Calling R from GAMA models
Introductionβ
The R language is a powerful tool for statistical computing and graphics, and its community is very large in the world (See the website). Adding a support for the R language is one of our strong endeavors to accelerate many statistical and data mining tools integration into the GAMA platform.
Installing R and rJavaβ
Install R on your computerβ
Please refer to the R official website, or to RStudio if you want in addition a nice IDE.
install the rJava library in Rβ
In the R (RStudio) console, write:
install.packages("rJava")
to install the library. To check that the install is correct, you load the library using library(rJava)
(in the R console). If no error message appears, it means the installation is correct.
In case of troubleβ
For MacOSXβ
in recent versions you should first write in a terminal:
R CMD javareconf
sudo ln -f -s $(/usr/libexec/java_home)/jre/lib/server/libjvm.dylib /usr/local/lib
For Linuxβ
make sure you have the default-jdk
and default-jre
packages installed and then execute the command sudo R CMD javareconf
For Windowsβ
make sure you have a JAVA_HOME
and a CLASSPATH
environment variable setup, if not you need to create and set them, for example:
JAVA_HOME="C:\Program Files\Java\OpenJDK17\"
CLASSPATH="C:\Program Files\Java\OpenJDK17\bin\"
Set the environment variable R_HOME
β
On Windowsβ
set the environment variables as follows.
R_HOME
is the root directory where we can find the library
folder in your R
installation, so it should look like this:
R_HOME="C:\Program Files\R\R-4.2.2\"
R_PATH
should point to the folder containing your R
interpreter, the variable should be set with something similar to this (adapting with your R version and R installation path):
R_PATH="C:\Program Files\R\R-4.2.0\bin\x64"
On Linuxβ
By default it should be /usr/lib/R
, you can thus just append the line R_HOME=/usr/lib/R
to your /etc/environment
file and reboot your computer
On macOSβ
You need to create (or update) the file environment.plist
in the folder: ~/Library/LaunchAgents/
Β (for the current user, note that this folder is a hidden folder) or inΒ /Library/LaunchAgents/
Β (for all users)
It should look like:
<?xml version="1.0" encoding="UTF-8"?>
<!DOCTYPE plist PUBLIC "-//Apple//DTD PLIST 1.0//EN" "http://www.apple.com/DTDs/PropertyList-1.0.dtd">
<plist version="1.0">
<dict>
<key>Label</key>
<string>my.startup</string>
<key>ProgramArguments</key>
<array>
<string>sh</string>
<string>-c</string>
<string>Β launchctl setenv R_HOME /Library/Frameworks/R.framework/Resources/Β </string>
</array>
<key>RunAtLoad</key>
<true/>
</dict>
</plist>
Recommendedβ
If the rJava library doesn't appear in the R library directory, copy the installed rJava library from where it was installed (with install.packages("rJava")
) to the library
folder in your R_HOME
.
Updating the Path
variable (Windows only)β
In addition, on Windows you also need to add to your Path
environment variable the path to your R
binaries, by default located in C:\Program Files\R\R-4.2.2\bin\x64
for R-4.2.2 64bits
.
The Path
variable is a variable already created by Windows, so you just have to edit it to add a new path, no need to delete anything.
Configuration in GAMAβ
Linking the R connectorβ
From GAMA 1.9.0, you need to specify the path to the R connector library in the GAMA launching arguments. To this purpose, you need to add to either:
-
the
GAMA.ini
file if you use the release version of GAMA -
or the launching configuration (if you use the source code version) the following line: (replace
PATH_TO_R
by the path to R, i.e. the value in$R_HOME
):- on macOS:
-Djava.library.path=PATH_TO_R/library/rJava/jri/rlibjri.jnilib
- on Windows:
-Djava.library.path=PATH_TO_R\library\rJava\jri\
- on Linux:
-Djava.library.path=PATH/TO/JRI
- on macOS:
As an example, under macOS, you need to add:
-Djava.library.path=/Library/Frameworks/R.framework/Resources/library/rJava/jri/
On Windows and Linux, the jri library could be in a different location than the R_HOME
, for example on Linux by default it would be in:
-Djava.library.path=/home/user_name/R/x86_64-pc-linux-gnu-library/3.6/rJava/jri/
On Windows it can be located in the user's AppData\local
or in Documents\R
.
Installing the R pluginβ
Next you need to install the R plugin from Gama. To do it, select "Install new plugins..." in the "Help" menu of Gama.
In the Work with
drop down select the repository ending with "experimental/" followed by your Gama version.
Once done, you need to select the plugin rJava
, click on next
and then finish
.
After this, you could be asked to "trust" the plugin, simply select the first line and click on Trust selected
Finally, you will be asked to restart Gama, click on Restart now
.
For more details, readers can refer to the page dedicated to the installation of additional plugins.
Calling R from GAMLβ
Before computationβ
Any agent aiming at using R for some computation needs to be provided with the RSkill
.
Before calling any computation, this agent needs to start a connection with the R software.
As an example, if we want that the global
agent can use R, we need to have the following minimal model:
global skills: [RSkill] {
init {
do startR;
}
}
Computationβ
Evaluate an R expressionβ
The R_eval
operator can be used to evaluate any R expression. It can also be used to initialize a variable or call any function. It can return any data type (depending on the R output).
As in an R session, the various evaluations are dependent on the previous ones.
Example:
global skills: [RSkill] {
init{
do startR;
write R_eval("x<-1");
write R_eval("rnorm(50,0,5)");
}
}
Evaluate an R scriptβ
To evaluate an R script, stored in a (text) file, open the file and execute each of its lines.
global skills:[RSkill]{
file Rcode <- text_file("../includes/rScript.txt");
init{
do startR;
// Loop that takes each line of the R script and execute it.
loop s over: Rcode.contents{
unknown a <- R_eval(s);
write "R>"+s;
write a;
}
}
}
Convert GAMA object to R objectβ
To use GAMA complex objects into R functions, we need to transform them using the to_R_data
operator: it transforms any GAMA object into a R object.
global skills:[RSkill] {
init {
do startR();
string s2 <- "s2";
list<int> numlist <- [1,2,3,4];
write R_eval("numlist = " + to_R_data(numlist));
}
}
Convert a species to a dataframeβ
Dataframe is a powerful R data type allowing to ease data manipulation...
Dataframe wan of course be defined at hand using R commands. But GAML provides the to_R_dataframe
operator to directly transform a species of agents into a dataframe for future analysis.
global skills: [RSkill] {
init{
do startR();
create people number: 10;
do R_eval("df<-" + to_R_dataframe(people));
write R_eval("df");
write R_eval("df$flipCoin");
}
}
species people {
bool flipCoin <- flip(0.5);
}
Troubleshootingβ
It is possible that after installing everything, Gama works normally but crashes every time you try to use the skill RSkill
without any error message. If that's the case, the problem is certainly that Gama is unable to load the jri
library or its dependencies (other R packages). Make sure that the path you wrote in the .ini
file is correct and that every environment variable is set with proper values.
Also on windows, check that the Path
variable contains the path to your R
installation.
If after checking everything the problem is still there, you can try copying the .dll
files at the R_PATH
location and the jri.dll
and paste them into your JAVA_PATH
directory (the bin
folder of your jdk).