Please add material here as you learn...
If you have questions not answers, then add those here: That's how a wiki works.
We ask that use the following two citations in your publications that use OpenMx. As of OpenMx 1.2.1, these citations are available from R by invoking
Steven M. Boker, Michael C. Neale, Hermine H. Maes, Michael J. Wilde, Michael Spiegel, Timothy R. Brick, Jeffrey Spies, Ryne Estabrook, Sarah Kenny, Timothy C. Bates, Paras Mehta, and John Fox. (2011) OpenMx: An Open Source Extended Structural Equation Modeling Framework. Psychometrika. Steven M. Boker, Michael C. Neale, Hermine H. Maes, Michael J. Wilde, Michael Spiegel, Timothy R. Brick, Ryne Estabrook, Timothy C. Bates, Paras Mehta, Timo von Oertzen, Ross J. Gore, Michael D. Hunter, Daniel C. Hackett, Julian Karch and Andreas Brandmaier. (2012) OpenMx 1.2 User Guide.
If you download a new version of the package, or compile a new from from the svn, you should quit R and restart to ensure you have the new version loaded
1. Download the appropriate binary package.
2. Save it in your USB drive
3. Use the following command to install it, e.g.,
You can view the list of demo files with this command:
Then this will show you where they are on your system
system.file("demo", "BivariateSaturated_PathCov.R", package="OpenMx")  "/Users/~/Library/R/2.9/library/OpenMx/demo/BivariateSaturated_PathCov.R"
You can take a look at our performance tests as they are updated on a wiki page. The scripts can be found in our svn repository under the /models/nightly directory. If you have a script you'd like to add to the repository please let us know.
See the answer below.
The R interpreter includes two implementations of BLAS on OS X. The reference implementation uses naive algorithms to provide functionality, and the vecLib implementation uses a library from Apple. The vecLib implementation has performance regressions on small matrices, which is why the binary release of OS X has been dynamically linked to the reference implementation. You can mimic this behavior on your system by altering the symbolic link at /Library/Frameworks/R.framework/Versions/Current/Resources/lib/libRblas.dylib to point to
A named entity is the term used by the OpenMx library for any S4 object that includes the slot "name". Examples of S4 classes in the OpenMx library that create named entities include MxModel, MxMatrix, MxAlgebra, MxObjectiveFunction, and MxData. An equivalent answer to this question would be any S4 object 'foo' such that 'foo@name' will return a value.
If model refers to a MxModel object, then use "?summary(model)" or the more generic "method ? summary('MxModel')". Documentation for the summary function was added in version 0.2.5.
Yes, there is. You can access it from Preferences:Bundles or github.com/tbates/OpenMx.tmbundle
OpenMx now runs on 64-bit windows.
Not yet. There may be support for the OpenMP mulithreading API in a future version of Rtools. Until this is available, we can only provide sequential OpenMx in Windows. If you wish to compile OpenMx from source you may be able to get multithreading support by placing a copy of libpthread-2.dll in your path. This option however has not been tested. Please post to the forum if you have questions regarding compiling OpenMx.
There is a known bug with OpenMP and OS X 10.7. You can install a more recent version of gcc on your machine, this should replace the default runtime with a more stable library. A more recent gcc version is available from macports. Another alternative is to install the sequential OpenMx binary.