You are here

Third OpenMx 2.0 beta is released

The third beta of OpenMx version 2.0 is now publicly available! Interested users can download and install this new version by copy-pasting source('https://openmx.ssri.psu.edu/getOpenMxBeta.R') into the R command line and pressing 'Return'. Documentation for the beta is available here.

Compared to the second beta, the performance of this release is much improved. The second beta actually introduced a performance decrement in full information maximum likelihood (FIML) analyses that has now been corrected. Further, Item Factor Analysis (IFA) models now converge much more quickly, due to implementation of an improved expectation-maximization algorithm.

CSOLNP, a new, open-source numerical optimizer written in C++, was introduced in the first beta. Several CSOLNP-related bugs have now been fixed. Additionally, a bug with mxFitFunctionMultigroup() has been fixed.

This release also features a new function, mxTryHard(). This function makes multiple attempts at running a model until an acceptable solution is reached. Between attempts, it randomly "jiggles" the most recent set of parameter estimates and uses them as start values on the next attempt.

Users should be aware of the following issues:

  • The Windows binaries of this third beta were compiled without enabling multiple threads (i.e., parallel processing with multiple cores). We discovered that, although running multiple threads did work under Windows, it was not safe and would crash R for certain kinds of OpenMx analyses. Savvy Windows users can of course try to build parallel binaries themselves from OpenMx source code.
  • CSOLNP can still crash or freeze up R, or terminate calculation prematurely, when calculating confidence intervals. This behavior appears to be Windows-specific, and almost entirely confined to 64-bit R.
  • Users should get into the habit of ALWAYS using the $ accessor instead of the @ accessor. For example, myModelRun@output$estimate would now be myModelRun$output$estimate. We CANNOT guarantee that every usage of the @ accessor that worked with versions 1.3/1.4 will continue to work in 2.0.

We are eager to receive user feedback about this beta of OpenMx 2.0! Please post questions, comments, and issues to the OpenMx Forums.