> http://admb-project.org.
>>
>> ADMB-based computer models are used globally to monitor populations of
>> many endangered and commercially valuable species, to develop
>> place-based resource management policies, and to reconstruct movements
>> of animals tracked with electronic tags. ADMB- based stock assessments
>> are critical to the management of commercially important fisheries
>> stocks worth billions of dollars as well as ecologically sensitive
>> species, in the United States and
>> internationally. Every NOAA Fisheries Science Center uses the ADMB
>> software.
>>
>> The recently released EURING 2007 proceedings has a manuscript
>> describing the use of ADMB for mark-recapture analysis: "Comparison of
>> Fixed Effect, Random Effect, and Hierarchical Bayes Estimators for Mark
>> Recapture Data Using AD Model Builder"
>>
>> Please feel free to e-mail me if you have any questions about the ADMB
>> software
>>
>> Mark Maunder
>> Inter-American  Tropical Tuna Commission
>> 8604 La Jolla Shores Drive
>> La Jolla, CA, 92037-1508, USA
>>   Tel: (858) 546-7027
>> Fax: (858) 546-7133
>> mmaunder at iattc.org
>> http://iattc.org/iattc-staffMMaunder.htm
>>
>> Spatial Analysis for Stock Assessment workshop 14-17 October 2008
>> http://www.fisheriesstockassessment.com/TikiWiki/tiki-index.php?page=Spa
>> tial+Analysis+for+Stock+Assessment+2008
>>
>> See the following website for information on fisheries stock assessment
>> http://www.fisheriesstockassessment.com/
>>
>>     
>> ------------------------------------------------------------------------
>>
>> _______________________________________________
>> Users mailing list
>> Users at admb-project.org
>> http://lists.admb-project.org/mailman/listinfo/users
>>   
>

-- 
Michael Prager, Ph.D.
Senior Scientist
NOAA Southeast Fisheries Science Center
Beaufort, North Carolina  28516
http://www.sefsc.noaa.gov/mprager/
** Opinions expressed are personal, not official.  No
** official endorsement of any product is made or implied.